CN113991640A - Multi-energy complementary energy base energy configuration planning method based on thermal power - Google Patents

Abstract

The invention relates to a thermal power-based multi-energy complementary energy base energy configuration planning method, which comprises the following steps: constructing a wind power output model based on a solar output scene; building a photovoltaic output model based on a solar output scene; constructing a thermal power generating unit output model; constructing a historical load model of an original thermal power line; building a digestion model based on a time sequence production simulation method; and obtaining a power supply combination scheme of the multi-energy complementary energy base according to a digestion model based on a time sequence production simulation method. From the perspective of power generation enterprises, the invention fully utilizes the advantages of the existing thermal power output line and combines the basis of the thermal power flexible peak regulation of the thermal power output line to form the active and prospective energy development structure planning.

Description

Multi-energy complementary energy base energy configuration planning method based on thermal power

Technical Field

The invention relates to the technical field of electric power, in particular to a thermal power-based multi-energy complementary energy base energy configuration planning method.

Background

The development and consumption of non-fossil energy mainly based on new energy are increased, and the decisive force for improving the occupation ratio of the non-fossil energy is achieved. By preferentially utilizing clean energy resources, fully exerting the regulation performance of water, electricity and coal, properly configuring energy storage facilities and mobilizing the flexible response enthusiasm of a demand side, the advantages of new energy resource enrichment areas are favorably exerted, the large-scale consumption of clean electricity is realized, the energy structure is optimized, the resource environment constraint is broken, the coordinated sustainable development of the energy field and the ecological environment is promoted, and the ecological civilization construction is promoted. The energy pattern of rich coal, poor oil and less gas in China is that new energy (photovoltaic and wind power) is equipped around the existing thermal power generating unit, energy is stored, and the new energy is sent out by means of the original power transmission line, so that a multi-energy complementary energy base is formed, and the energy pattern is the development direction of future thermal power.

The complementation, storage and conversion of various energy sources are realized through the integration of wind, light, water, fire and storage, the problems caused by the intermittence and instability of wind and light resources can be reduced, a stable comprehensive output curve of various energy sources is obtained, and the stability and the reliability of a power system are improved. By means of on-site development and consumption of new energy, flexible adjusting resources of the power grid can be occupied as little as possible or even not, the phenomena of wind and light abandonment are reduced, and dependence on the power grid is reduced to a great extent. Meanwhile, the allocation capacity of the energy storage device can be obviously reduced, and the system cost tends to be reasonable.

At present, most researches consider the energy structure proportion of a multi-energy complementary energy base from the perspective of a power grid, and for power generation enterprises, the early-stage prejudgment and the initiative of the power generation enterprises are lacked, so that the power generation enterprises are passive.

For a multi-energy complementary energy base formed by thermal power generating units, from the perspective of thermal power generation enterprises, how to accurately calculate the maximum optimal proportion of new energy is a problem existing in the current thermal power enterprises by taking the original thermal power generating units as adjusting power supplies under the condition of lacking an accurate power load model of a power system.

Disclosure of Invention

The invention aims to provide a thermal power-based multi-energy complementary energy base energy configuration planning method, which is used for forming an active prospective energy development structure planning by fully utilizing the advantages of thermal power output lines of a power generation enterprise and combining the basis of thermal power flexible peak regulation of the power generation enterprise.

The invention provides a thermal power-based multi-energy complementary energy base energy configuration planning method, which comprises the following steps:

step 1, constructing a wind power output model based on a sunpower scene, comprising the following steps of:

step 1.1, establishing a power balance scene and a peak shaving balance scene;

the power balance scene establishment steps are as follows:

6) the minimum output P of the wind power generation curve in each season according to the late peak period_EPminRanked in order of minimum output P at late peak hours_EPminAs a main characteristic index of the power balance scene, the daily wind power generation curve in each season is expressed according to P_EPminSorting from small to large to form order statistics;

7) given P_EPminDetermining a wind power generation curve set P based on the confidence level alpha_WΩ1Given P_EPminThe confidence level alpha of the wind power generation curve P is screened out_W(d) With a minimum output of P at late peak hours_EPmin(d) Ensuring that the minimum output of the late peak period of the wind power generation curve of the season is not less than P with the probability exceeding alpha_EPmin(d) Recording the set of wind power generation curves meeting the conditions as P_WΩ1：

Note P_WΩ1The power generation curve with the maximum guarantee rate of the electric quantity in the middle day is P_μ1；

3) At P_WΩ1In the method, a wind power daily power generation curve with higher daily electric quantity guarantee rate or smaller daily average output is selected as a power balance scene, wherein,

the first method is as follows: selecting the daily electric quantity guarantee rate:

at a given daily charge guarantee rate β, in set P_WΩ1Middle screening out wind power daily power generation curve P_W(d) Let P stand_W(d) Daily electricity quantity guarantee rate lambda_E(d) Not less than beta, and recording that the set of wind power daily generation curves meeting the conditions is P'_WΩ1：

Note P'_WΩ1Minimum output P at peak time of middle and late_EPminMinimum power generation curve is P_ν1；

The second method comprises the following steps: selecting average daily output:

confidence level β given mean output of days, in set P_WΩ1Middle screening out wind power daily power generation curveP_W(d) And ensuring that the average output of the wind power daily generation curve in the season is not low P with the probability of exceeding beta_DaveThe set of wind power daily generation curves satisfying the conditions is recorded as P'_WΩ1：

Note P'_WΩ1Mean output P in the middle of the day_DaveThe maximum power generation curve is P_ν1；

4) Selecting wind power output power balance scene P of each season_W1

If P 'is determined according to the mode I in 3)'_WΩ1In the presence of a catalyst satisfying P_EPminSelecting P according to the confidence level alpha and the electric quantity guarantee rate beta_EPminThe minimum power generation curve is a power balance scene to reflect the most extreme influence of wind power on the power balance of the system; when the requirement of the electric quantity guarantee rate beta cannot be met, taking a power generation curve with the maximum electric quantity guarantee rate as an electric power balance scene;

if P 'is determined in the mode of 3)'_WΩ1In the presence of a catalyst satisfying P_EPminConfidence levels of alpha and P_DaveOn the requirement of the confidence level beta, selecting a power generation curve closest to the daily average output confidence level beta as a boundary power balance scene; when the requirement of the confidence level beta cannot be met, taking a power generation curve with the minimum daily average output as a power balance scene;

6) determining the exact probability of a power balance scenario:

in the formula, N_rThe total number of the quarterly output curves;

the peak regulation balance scene establishment steps are as follows:

1) for wind in all seasonsForce-day power generation curve according to delta P_PDmaxSorting

Peak load demand Δ P at daily maximum_PDmaxAs a main characteristic index of a peak-shaving balance scene, a wind power daily generation curve in each season is divided into delta P_PDmaxSorting from big to small to form order statistics;

2) given Δ P_PDmaxDetermining a wind power generation curve set P based on the confidence level gamma_WΩ2

Given Δ P_PDmaxThe confidence level gamma of the wind power generation curve P is screened out_W(d) The daily maximum peak shaver requirement is delta P_PDmax(d) Make sure that the season Δ P is certain with an excess probability_PDmaxNot more than Δ P_PDmax(d) Recording the set of wind power generation curves meeting the conditions as P_WΩ2：

Recording daily minimum peak regulation demand delta P in set_PDminThe maximum power generation curve is P_μ2；

8) At P_WΩ2In the form of Δ P_PDminDetermining a wind power generation daily curve set P 'with the constraint of more than or equal to 0'_WΩ2At Δ P_PDminConstraint condition of more than or equal to 0 is from P_WΩ2Screening out a daily wind power generation curve P_W(d) The set of wind power daily generation curves satisfying the conditions is recorded as P'_WΩ2The formula is as follows:

P'_wΩ2＝{P_w(d)|ΔP_PDmin(d)≥0,P_w(d)∈P_wΩ2}

note P'_WΩ2Maximum peak load demand Δ P in mid-day_PDmaxThe maximum power generation curve is P_ν2；

9) Selecting wind power output peak regulation balance scene P of each season_W2

If P_'WΩ2Is not an empty collectionI.e. the presence of a gas satisfying Δ P_PDmaxConfidence levels gamma and delta P_PDminSelecting delta P in the wind power daily generation curve with constraint of more than or equal to 0_PDmaxThe maximum power generation curve is a peak-load balancing scene to reflect the most extreme influence of wind power on the peak-load balancing of the system; if P'_WΩ2Is an empty set, i.e. P_WΩ2All the daily wind power generation curves cannot satisfy delta P_PDminConstraint of not less than 0, then take Δ P_PDminThe maximum wind power daily generation curve is a peak-shaving balance scene and is close to reverse peak shaving to the maximum extent;

10) determining the exact probability of a peak shaver balance scene;

in the formula, N_rThe total number of the quarterly output curves;

step 1.2, establishing a wind power daily output clustering scene, which comprises the following steps:

(1) the K-means clustering method based on the weighted Euclidean distance effectively clusters and screens a large number of wind power output scenes, and comprises the following specific steps:

2) from N wind power output curves P_mSelecting k pieces of (M is 1,2,3, …, N) as initial clustering centers M_i(i＝1,2,3,…,k)；

2) According to the requirements of practical application, determining the weight coefficient omega occupied by the wind power output in the clustering process at the load early peak time, the load late peak time, the load valley time and the load waist time_t(t＝1,2,3,…,S)；

3) Calculating each wind power output curve P in sequence_mWith respective cluster centers M_iA distance l of_miDistributing the wind power output curve to the category closest to the clustering center;

in the formula, P_mtAnd M_itRespectively is a wind power output curve P_mAnd a clustering center M_ithe output value at the moment t;

4) calculating a new clustering center in each category and re-clustering, and performing iterative calculation in a loop until a criterion function of clustering is not changed, wherein the criterion function is as follows:

wherein

In the formula, e_miIs a state variable; r_iThe method comprises the steps of collecting all wind power output curves in the ith wind power output scene;

recording the exact probability of the ith type wind power output scene as p_i，

(2) Determining typical daily wind power generation curves of various scenes, which comprises the following specific steps:

2) calculating the daily power generation curve d of wind power and other power generation curves q in the class about delta P_PDmax、ΔP_PDminIs weighted by the average euler distance S_i(d)；

In the formula, k_αIs an index of Δ P_PDmaxThe weight of (c);

2) get S_i(d) The minimum power generation curve is a scene class R_iTypical daily wind power generation curve, denoted as P_Ri；

3) Correcting scene class R_iTypical wind power daily generation curve P_RiThe generated power of (3);

the correction process is as follows:

calculating a scene class R_iWind power total generating capacity E_Ri

In the formula, P_W(dt) represents scene class R_iThe output of the medium wind power generation curve d at the tth hour;

② calculating scene class R_iTypical wind power daily generation curve P_RiExpected power generation amount

In the formula, P_Ri(t) represents P_RiThe force at time t;

calculating scene class R_iTypical wind power daily generation curve P_RiCorrected electric quantity delta E of

Fourthly, calculating scene class R_iTypical wind power daily generation curve P_RiCorrected hourly power contribution δ P

Fifthly, obtaining the scene class R_iCorrected typical wind power generation daily curve P_Ri^△

Wherein the expression represents the output value at the t hour; after correction, if the wind power output at a certain moment overflows the scene wind power output range, the overflowing electric quantity at the moment is spread to other moments;

step 2, constructing a photovoltaic output model based on the sunoutput scene, comprising the following steps:

step 2.1, when the power balance of a system containing photovoltaic power generation needs to be evaluated, directly replacing a power balance scene with a photovoltaic scene with the minimum output in a clustering scene; taking the maximum output P of each season day_DmaxThe maximum photovoltaic daily power generation curve is used as a peak-load balancing scene of photovoltaic power generation; the exact probability calculation formula of the photovoltaic peak-shaving balance scene is the same as that of wind power;

2.2, clustering photovoltaic daily power generation curves in each season into a plurality of typical scenes according to photovoltaic historical output data by adopting a forward scene reduction method based on a Kantorovich Distance; through repeated iteration, a scene with the minimum KD distance to other scenes is selected from the original scene set omega and is placed into the scene set omega ', and the KD distances of the two scene sets omega and omega' are defined as follows:

in the formula, s and s 'are scenes in a scene set omega and omega' respectively; p is a radical of_sAnd p_s'Probability of scenes s and s 'in Ω and Ω', respectively; c (s, s') is a non-negative, continuous, symmetrical distance function; μ (s, s ') is the probability product of scene s and s';

the forward scene reduction based on the KD distance comprises the following steps:

(1) determining a scene requiring clipping: eliminating scenes omega meeting the following conditions_s'

In the formula, c (ω)_s',ω_m) The distance between the two photovoltaic sunrise force curves;

(4) total number of changed scenes: n-1; and screen out and cut scene omega_s'Nearest scene omega_sI.e. by

(5) Changing and culling scene omega_s'Nearest scene omega_sTo ensure that the sum of the probabilities of all remaining scenes is 1;

P_s＝P_s+P_s’

(4) the iterative calculation is circulated until the residual scene number meets the set target scene number N_sUntil the requirements are met; each representative scene omega_sHas a probability of p_s；

Step 3, constructing a thermal power generating unit output model, comprising the following steps:

constructing a condensing thermal power generating unit output model;

constructing a back pressure thermal power generating unit output model;

constructing a steam extraction type thermal power generating unit output model;

step 4, constructing a historical load model of the original thermal power circuit, comprising the following steps:

determining a reference daily load curve:

according to a typical working day and rest day curve every month, and considering typical holidays separately, selecting representative days every month to eliminate the influence of abnormal factors; or comprehensively analyzing and comparing typical curves of the month in each year in history to determine a representative curve of the month; the following analytical methods were used for each typical day data:

let T be 24, denote the number of time segments, and let the load data of the day to be analyzed be l_i(i 1,2,.. and T), and a daily maximum load of l₀In 1 with₀To l_iPerforming per unit to obtain the load curve d of the day_i(i ═ 1,2, · and T), the following relationships are established:

l₀＝max l_i 1≦i≦T

d_i＝l_i/l₀

and 5, building a digestion model based on a time sequence production simulation method, wherein the digestion model comprises the following steps:

step 5.1 determining the objective function

The objective function in the optimization cycle is:

in the formula: t represents the total length of time; t is the simulation time step length; p_w(t) is the wind power output of the wind-light-fire multi-energy base in the time period t, P_pv(t) generating power by solar energy in the wind, light and fire multi-energy base at the time t;

step 5.2, determining constraint conditions

(1) Regional load balancing constraints

P_j(t)×S_j(t)+P_W(t)+P_PV(t)＝P_I(t)

In the formula: p_j(t)×S_j(t) is the sum of the powers of the conventional units;

(2) unit output constraint

0≤ΔP_j(t)≤[P_j,max(t)-P_j,min(t)]×S_j(t)

P_j(t)＝P_j,min(t)×S_j(t)+ΔP_j(t)

In the formula, P_j(t) optimizing the power of the conventional unit;

(3) unit optimized power ramp rate constraint

P_j(t+1)-P_j(t)≤ΔP_j,up(n)

P_j(t)-P_j(t+1)≤ΔP_j,down(n)

In the formula,. DELTA.P_j,up，ΔP_j,downRespectively the climbing rate and the descending rate of the jth unit;

(4) output constraint of heat supply unit in heat supply period

The generated output and the thermal output of the back pressure type cogeneration thermal power unit are in a linear relationship:

P_BYJ(t)＝C_j,b×Q_j(t)

the linear constraint formula of the steam extraction type cogeneration thermal power unit is shown as the following formula:

Q_j(t)×C_j,b≤P_CQJ,max-Q_j(t)×C_j,v

(5) new energy output constraint

0≦P_w(t)≦P^*_w(t)

0≦P_pv(t,≦P^*_pv(t)

In the formula: p^*_w(t) wind-power time-series output at a time t when the capacity of the time-varying machine is constant, P^*_pv(t) a photovoltaic time series output at a time t when the capacity of the fashion machine is constant;

(6) proportional constraint of new energy abandonment

Allowing a certain proportion of new energy to be abandoned in the load valley period so as to replace larger new energy consumption;

step 5.3, model solving method

The new energy production simulation model is abbreviated as follows:

Minf(x)

s.t g_i(x)≧0(i＝1,2....m)g_i(x)≧0(i＝1,2....n)

Hj(x)＝0(j＝1,2....n)

wherein: x is the set of variables to be optimized, f (x) is the optimization objective function, g_i(x) ≧ 0 is the set of inequality constraints, h_j(x) 0 is the set of equality constraints;

and 6, obtaining a power supply combination scheme of the multi-energy complementary energy base according to a digestion model based on a time sequence production simulation method.

By means of the scheme, through the thermal power-based multi-energy complementary energy base energy configuration planning method, the load model of the multi-energy complementary energy base is built according to the historical output of the original thermal power output line, and the new energy daily output scene and the probability distribution thereof are built on the basis of the historical output time sequence record of the new energy historical output data. In order to comprehensively reflect the full-time output characteristics of the new energy and the influence of the full-time output characteristics on the power balance, the electric quantity balance and the peak regulation balance of the system, a new energy daily output scene is divided into a key scene and a clustering scene, and a high-precision new energy consumption analysis model is formed by combining a time sequence production simulation method. From the perspective of power generation enterprises, the invention fully utilizes the advantages of the existing thermal power output line and combines the basis of the thermal power flexible peak regulation of the thermal power output line to form the active and prospective energy development structure planning.

The foregoing description is only an overview of the technical solutions of the present invention, and in order to make the technical solutions of the present invention more clearly understood and to implement them in accordance with the contents of the description, the following detailed description is given with reference to the preferred embodiments of the present invention and the accompanying drawings.

Drawings

FIG. 1 is a schematic diagram of an energy configuration planning method for a multi-energy complementary energy base based on thermal power;

FIG. 2 is a flow chart of a wind power output model constructed based on a solar output scenario according to the present invention;

FIG. 3 is a graph of the relationship between electricity and heat output of a condensing unit;

FIG. 4 is a graph of electricity versus heat output for a back pressure unit and a steam extraction unit; (a) a back pressure machine set and a gas extraction machine set.

Detailed Description

The following detailed description of embodiments of the present invention is provided in connection with the accompanying drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.

Referring to fig. 1, the present embodiment provides a thermal power-based energy allocation planning method for a multi-energy complementary energy base, including:

step 1: wind power output model based on solar output scene

The wind power output key scene is divided into a power balance scene and a peak-load balancing scene. The power balance scene is a wind power output scene which has a remarkable influence on a system power balance operation simulation result and reflects the capacity value of a wind power plant; the peak-shaving balance scene is a wind power output scene which has a remarkable influence on a system peak-shaving balance operation simulation result, and influences the peak-shaving power supply requirement and the power supply structure of the system. After key scenes are deducted from the new energy daily output curve, a certain clustering method is adopted to form a new energy output clustering scene, which mainly influences the electric quantity balance of the system and the utilization rate of new energy. And in consideration of the difference of the wind-solar sunrise characteristics, clustering scenes are respectively clustered by adopting different methods.

The historical output data of the wind power plant implies the inherent statistical regularity, and meanwhile, the characteristic information of the wind speed in the area where the wind power plant is located, such as the time sequence, autocorrelation and the like, and the wind speed change information (long-term change trend, seasonal change rule, periodic change rule and the like) are completely kept in the historical output data. However, because the historical data amount of wind power output is huge, the output data of each day cannot be brought into the operation simulation system for calculation, the number of scenes can be reduced in a clustering mode, and the effect of several times or even tens of times of scenes can be achieved by using a representative scene with few measuring tools, so that the calculation efficiency is improved. In order to fully consider the characteristics of wind power uncertainty, volatility, regionality, reverse peak shaving, seasonal difference and the like, the overall flow chart is shown in the following figure 2:

step 1.1 wind power solar output key scene

Firstly, two wind power output key scenes, namely a power balance scene and a peak shaving balance scene, are established, the two scenes are taken as key scenes and brought into system operation simulation, the power balance scene is obvious to a power balance result of a system and a system planning overall scheme, and the peak shaving balance scene is obvious to influence the peak shaving requirement of the system.

(1) Power balance scenario

The power balance scene is a wind power daily output scene which has a remarkable influence on a system power balance operation simulation result. Different power balance scenes correspond to different wind power plant capacity credibility (namely, the effective capacity of the wind power plant), and the embodied wind power plant capacity values are also different. Because the uncertainty of wind power output and the uncertainty of the capacity provided by wind power cannot provide the capacity matched with the installed capacity of the wind power, the condition that the effective capacity of a wind power plant is low is considered on the planning level, and the wind power output power balance scene in each season is reasonably selected by combining historical output data, and the method mainly comprises the following steps:

11) the minimum output P of the wind power generation curve in each season according to the late peak period_EPminRanked in order of minimum output P at late peak hours_EPminAs a main characteristic index of the power balance scene, the daily wind power generation curve in each season is expressed according to P_EPminThe order statistics are formed by sorting from small to large.

12) Given P_EPminDetermining a wind power generation curve set P based on the confidence level alpha_WΩ1Given P_EPminThe confidence level alpha of the wind power generation curve P is screened out_W(d) (with a minimum output of P during late peak hours_EPmin(d) To make sure that the minimum output of the wind generation curve in the season is not less than P during the late peak period with a probability exceeding alpha_EPmin(d) Recording the set of wind power generation curves meeting the conditions as P_WΩ1：

Note P_WΩ1The power generation curve with the maximum guarantee rate of the electric quantity in the middle day is P_μ1。

3) At P_WΩ1In the method, a wind power day generation curve set P 'is determined according to a certain principle'_WΩ1Considering the different requirements of power supply planning, in P_WΩ1On the basis, a wind power daily power generation curve with high daily electric quantity guarantee rate or low daily average output can be selected as a power balance scene. Wherein

The first method is as follows: and selecting the daily electric quantity guarantee rate. At a given daily charge guarantee rate β, in set P_WΩ1Middle screening out wind power daily power generation curve P_W(d) Let P stand_W(d) Daily electricity quantity guarantee rate lambda_E(d)Not less than beta, and recording that the set of wind power daily generation curves meeting the conditions is P'_WΩ1：

Note P'_WΩ1Minimum output P at peak time of middle and late_EPminMinimum power generation curve is P_ν1。

The second method comprises the following steps: selecting the average daily output. Confidence level β given mean output of days, in set P_WΩ1Middle screening out wind power daily power generation curve P_W(d) And ensuring that the average output of the wind power daily generation curve in the season is not low P with the probability of exceeding beta_DaveThe set of wind power daily generation curves satisfying the conditions is recorded as P'_WΩ1：

Note P'_WΩ1Mean output P in the middle of the day_DaveThe maximum power generation curve is P_ν1。

4) Selecting wind power output power balance scene P of each season_W1

The meaning of this formula is: if P 'is determined according to the mode I in 3)'_WΩ1In the presence of a catalyst satisfying P_EPminSelecting P according to the confidence level alpha and the electric quantity guarantee rate beta_EPminThe minimum power generation curve is a power balance scene which can reflect the most extreme influence of wind power on the power balance of the system; when the requirement of the electric quantity guarantee rate beta cannot be met, determining P 'according to the mode II in the step 3) if the power generation curve with the maximum electric quantity guarantee rate is the power balance scene'_WΩ1In the presence of a catalyst satisfying P_EPminConfidence levels of alpha and P_DaveOn the requirement of the confidence level beta, selecting a power generation curve closest to the daily average output confidence level beta as a boundary power balance scene; when the confidence water is not satisfiedAnd when the average output per day is required, taking the power generation curve with the minimum average output per day as a power balance scene.

7) The exact probability of a power balance scenario is determined.

In the formula, N_rThe total number of the quarterly output curves.

(2) Peak-shaving balance scene

The peak-shaving balance scene is a wind power daily output scene which has obvious influence on a system peak-shaving balance operation simulation result. Different peak regulation balance scenes correspond to different wind power peak regulation capacities, and the peak regulation power supply requirement and the power supply structure of the system are influenced. Because wind power output often has a peak-reversal regulation characteristic, the condition of wind power plant peak-reversal regulation should be considered on a planning level, and a peak-regulation balance scene is reasonably screened by combining historical output data. The method comprises the following specific steps:

1) the daily wind power generation curve is expressed by delta P for each season_PDmaxSorting

Peak load demand Δ P at daily maximum_PDmaxAs a main characteristic index of a peak-shaving balance scene, a wind power daily generation curve in each season is divided into delta P_PDmaxThe order statistics are formed by sorting from large to small.

2) Given Δ P_PDmaxDetermining a wind power generation curve set P based on the confidence level gamma_WΩ2

Given Δ P_PDmaxThe confidence level gamma of the wind power generation curve P is screened out_W(d) (daily maximum Peak Regulation requirement. DELTA.P_PDmax(d) To make sure that the season Δ P is satisfied with the probability of being exceeded_PDmaxNot more than Δ P_PDmax(d) Recording the set of wind power generation curves meeting the conditions as P_WΩ2：

Recording daily minimum peak regulation demand delta P in set_PDminThe maximum power generation curve is P_μ2；

13) At P_WΩ2In the form of Δ P_PDminDetermining a wind power generation daily curve set P 'with the constraint of more than or equal to 0'_WΩ2To fully ensure the inverse peak-shaving characteristic of the peak-shaving balance scene, the delta P is used_PDminConstraint condition of more than or equal to 0 is from P_WΩ2Screening out a daily wind power generation curve P_W(d) The set of wind power daily generation curves satisfying the conditions is recorded as P'_WΩ2The formula is as follows:

P'_wΩ2＝{P_w(d)|ΔP_PDmin(d)≥0,P_w(d)∈P_wΩ2}

note P'_WΩ2Maximum peak load demand Δ P in mid-day_PDmaxThe maximum power generation curve is P_ν2；

14) Selecting wind power output peak regulation balance scene P of each season_W2

The meaning of this formula is: if P_'WΩ2Not being empty, i.e. there is a Δ P_PDmaxConfidence levels gamma and delta P_PDminSelecting delta P in the wind power daily generation curve with constraint of more than or equal to 0_PDmaxThe maximum power generation curve is a peak-load balancing scene which can reflect the most extreme influence of wind power on the peak-load balancing of the system; if P'_WΩ2Is an empty set, i.e. P_WΩ2All the daily wind power generation curves cannot satisfy delta P_PDminConstraint of not less than 0, then take Δ P_PDminThe maximum daily wind power generation curve is a peak-shaving balance scene and is close to inverse peak shaving to the maximum extent.

15) The exact probability of a peak shaver balance scenario is determined.

In the formula, N_rThe total number of the quarterly output curves.

Step 1.2 clustering scene of wind power daily output

The clustering of the wind power solar output scenes can reduce the number of original scenes, and the full-time-space characteristic of the wind power output is represented with smaller workload and higher precision. According to the method, wind power output characteristics of load in early peak, late peak, low valley and waist load periods are used as a basis, wind power solar output curves with key scenes removed are clustered, and a wind power output clustering scene and probability distribution thereof are generated.

(1) K-means clustering method based on weighted Euclidean distance

The K-means clustering algorithm is to randomly select K objects as initial clustering centers, then calculate the distance between each object and each initial clustering center, assign each object to the closest clustering center, and the clustering centers and the objects assigned to the clustering centers represent a category. When the distribution of all the objects is completed, the cluster center of each category is reselected according to the existing objects in the category, and the iterative computation is circulated until a certain termination condition is met. The termination condition may be that the cluster center is no longer changing or that the sum of squared errors in each class is minimal, etc.

However, the traditional K-means clustering algorithm does not consider that each variable in an object has different effects in a clustering process, but considers the same, and the similarity of two objects represented by the method may have certain limitations in practical application. The distance between the objects represents the degree of closeness of the objects, and the degree of similarity depends not only on the degree of closeness between the objects, but also on the inherent properties of the objects, i.e., the importance of each variable in the objects is different.

For a wind power output sequence, the load demand of a system reaches the peak at a late peak period, and the output of the system mainly influences the power supply capacity demand of the system and the formulation of a start-up and shut-down plan; the load level at the early peak time is lower than that at the later peak time, and the wind power output further influences the reliability index of the system under the condition of meeting the power balance; in the load valley period, the wind power output mainly influences the peak regulation requirement of the power system and the consumption level of new energy.

Therefore, in order to fully consider different influences of wind power output on power system scheduling and power system operation simulation at different time periods in the day, the report provides a weighted Euclidean distance-based K-means clustering algorithm. According to the requirements in practical application, different weights are given to the wind power output in different time periods, so that effective clustering and screening of a large number of wind power output scenes are achieved. The method comprises the following specific steps:

3) from N wind power output curves P_mSelecting k pieces of (M is 1,2,3, …, N) as initial clustering centers M_i(i＝1,2,3,…,k)；

2) According to the requirements of practical application, determining the weight coefficient omega occupied by the wind power output in the clustering process at the load early peak time, the load late peak time, the load valley time and the load waist time_t(t＝1,2,3,…,S)；

3) Calculating each wind power output curve P in sequence_mWith respective cluster centers M_iA distance l of_miDistributing the wind power output curve to the category closest to the clustering center;

in the formula, P_mtAnd M_itRespectively is a wind power output curve P_mAnd a clustering center M_ithe force output value at the moment t.

4) And calculating a new clustering center in each category and re-clustering, and performing iterative calculation in a loop until the criterion function of the clustering is not changed. The criterion function is as follows:

wherein

In the formula, e_miIs a state variable; r_iAnd collecting all wind power output curves in the ith wind power output scene.

Recording the exact probability of the ith type wind power output scene as p_i，

(2) Determining typical daily wind power generation curve of various scenes

Firstly, the peak regulation pressure of a power grid is further intensified after the wind power is connected to the grid due to the anti-peak regulation characteristic of the wind power, and great difficulty is brought to the dispatching operation of the power grid, so that typical daily output curves of the wind power in various scenes can well reflect the peak regulation benefits of the original output curves in various scenes.

Secondly, in order to ensure the accuracy of the typical wind power output curves of various scenes in the system electric quantity balance analysis, the electric quantity characteristics of the original power generation curve must be correctly reflected, namely the electric quantity characteristics are consistent with the expected generated energy of the original power generation curve.

Based on the method, typical wind power output curves of various scenes are determined according to the following steps:

3) calculating the daily power generation curve d of wind power and other power generation curves q in the class about delta P_PDmax、ΔP_PDminIs weighted by the average euler distance S_i(d)。

In the formula, k_αIs an index of Δ P_PDmaxThe weight of (c).

2) Get S_i(d) The minimum power generation curve is a scene class R_iTypical daily wind power generation curve, denoted as P_Ri。

3) Correcting scene class R_iTypical wind power daily generation curve P_RiThe power generation amount of (1).

Under the condition of ensuring that the output value of the corrected power generation curve at each moment does not overflow the scene wind power output range, the electric quantity to be corrected is spread to the whole day, namely the whole curve is longitudinally translated, and the peak regulation requirement of the typical wind power daily power generation curve is not changed as much as possible. The correction process is as follows:

calculating a scene class R_iWind power total generating capacity E_Ri

In the formula, P_W(dt) represents scene class R_iAnd (4) the output of the Tth hour of the medium wind power daily generation curve d.

② calculating scene class R_iTypical wind power daily generation curve P_RiExpected power generation amount

In the formula, P_Ri(t) represents P_RiForce at time t.

Calculating scene class R_iTypical wind power daily generation curve P_RiCorrected electric quantity delta E of

Fourthly, calculating scene class R_iTypical wind power daily generation curve P_RiCorrected hourly power contribution δ P

Fifthly, obtaining the scene class R_iCorrected typical wind power generation daily curve P_Ri^△

In the formula, the output value at the t-th hour is shown. And after correction, if the wind power output at a certain moment overflows the scene wind power output range, leveling the overflowing electric quantity at the moment to other moments.

The output model based on the wind power daily output scene and the probability distribution thereof comprehensively considers typical daily load characteristics of all seasons, time correlation of wind power daily power generation, and power structure and power generation scheduling characteristics of China, can reflect the influence of wind power generation on system power balance and peak shaving balance, also considers the influence of wind power output on system power balance in different load periods, and can reflect full-time-space characteristics of wind power output.

Step 2: photovoltaic output model based on solar output scene

The correlation between the highest peak time period of photovoltaic power generation and the highest peak time period of load is not obvious, and the power balance of the system is indirectly influenced only by influencing the capacity utilization rate of the energy-limiting power station, so that the peak-load balancing scene is only considered in the key photovoltaic output scene, and the power balancing scene can be directly replaced by the photovoltaic scene with the minimum output in the clustering scene.

Step 2.1 photovoltaic solar output key scene

(1) Power balance scenario

The correlation between the peak time of the photovoltaic power generation and the peak time of the load is not obvious, the system installation level is not directly influenced, and the power balance of the system is indirectly influenced only by influencing the capacity utilization rate of the energy-limiting power station. Therefore, a photovoltaic output power balance scene is not independently selected, and when the power balance of a system containing photovoltaic power generation needs to be evaluated, the photovoltaic scene with the minimum output in the clustering scenes can be directly used for replacing the photovoltaic scene.

(2) Peak-shaving balance scene

The output of the photovoltaic power generation load in the low-valley period is 0, the output of the photovoltaic power generation load in the high-peak period is relatively fixed and generally inconsistent with the highest-peak period of the load, so the output level of the photovoltaic maximum output day obviously influences the peak regulation balance of the system and the light abandoning level of the photovoltaic power station, and the maximum output P of the photovoltaic power station in each season day can be taken_DmaxAnd taking the maximum photovoltaic daily power generation curve as a peak-load balancing scene of photovoltaic power generation. The exact probability calculation formula of the photovoltaic peak-shaving balance scene is the same as that of wind power.

Step 2.2 photovoltaic solar output clustering scene

Compared with wind power, the randomness of photovoltaic power generation in one day is weak, the day and night characteristics are obvious, and the output level in the same load period generally does not change violently, so that the output characteristics of photovoltaic output in each load period are not considered when a photovoltaic output clustering scene is determined. Based on the method, after a photovoltaic output key scene is deducted, a forward scene reduction technology based on a Kantorovich Distance (KD) Distance is adopted, and photovoltaic daily power generation curves in all seasons are clustered into a plurality of typical scenes according to photovoltaic historical output data. A forward scene reduction technology based on KD distance is an optimization process, and a scene with the minimum KD distance to other scenes is selected from an original scene set omega through repeated iteration and is placed into a scene set omega'. The KD distance of two scene sets Ω, Ω' is defined as follows

In the formula, s and s 'are scenes in a scene set omega and omega' respectively; p is a radical of_sAnd p_s'Probability of scenes s and s 'in Ω and Ω', respectively; c (s, s') is a non-negative, continuous, symmetrical distance function; μ (s, s ') is the probability product of scene s and s'.

The forward scene reduction based on the KD distance comprises the following steps:

(1) determining a scene requiring clipping: eliminating scenes omega meeting the following conditions_s'

In the formula, c (ω)_s',ω_m) The distance between the two photovoltaic sunrise force curves;

(6) total number of changed scenes:n-1; and screen out and cut scene omega_s'Nearest scene omega_sI.e. by

(7) Changing and culling scene omega_s'Nearest scene omega_sTo ensure that the sum of the probabilities of all remaining scenes is 1;

P_s＝P_s+P_s’

(4) the iterative calculation is circulated until the residual scene number meets the set target scene number N_sUntil the requirements are met; each representative scene omega_sHas a probability of p_s。

And step 3: method for constructing thermal power generating unit output model

The thermal power plant is an important component of the active power source of the power system. The peak regulation characteristic of the thermal power generating unit is important for receiving new energy, so that the output characteristics of different types of thermal power generating units are mainly considered during production simulation, and particularly the output adjustable change of the heat supply unit in the heat supply period is considered.

The thermal power generating set is divided into a condensing steam turbine set for power generation and a back pressure type and steam extraction type steam turbine set for heat supply. The condensing steam turbine set does not supply heat load, and the output of the condensing steam turbine set is adjusted between the minimum technical output and the rated output; the steam turbine unit with heat supply firstly ensures heat supply, and in order to improve the fuel utilization rate, a certain amount of steam must be conveyed from a boiler to a steam turbine, namely a certain amount of active power corresponding to the heat supply must be emitted.

(1) Condensing thermal power generating unit

The condensing thermal power generating unit does not supply heat, so that the output of the condensing thermal power generating unit is unrelated to the heat supply, and can be represented as a curve relation shown in fig. 3. The relation between the efficiency and the fuel consumption of the condensing thermal power generating unit is as follows: if the unit efficiency is f, then fGJ electricity is generated from 1GJ of coal (3600000J corresponds to 1kWh of electricity).

(2) Back pressure type thermal power generating unit

The back-pressure thermal power generating unit is one of the heat supply units, and the operating characteristics thereof are as shown in the following formula and fig. 4 (a). In the formula C_bThe value is the ratio of the power output and the heat output of the unit.

Pi,t＝Hi,t×Cb

(3) Steam extraction type thermal power generating unit

Another operation characteristic of the extracted steam thermal power generating unit is shown in the following formula and fig. 4 (b). When the thermal output is fixed, the electric output of the unit can be in a certain range, namely C_bValue and C_vAnd (4) determining the value.

And 4, step 4: historical load model for constructing original thermal power line

The daily load curve is the basis for carrying out power system production simulation under the time sequence load curve, and the load position of each unit in the system, whether the peak regulation capacity of the system is enough, the magnitude of peak shifting benefit of the interconnected system and the like all depend on the shape of the daily load curve. Therefore, daily load distribution curve analysis and modeling are mainly performed.

(1) And (5) determining a reference daily load curve. The study is carried out according to a typical working day and rest day curve every month, and typical holidays such as New year, spring festival, five-one, eleven and the like can be considered independently. The selection of the representative day of each month should eliminate the influence of abnormal factors such as switching-off, power limiting, accidents and the like, and is as close as possible to the actual situation. It is also possible to make comprehensive analysis and comparison on the typical curves of the month in each year in history, for example, to perform weighted integration (the recent curves should be weighted more heavily) to determine the representative curves of the month. The following analytical methods were used for each typical day data:

let T be 24 (representing the number of time segments) and let the load data on the day to be analyzed be l_i(i 1,2,.. and T), and a daily maximum load of l₀In 1 with₀To l_iPerforming per unit to obtain the load curve d of the day_i(i ═ 1,2, · and T), the following relationships are established:

l₀＝max l_i 1≦i≦T

d_i＝l_i/l₀

and 5: building a digestion model based on a time sequence production simulation method

The time sequence-based production simulation is a time sequence simulation method for simulating the operation conditions of each generator set under a given load condition and calculating the production cost of a power generation system. And considering the system load and the generator set output as a time sequence which changes along with time. And the balance relation between the system load and the unit output is regarded as the supply and demand balance relation between the product and the demand, and the objective function is optimized under the constraint to obtain the optimal index. The time sequence production simulation plays an important role in the operation and decision of the power generation system, wherein the production simulation of a short time scale is generally different from several to dozens of hours, the system operation mode can be optimized, the new energy consumption capability is improved, more new energy electric quantity is consumed, and a reasonable power generation plan is provided for a scheduling department.

Step 5.1 determining the objective function

The established new energy acceptance capacity calculation model must fully consider the operation and output characteristics of various conventional units of an actual power system, including thermal power units and hydroelectric power units, including the start-stop characteristic of the units, the climbing characteristic of the units, the minimum output characteristic and the like, and further needs to consider the thermocouple characteristic of certain special types of units, such as a cogeneration unit. Therefore, the new energy admission capacity mathematical optimization analysis model comprehensively considers system balance constraint, power grid safety constraint, standby constraint, electric quantity constraint and unit operation constraint according to information such as a tie line exchange plan, a maintenance plan, a new energy power prediction curve, a system load prediction curve, a bus load prediction curve, a network topology, a unit power generation capacity and power plant operation constraint, and obtains an evaluation result of the new energy admission capacity by adopting an optimization evaluation algorithm considering the safety constraint.

The optimization target of the new energy time sequence production simulation model is that the new energy consumption is maximum in the optimization period, so that the objective function in the optimization period is as follows:

in the formula: t represents the total length of time; t is the simulation time step length; p_w(t) is the wind power output of the wind-light-fire multi-energy base in the time period t, P_pvAnd (t) the wind, light and fire multi-energy base generates power by solar energy in the time period t.

Step 5.2 constraint Condition

(1) Regional load balancing constraints

P_j(t)×S_j(t)+P_W(t)+P_PV(t)＝P_I(t)

In the formula: p_j(t)×S_j(t) is the sum of the powers of the conventional units;

(2) unit output constraint

0≤ΔP_j(t)≤[P_j,max(t)-P_j,min(t)]×S_j(t)

P_j(t)＝P_j,min(t)×S_j(t)+ΔP_j(t)

In the formula, P_jAnd (t) optimizing the power of the conventional unit.

(3) Unit optimized power ramp rate constraint

P_j(t+1)-P_j(t)≤ΔP_j,up(n)

P_j(t)-P_j(t+1)≤ΔP_j,down(n)

In the formula,. DELTA.P_j,up，ΔP_j,downRespectively the climbing rate and the descending rate of the jth unit.

(4) Output constraint of heat supply unit in heat supply period

According to the definition of the heat supply unit and the actual situation of the cogeneration development in China, mathematical modeling is respectively carried out on the back pressure type cogeneration thermal power unit and the extraction type cogeneration thermal power unit. The generated output and the thermal output of the back pressure type cogeneration thermal power unit are in a linear relationship:

P_BYJ(t)＝C_j,b×Q_j(t)

the working condition curve of the steam extraction type cogeneration thermal power generating unit is more complex, and the linear constraint formula is shown as the following formula:

Q_j(t)×C_j,b≤P_CQJ,max-Q_j(t)×C_j,v

(5) new energy output constraint

0≦P_w(t)≦P^*_w(t)

0≦P_pv(t,≦P^*_pv(t)

In the formula: p^*_w(t) wind-power time-series output at a time t when the capacity of the time-varying machine is constant, P^*_pvAnd (t) refers to the photovoltaic time series output of the time t, which is the moment when the capacity of the packaging machine is constant.

(6) Proportional constraint of new energy abandonment

If the aim of completely absorbing the new energy output is to ignore the absorbable space of the new energy at the non-load valley time, the new energy absorbed quantity calculated by the method is lower. If a certain proportion of abandoned new energy is allowed to be used for taking larger new energy consumption in the load valley period, the peak shaving margin of most periods except the low-load period is fully utilized, and the new energy consumption calculated by the method is greatly improved.

Step 5.3 model solving method

The new energy production simulation model can be mathematically reduced to solve the mixed integer linear programming problem, and the mathematical model is abbreviated as follows:

Minf(x)

s.t g_i(x)≧0(i＝1,2....m)g_i(x)≧0(i＝1,2....n)

Hj(x)＝0(j＝1,2....n)

wherein: x is the set of variables to be optimized, f (x) is the optimization objective function, g_i(x) ≧ 0 is the set of inequality constraints, h_j(x) 0 is the constraint set of equations.

Mixed integer programming is a type of mathematical programming problem in which all or a portion of the variables in the solution to the optimization problem are integers. To satisfy the requirement that variables are integers, at first glance, it seems that only rounding the obtained non-integer solution is needed, and in fact, the rounded number is not a feasible solution or an optimal solution, so a special method should be provided for solving the integer programming problem.

The core algorithm for solving mixed integer programming is the branch-and-bound method, the basic idea of which is to search all feasible solution (limited number) spaces of the constrained optimization problem. For a mixed integer programming problem A with the maximum, loosening integer variables into continuous variables to obtain a corresponding programming problem B, starting from the solution problem B, if the optimal solution does not accord with the integer condition of A, the optimal objective function of B must be the upper bound of the optimal objective function of A, the objective function value of any integer feasible solution of A is the lower bound, the variable value range is divided by taking the integer as the bound according to the optimization result of B of the problem, the branch problems B1 and B2 of B are obtained, B1 and B2 are solved, branches are continuously defined according to the result, and iteration is performed in sequence until the solution which accords with the constraint condition of A is generated in the branch problem, namely the optimal solution. The branch-and-bound method is to divide the feasible region of B into sub-regions, continuously branch, cut and bound, update the lower bound after finding a better feasible integer solution, gradually increase the lower bound and decrease the upper bound, and finally obtain the optimal solution of the problem A.

For the nonlinear mixed integer programming problem of unit output optimization, the branch-and-bound method is reasonably applied to split the original mixed integer programming problem, and simultaneously, the interior point method or the exterior point method is applied to carry out nonlinear optimization calculation. In order to ensure the solving speed and the global optimality of the result, the optimized calculation code needs to be scientifically and reasonably written, and various algorithms are organically combined together.

Common integer programming models include Set Covering, Packing and Partitioning problems (Set Covering, Packing and Partitioning), Functions with L Values (Functions with L Possible Values) with at least K of L Constraints satisfied, Fixed-cost problems (The Fixed-Charge Problem), If-Then Constraints (If-Then Constraints), and piece-wise Linear Functions (piece with Linear Functions As MIP). The If-Then constraint is mainly introduced.

In many applications: if the constraint f (x) is satisfied₁,x₂,L,x_n)>0, then the constraint g (x) must also be satisfied₁,x₂,L,x_n) ≧ 0. To ensure this, 0-variable y can be introduced, when f (x)₁,x₂,L,x_n)>When y is 0, then g (x) is required when y is 0₁,x₂,L,x_n) ≧ 0. The requirement can thus be expressed as:

where M is a sufficiently large constant that it should be guaranteed that all (x) of the other constraints in the problem are satisfied₁,x₂,L,x_n) All satisfy f (x)₁,x₂,L,x_n) M and-g (x)₁,x₂L, xn) ≧ M. It can be seen that if f>If y is 0, then y is 0, and if-g ≦ 0 or g ≧ 0, then the constraint indicates that this is the desired result.

From the perspective of power generation enterprises, the invention fully utilizes the advantages of the existing thermal power output line and combines the basis of the thermal power flexible peak regulation of the thermal power output line to form the active prospective energy development structure planning.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, it should be noted that, for those skilled in the art, many modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.

Claims (1)

1. A multi-energy complementary energy base energy configuration planning method based on thermal power is characterized by comprising the following steps:

step 1, constructing a wind power output model based on a sunpower scene, comprising the following steps of:

step 1.1, establishing a power balance scene and a peak shaving balance scene;

the power balance scene establishment steps are as follows:

1) the minimum output P of the wind power generation curve in each season according to the late peak period_EPminRanked in order of minimum output P at late peak hours_EPminAs a main characteristic index of the power balance scene, the daily wind power generation curve in each season is expressed according to P_EPminSorting from small to large to form order statistics;

2) given P_EPminDetermining a wind power generation curve set P based on the confidence level alpha_WΩ1Given P_EPminThe confidence level alpha of the wind power generation curve P is screened out_W(d) With a minimum output of P at late peak hours_EPmin(d) Ensuring that the minimum output of the late peak period of the wind power generation curve of the season is not less than P with the probability exceeding alpha_EPmin(d) Recording the set of wind power generation curves meeting the conditions as P_WΩ1：

Note P_WΩ1The power generation curve with the maximum guarantee rate of the electric quantity in the middle day is P_μ1；

3) At P_WΩ1In the method, a wind power daily power generation curve with higher daily electric quantity guarantee rate or smaller daily average output is selected as a power balance scene, wherein,

the first method is as follows: selecting the daily electric quantity guarantee rate:

at a given daily charge guarantee rate β, in set P_WΩ1Middle screening out wind power daily power generation curve P_W(d) Let P stand_W(d) Daily electricity quantity guarantee rate lambda_E(d) Not less than beta, and recording that the set of wind power daily generation curves meeting the conditions is P'_WΩ1：

Note P'_WΩ1Minimum output P at peak time of middle and late_EPminMinimum power generation curve is P_ν1；

The second method comprises the following steps: selecting average daily output:

confidence level β given mean output of days, in set P_WΩ1Middle screening out wind power daily power generation curve P_W(d) And ensuring that the average output of the wind power daily generation curve in the season is not low P with the probability of exceeding beta_DaveThe set of wind power daily generation curves satisfying the conditions is recorded as P'_WΩ1：

Note P'_WΩ1Mean output P in the middle of the day_DaveThe maximum power generation curve is P_ν1；

4) Selecting wind power output power balance scene P of each season_W1

If P 'is determined according to the mode I in 3)'_WΩ1In the presence of a catalyst satisfying P_EPminSelecting P according to the confidence level alpha and the electric quantity guarantee rate beta_EPminThe minimum power generation curve is a power balance scene to reflect the most extreme influence of wind power on the power balance of the system; when the requirement of the electric quantity guarantee rate beta cannot be met, taking a power generation curve with the maximum electric quantity guarantee rate as an electric power balance scene;

if P 'is determined in the mode of 3)'_WΩ1In the presence of a catalyst satisfying P_EPminConfidence levels of alpha and P_DaveOn the requirement of the confidence level beta, selecting a power generation curve closest to the daily average output confidence level beta as a boundary power balance scene; when the requirement of the confidence level beta cannot be met, taking a power generation curve with the minimum daily average output as a power balance scene;

5) determining the exact probability of a power balance scenario:

in the formula, N_rThe total number of the quarterly output curves;

the peak regulation balance scene establishment steps are as follows:

1) the daily wind power generation curve is expressed by delta P for each season_PDmaxSorting

Peak load demand Δ P at daily maximum_PDmaxAs a main characteristic index of a peak-shaving balance scene, a wind power daily generation curve in each season is divided into delta P_PDmaxSorting from big to small to form order statistics;

2) given Δ P_PDmaxDetermining a wind power generation curve set P based on the confidence level gamma_WΩ2

Given Δ P_PDmaxThe confidence level gamma of the wind power generation curve P is screened out_W(d) The daily maximum peak shaver requirement is delta P_PDmax(d) Make sure that the season Δ P is certain with an excess probability_PDmaxNot more than Δ P_PDmax(d) Recording the set of wind power generation curves meeting the conditions as P_WΩ2：

Recording daily minimum peak regulation demand delta P in set_PDminThe maximum power generation curve is P_μ2；

3) At P_WΩ2In the form of Δ P_PDminDetermining a wind power generation daily curve set P 'with the constraint of more than or equal to 0'_WΩ2At Δ P_PDminConstraint condition of more than or equal to 0 is from P_WΩ2Screening out a daily wind power generation curve P_W(d) The set of wind power daily generation curves satisfying the conditions is recorded as P'_WΩ2The formula is as follows:

P'_wΩ2＝{P_w(d)|ΔP_PDmin(d)≥0,P_w(d)∈P_wΩ2}

note P'_WΩ2Maximum peak load demand Δ P in mid-day_PDmaxThe maximum power generation curve is P_ν2；

4) Selecting wind power output peak regulation balance scene P of each season_W2

If P_'WΩ2Not being empty, i.e. there is a Δ P_PDmaxConfidence levels gamma and delta P_PDminSelecting delta P in the wind power daily generation curve with constraint of more than or equal to 0_PDmaxThe maximum power generation curve is a peak-load balancing scene to reflect the most extreme influence of wind power on the peak-load balancing of the system; if P'_WΩ2Is an empty set, i.e. P_WΩ2All the daily wind power generation curves cannot satisfy delta P_PDminConstraint of not less than 0, then take Δ P_PDminThe maximum wind power daily generation curve is a peak-shaving balance scene and is close to reverse peak shaving to the maximum extent;

5) determining the exact probability of a peak shaver balance scene;

in the formula, N_rThe total number of the quarterly output curves;

step 1.2, establishing a wind power daily output clustering scene, which comprises the following steps:

(1) the K-means clustering method based on the weighted Euclidean distance effectively clusters and screens a large number of wind power output scenes, and comprises the following specific steps:

1) from N wind power output curves P_mSelecting k pieces of (M is 1,2,3, …, N) as initial clustering centers M_i(i＝1,2,3,…,k)；

2) According to the requirements of practical application, determining the weight coefficient omega occupied by the wind power output in the clustering process at the load early peak time, the load late peak time, the load valley time and the load waist time_t(t＝1,2,3,…,S)；

3) Calculating each wind power output curve P in sequence_mWith respective cluster centers M_iA distance l of_miDistributing the wind power output curve to the category closest to the clustering center;

in the formula, P_mtAnd M_itRespectively is a wind power output curve P_mAnd a clustering center M_ithe output value at the moment t;

4) calculating a new clustering center in each category and re-clustering, and performing iterative calculation in a loop until a criterion function of clustering is not changed, wherein the criterion function is as follows:

wherein

In the formula, e_miIs a state variable; r_iThe method comprises the steps of collecting all wind power output curves in the ith wind power output scene;

recording the exact probability of the ith type wind power output scene as p_i，

(2) Determining typical daily wind power generation curves of various scenes, which comprises the following specific steps:

1) calculating the daily power generation curve d of wind power and other power generation curves q in the class about delta P_PDmax、ΔP_PDminIs weighted by the average euler distance S_i(d)；

In the formula, k_αIs an index of Δ P_PDmaxThe weight of (c);

2) get S_i(d) The minimum power generation curve is a scene class R_iTypical daily wind power generation curve, denoted as P_Ri；

3) Correcting scene class R_iTypical wind power daily generation curve P_RiThe generated power of (3);

the correction process is as follows:

calculating a scene class R_iWind power total generating capacity E_Ri

In the formula, P_W(dt) represents scene class R_iThe output of the medium wind power generation curve d at the tth hour;

② calculating scene class R_iTypical wind power daily generation curve P_RiExpected power generation amount

In the formula, P_Ri(t) represents P_RiThe force at time t;

calculating scene class R_iTypical wind power daily generation curve P_RiCorrected electric quantity delta E of

Fourthly, calculating scene class R_iTypical wind power daily generation curve P_RiCorrected hourly power contribution δ P

Fifthly, obtainingScene class R_iCorrected typical wind power generation daily curve P_Ri^△

Wherein the expression represents the output value at the t hour; after correction, if the wind power output at a certain moment overflows the scene wind power output range, the overflowing electric quantity at the moment is spread to other moments;

step 2, constructing a photovoltaic output model based on the sunoutput scene, comprising the following steps:

step 2.1, when the power balance of a system containing photovoltaic power generation needs to be evaluated, directly replacing a power balance scene with a photovoltaic scene with the minimum output in a clustering scene; taking the maximum output P of each season day_DmaxThe maximum photovoltaic daily power generation curve is used as a peak-load balancing scene of photovoltaic power generation; the exact probability calculation formula of the photovoltaic peak-shaving balance scene is the same as that of wind power;

2.2, clustering photovoltaic daily power generation curves in each season into a plurality of typical scenes according to photovoltaic historical output data by adopting a forward scene reduction method based on a Kantorovich Distance; through repeated iteration, a scene with the minimum KD distance to other scenes is selected from the original scene set omega and is placed into the scene set omega ', and the KD distances of the two scene sets omega and omega' are defined as follows:

in the formula, s and s 'are scenes in a scene set omega and omega' respectively; p is a radical of_sAnd p_s'Probability of scenes s and s 'in Ω and Ω', respectively; c (s, s') is a non-negative, continuous, symmetrical distance function; μ (s, s ') is the probability product of scene s and s';

the forward scene reduction based on the KD distance comprises the following steps:

(1) determining a scene requiring clipping: eliminating scenes omega meeting the following conditions_s'

In the formula, c (ω)_s',ω_m) The distance between the two photovoltaic sunrise force curves;

(2) total number of changed scenes: n-1; and screen out and cut scene omega_s'Nearest scene omega_sI.e. by

(3) Changing and culling scene omega_s'Nearest scene omega_sTo ensure that the sum of the probabilities of all remaining scenes is 1;

P_s＝P_s+P_s’

(4) the iterative calculation is circulated until the residual scene number meets the set target scene number N_sUntil the requirements are met; each representative scene omega_sHas a probability of p_s；

Step 3, constructing a thermal power generating unit output model, comprising the following steps:

constructing a condensing thermal power generating unit output model;

constructing a back pressure thermal power generating unit output model;

constructing a steam extraction type thermal power generating unit output model;

step 4, constructing a historical load model of the original thermal power circuit, comprising the following steps:

determining a reference daily load curve:

according to a typical working day and rest day curve every month, and considering typical holidays separately, selecting representative days every month to eliminate the influence of abnormal factors; or comprehensively analyzing and comparing typical curves of the month in each year in history to determine a representative curve of the month; the following analytical methods were used for each typical day data:

let T be 24, denote the number of time segments, and let the load data of the day to be analyzed be l_i(i 1,2,.. and T), and a daily maximum load of l₀In 1 with₀To l_iPerforming per unit to obtain the load curve d of the day_i(i ═ 1,2, · and T), the following relationships are established:

l₀＝max l_i 1≦i≦T

d_i＝l_i/l₀

and 5, building a digestion model based on a time sequence production simulation method, wherein the digestion model comprises the following steps:

step 5.1 determining the objective function

The objective function in the optimization cycle is:

in the formula: t represents the total length of time; t is the simulation time step length; p_w(t) is the wind power output of the wind-light-fire multi-energy base in the time period t, P_pv(t) generating power by solar energy in the wind, light and fire multi-energy base at the time t;

step 5.2, determining constraint conditions

(1) Regional load balancing constraints

P_j(t)×S_j(t)+P_W(t)+P_PV(t)＝P_I(t)

In the formula: p_j(t)×S_j(t) is the sum of the powers of the conventional units;

(2) unit output constraint

0≤ΔP_j(t)≤[P_j,max(t)-P_j,min(t)]×S_j(t)

P_j(t)＝P_j,min(t)×S_j(t)+ΔP_j(t)

In the formula, P_j(t) optimizing the power of the conventional unit;

(3) unit optimized power ramp rate constraint

P_j(t+1)-P_j(t)≤ΔP_j,up(n)

P_j(t)-P_j(t+1)≤ΔP_j,down(n)

In the formula,. DELTA.P_j,up，ΔP_j,downRespectively the climbing rate and the descending rate of the jth unit;

(4) output constraint of heat supply unit in heat supply period

The generated output and the thermal output of the back pressure type cogeneration thermal power unit are in a linear relationship:

P_BYJ(t)＝C_j,b×Q_j(t)

the linear constraint formula of the steam extraction type cogeneration thermal power unit is shown as the following formula:

Q_j(t)×C_j,b≤P_CQJ,max-Q_j(t)×C_j,v

(5) new energy output constraint

0≦P_w(t)≦P^*_w(t)

0≦P_pv(t,≦P^*_pv(t)

In the formula: p^*_w(t) wind-power time-series output at a time t when the capacity of the time-varying machine is constant, P^*_pv(t) a photovoltaic time series output at a time t when the capacity of the fashion machine is constant;

(6) proportional constraint of new energy abandonment

Allowing a certain proportion of new energy to be abandoned in the load valley period so as to replace larger new energy consumption;

step 5.3, model solving method

The new energy production simulation model is abbreviated as follows:

Minf(x)

s.t g_i(x)≧0(i＝1,2....m)g_i(x)≧0(i＝1,2....n)

Hj(x)＝0(j＝1,2....n)

wherein: x is the set of variables to be optimized, f (x) is the optimization objective function, g_i(x) ≧ 0 is the set of inequality constraints, h_j(x) 0 is the set of equality constraints;

and 6, obtaining a power supply combination scheme of the multi-energy complementary energy base according to a digestion model based on a time sequence production simulation method.

Images