技术领域Technical field
本发明属于虚拟电厂技术领域，尤其涉及一种虚拟电厂优化调度方法。The invention belongs to the technical field of virtual power plants, and in particular relates to a virtual power plant optimization dispatching method.
背景技术Background technique
随着能源危机和环境污染的日益严重，清洁能源的开发和利用已受到各国的高度重视。然而风能的随机性和波动性使得大规模风电场并网下的电力系统运行中不确定因素增多。近年来，中国风光装机容量不断增加，但因其具有单机容量小、地域分散、并网具有较大的随机性和波动性的特性，大规模接入会给电网的可靠性带来巨大的挑战。为了有效解决利用可再生能源所带来的威胁，虚拟电厂(virtual power plant，VPP)应运而生。VPP是通过将各种分布式电源、储能系统营，缓解了可再生能源接入电网造成波动的同时，还增加了VPP各组成部分的经济效益。As the energy crisis and environmental pollution become increasingly serious, the development and utilization of clean energy has attracted great attention from all countries. However, the randomness and volatility of wind energy increase uncertainties in the operation of power systems with large-scale wind farms connected to the grid. In recent years, China's wind and solar installed capacity has continued to increase. However, due to its small single-machine capacity, geographical dispersion, and large randomness and volatility in grid connection, large-scale integration will bring huge challenges to the reliability of the power grid. . In order to effectively solve the threats posed by the use of renewable energy, virtual power plant (VPP) came into being. By integrating various distributed power supplies and energy storage systems, VPP not only alleviates the fluctuations caused by the integration of renewable energy into the power grid, but also increases the economic benefits of each component of VPP.
国内外对于含风电场电力系统调度模型方面做了大量研究。虚拟电厂存在源荷不确定性的问题会导致虚拟电厂在获取收益同时也要面对一定的风险损失。随着风电并网容量的不断增加，传统的确定性优化调度方法已难以满足电力系统安全运行要求。且因调节能力不足会导致大量的弃风、弃光问题。A lot of research has been done at home and abroad on the dispatch model of power systems including wind farms. The problem of source load uncertainty in virtual power plants will cause virtual power plants to face certain risks and losses while gaining profits. With the continuous increase of wind power grid-connected capacity, traditional deterministic optimization dispatch methods are no longer able to meet the safe operation requirements of the power system. In addition, insufficient adjustment capabilities will lead to a large number of wind and light abandonment problems.
发明内容Contents of the invention
本发明目的在于提供一种虚拟电厂优化调度方法，以解决虚拟电厂存在源荷不确定性的问题会导致虚拟电厂在获取收益同时也要面对一定的风险损失的技术问题。The purpose of the present invention is to provide an optimal dispatching method for a virtual power plant to solve the technical problem that the uncertainty of source and load in the virtual power plant will cause the virtual power plant to face certain risks and losses while obtaining profits.
为实现上述目的，本发明的一种虚拟电厂优化调度方法的具体技术方案如下：In order to achieve the above objectives, the specific technical solution of a virtual power plant optimization dispatching method of the present invention is as follows:
一种虚拟电厂优化调度方法，包括以下步骤：An optimized dispatching method for a virtual power plant, including the following steps:
步骤S1、虚拟电厂(virtual power plant，VPP)机组建模：Step S1. Virtual power plant (VPP) unit modeling:
步骤S1-1、风电机组Step S1-1, wind turbine
风电机组的输出功率为：The output power of the wind turbine is:
式中，PWPP(t)为风电机组在t时刻的输出功率；rt为在t时刻的自然风风速；ri为风电机组的切入风速；ro为风电机组的切出风速；rq为风电机组的额定风速；pq为风电机组的额定输出功率。In the formula, PWPP (t) is the output power of the wind turbine at time t; rt is the natural wind speed at time t; ri is the cut-in wind speed of the wind turbine; ro is the cut-out wind speed of the wind turbine; rq is the rated wind speed of the wind turbine; pq is the rated output power of the wind turbine.
步骤S1-2、光伏机组Step S1-2, photovoltaic unit
光伏机组的输出功率为：The output power of the photovoltaic unit is:
PPV(t)＝ηPVSPVθt (2)PPV (t)＝ηPV SPV θt (2)
式中，PPV(t)为光伏机组在t时刻的输出功率；ηPV为光伏板的转换效率；SPV为光伏板的面积；θt为在t时刻的光照强度；In the formula, PPV (t) is the output power of the photovoltaic unit at time t; etaPV is the conversion efficiency of the photovoltaic panel; SPV is the area of the photovoltaic panel; θt is the light intensity at time t;
步骤S1-3、微型燃气轮机Step S1-3, micro gas turbine
微型燃气轮机发电成本为：The power generation cost of micro gas turbine is:
式中，CMGT(t)为微型燃气轮机在t时段的发电成本；PMGT,t为微型燃气轮机在t时段的输出功率；a为微型燃气轮机发电成本的二次项系数；b为微型燃气轮机发电成本的一次项系数；c为微型燃气轮机发电成本的常数项；In the formula, CMGT (t) is the power generation cost of the micro gas turbine in period t; PMGT,t is the output power of the micro gas turbine in period t; a is the quadratic coefficient of the power generation cost of the micro gas turbine; b is the power generation cost of the micro gas turbine. The coefficient of the linear term; c is the constant term of the power generation cost of the micro gas turbine;
步骤S1-4、储能系统Step S1-4, energy storage system
储能系统的充放电双向特性如下：The two-way charging and discharging characteristics of the energy storage system are as follows:
-PES,max≤PES,e,t≤PES,man (4)-PES,max ≤PES,e,t ≤PES,man (4)
其中，PES,e,t为储能系统在t时段的充放电功率；PES,man为储能系统充放电功率的最大值；Among them, PES,e,t is the charge and discharge power of the energy storage system in period t; PES,man is the maximum value of the charge and discharge power of the energy storage system;
步骤S2、VPP的收益与成本建模：Step S2, VPP revenue and cost modeling:
步骤S2-1、虚拟电厂的购售电收益Step S2-1. Revenue from electricity purchase and sale of the virtual power plant
其中，εo(t)和εt(t)分别为虚拟电厂在t时刻的售电价格和购电价格；PVPP,o(t)和PVPP,t(t)分别为虚拟电厂在t时刻的售电量和购电量；Among them, εo (t) and εt (t) are the electricity sales price and power purchase price of the virtual power plant at time t respectively; PVPP,o (t) and PVPP,t (t) are the electricity sales price and purchase price of the virtual power plant at time t respectively. Electricity sales and purchases at any time;
步骤S2-2、虚拟电厂的环境惩罚成本Step S2-2. Environmental penalty cost of virtual power plant
其中，xOR,n为数值取0或1的变量，在值为1时，n机组开启，在值为0时，n机组关闭；ep(m)为机组在功率P下m类污染物的排放量；f(m)为m类排放物的罚款金额；POR,n(t)为t时刻机组的功率；Among them, xOR, n is a variable with a value of 0 or 1. When the value is 1, the n unit is turned on, and when the value is 0, the n unit is turned off; ep (m) is the m type of pollutants under the power P of the unit. The amount of emissions; f(m) is the fine amount for type m emissions; POR,n (t) is the power of the unit at time t;
步骤S2-3、虚拟电厂所选机组的租赁成本Step S2-3. Rental cost of the selected units of the virtual power plant
其中，xPV,i、xWPP,j、xOR,k、xES,e分别为光伏机组i、风电机组j、微燃机组k、第e台储能系统的0/1变量，该机组被虚拟电厂租赁取值为1，该机组未被虚拟电厂租赁取值为0；PES,e,ec(t)、PES,e,ed(t)分别为第e台储能系统在时刻的充电、放电功率；PIL(t)电厂中可中断负荷在t时刻的功率；CIL为可中断负荷单位电量的补偿成本；LPPV,i、LPWPP,j、OCES,e,op、LPES,e分别为光伏机组i、风电机组j、微燃机组k、第e台储能系统的单位输出功率的租赁费用；Among them, xPV,i , xWPP,j , xOR,k , xES,e are the 0/1 variables of the photovoltaic unit i, the wind turbine unit j, the micro-combustion unit k, and the e-th energy storage system respectively. The value of being leased by a virtual power plant is 1, and the value of the unit not being leased by a virtual power plant is 0; PES,e,ec (t) and PES,e,ed (t) are respectively the e-th energy storage system at time charging and discharging power; PIL (t) the power of the interruptible load in the power plant at time t; CIL is the compensation cost per unit of power of the interruptible load; LPPV,i , LPWPP,j ,OC ES,e,op , LPES,e are the rental costs per unit output power of photovoltaic unit i, wind turbine unit j, micro-combustion unit k, and e-th energy storage system respectively;
步骤S2-4、虚拟电厂所选机组的运行维护成本Step S2-4. Operation and maintenance costs of the selected units of the virtual power plant
其中，OCMGT,k,op、OCES,e,op分别为微燃机组k、第e台储能系统的单位输出功率的运行费用；Among them, OCMGT,k,op and OCES,e,op are the operating costs per unit output power of the micro-combustion unit k and the e-th energy storage system respectively;
步骤S2-5、虚拟电厂的条件风险价值(conditional value-at-risk，CVaR)Step S2-5, conditional value-at-risk (CVaR) of the virtual power plant
其中，v为边界值；b为提前给定的置信度，取0.5；dv为决策向量；Lf为损失函数，其值取虚拟电厂的收益负值；Among them, v is the boundary value; b is the confidence level given in advance, which is 0.5; dv is the decision vector; Lf is the loss function, and its value is the negative value of the virtual power plant's income;
步骤S3、虚拟电厂VPP的运行约束条件：Step S3, operating constraints of the virtual power plant VPP:
步骤S3-1、微燃机运行约束Step S3-1. Micro-gas turbine operation constraints
PMGT,i,min≤PMGT,i(t)≤PMGT,i,max (10)PMGT,i,min ≤PMGT,i (t)≤PMGT,i,max (10)
其中，PMGT,i(t)、PMGT,i(t-1)分别为微燃机组i在t时刻和t-1时刻的功率；PMGT,i,min、PMGT,i,max分别为微燃机组i的最小、最大输出功率；分别为微燃机组i的向下、向上爬坡功率；Among them, PMGT,i (t) and PMGT,i (t-1) are the power of micro-combustion unit i at time t and time t-1 respectively; PMGT,i,min and PMGT,i,max respectively are the minimum and maximum output power of micro-combustion unit i; are the downward and upward climbing powers of micro-combustion unit i respectively;
步骤S3-2、储能系统的荷电状态和充、放电约束Step S3-2. State of charge and charge and discharge constraints of the energy storage system
(1-Cd,max)Si,man≤Si(t)≤Si,man (13)(1-Cd,max )Si,man ≤Si (t)≤Si,man (13)
0≤PES,i,ec(t)≤cec(t)PES,i,ec,max (14)0≤PES,i,ec (t)≤cec (t)PES,i,ec,max (14)
0≤PES,i,ed(t)≤ced(t)PES,i,ed,max (15)0≤PES,i,ed (t)≤ced (t)PES,i,ed,max (15)
cec(t)+ced(t)≤1 (16)cec (t)+ced (t)≤1 (16)
其中，Si(t)、Si(t-1)分别为第i台储能系统在t时刻、t-1时刻的荷电状态；γec、γed分别为储能系统的充电、放电效率系数；Si,man为第i台储能系统荷电状态的上限值；Cd,max为储能系统的最大放电深度；PES,i,ec(t)、PES,i,ed(t)分别为第i台储能系统在t时刻的充电、放电功率；PES,i,ec,max、PES,i,ed,max分别为第i台储能系统的最大允许充电、最大允许放电功率：cec(t)、ced(t)分别为储能系统在t时刻是否处于充电、放电的状态值，是则取1，否则取0，两者不可以同时为1；Among them, Si (t) and Si (t-1) are the state of charge of the i-th energy storage system at time t and time t-1 respectively; γec and γed are the charging and discharging of the energy storage system respectively. Efficiency coefficient; Si,man is the upper limit of the state of charge of the i-th energy storage system; Cd,max is the maximum discharge depth of the energy storage system; PES,i,ec (t), PES,i, ed (t) are the charging and discharging power of the i-th energy storage system at time t respectively; PES,i,ec,max and PES,i,ed,max are respectively the maximum allowable charging of the i-th energy storage system. , Maximum allowable discharge power: cec (t), ced (t) are respectively whether the energy storage system is in the state of charging and discharging at time t. If so, it takes 1, otherwise it takes 0. Both cannot be 1 at the same time. ;
步骤S3-3、可中断负荷约束Step S3-3, interruptible load constraints
0≤PIL(t)≤sfILPL(t) (17)0≤PIL (t)≤sfIL PL (t) (17)
其中，PL(t)为虚拟电厂中的电力负荷；sfIL为虚拟电厂中电力负荷中的可中断负荷的比例系数；Among them, PL (t) is the electric load in the virtual power plant; sfIL is the proportional coefficient of the interruptible load in the electric load in the virtual power plant;
步骤S3-4、功率平衡约束Step S3-4, power balance constraints
步骤S3-5、备用容量约束Step S3-5, spare capacity constraints
预先设置备用容量约束，具体的约束公式为：Set the spare capacity constraint in advance. The specific constraint formula is:
其中，R+(t)为VPP在t时刻所需要的上旋转备用容量；R+(t)为VPP在t时刻所需要的下旋转备用容量；Among them, R+ (t) is the up-spinning reserve capacity required by VPP at time t; R+ (t) is the down-spinning reserve capacity required by VPP at time t;
步骤S3-6、系统风险约束Step S3-6, system risk constraints
根据调度人员对于风险性的要求，令风险系数小于某一阈值：According to the dispatcher's risk requirements, the risk coefficient is made less than a certain threshold:
Rf(P)≤δ (21)Rf (P)≤δ (21)
其中Rf(P)为风险系数，风险系数Rf(P)定义如下：Among them, Rf (P) is the risk coefficient, and the risk coefficient Rf (P) is defined as follows:
Rf(P)＝max{max(Ps,t,Pn,t)|t＝1,2,…,24} (22)Rf (P)＝max{max(Ps,t ,Pn,t )|t＝1,2,…,24} (22)
Ppl,t＝Pal,t-Pw,t (24)Ppl,t =Pal,t -Pw,t (24)
Si,t＝min(Pi,t,max-Pi,t,uiT10) (25)Si,t =min(Pi,t,max -Pi,t ,ui T10 ) (25)
Ni,t＝min(Pi,t-Pi,t,max,diT10) (26)Ni,t =min(Pi,t -Pi,t,max ,di T10 ) (26)
其中，Rf(P)表示调度方案在所有调度时段正或负旋转备用不足概率最大值；Si,t、Ni,t别表示机组i在时刻t可提供的正负旋转备用容量；Ppl,t、Pal,t、Pw,t分别表示系统时刻t的净负荷、实际负荷、风电场总出力；ui、di分别表示机组i的分钟级上下爬坡速率；T10表示旋转备用响应时间，取值10min；E(Ppl,t)表示Ppl,t的期望值；Among them, Rf (P) represents the maximum probability of insufficient positive or negative spinning reserve of the dispatching plan in all scheduling periods; Si,t and Ni,t respectively represent the positive and negative spinning reserve capacity that unit i can provide at time t; Ppl,t , Pal,t and Pw,t respectively represent the net load, actual load and total wind farm output of the system at time t; ui and di respectively represent the minute-level up and down ramp rate of unit i; T10 represents Spinning reserve response time, value 10min; E(Ppl,t ) represents the expected value of Ppl,t ;
步骤S4、场景选择：Step S4, scene selection:
风光随机场景生成Scenery random scene generation
采用随机模拟法来计算风险价值；Use stochastic simulation method to calculate value at risk;
概率密度分布函数分别为：The probability density distribution functions are:
式中：Γ为Gamma函数；α和β为形状系数；r为太阳辐照度；rmax为太阳辐照度的最大值；P为实时风电功率；k为形状参数；c为分布系数；In the formula: Γ is the Gamma function; α and β are shape coefficients; r is the solar irradiance; rmax is the maximum value of the solar irradiance; P is the real-time wind power power; k is the shape parameter; c is the distribution coefficient;
根据风光概率密度分布，利用抽样算法获得大量概率预测场景；Based on the scenery probability density distribution, a sampling algorithm is used to obtain a large number of probability prediction scenarios;
步骤S5、确认目标函数：Step S5: Confirm the objective function:
其中，ηi为场景i发生的概率；L为权重系数，其表现VPP管理者的风险偏好，L≥0；当L取较小的值时，管理者的选择比较激进，希望获取较高的收益，但同时会面对较大的风险损失；当取较大的值时，管理者选择比较保守，获取的收益较低，但同时会面对的风险损失较小；Among them, ηi is the probability of scenario i occurring; L is the weight coefficient, which expresses the risk preference of VPP managers, L≥0; when L takes a smaller value, the manager’s choice is more aggressive, hoping to obtain higher returns income, but at the same time they will face greater risk losses; when taking a larger value, the manager will choose to be more conservative and obtain lower income, but at the same time face smaller risk losses;
步骤S6、结果分析。Step S6: Result analysis.
进一步，所述步骤S4中，利用抽样算法获得大量概率预测场景，拉丁超立方抽样(Latin hypercube sampling，LHS)主要包括区间抽样和样本相关性排序：Furthermore, in step S4, a sampling algorithm is used to obtain a large number of probability prediction scenarios. Latin hypercube sampling (LHS) mainly includes interval sampling and sample correlation sorting:
步骤S4-1、抽样Step S4-1, sampling
假设N个服从某概率分布的独立随机变量Xn，的概率累计函数Zn为：Assume that N independent random variables Xn obey a certain probability distribution, and the probability accumulation functionZn is:
Zn＝fn(Xn),n＝1,2,…,N (30)Zn =fn (Xn ),n＝1,2,…,N (30)
设抽样规模为M，将取值区间[0，1]均分为M个区间，并从每个等距区间中随机生成一个抽样值，每个区间只生成一个抽样值；抽样值可表示为：Assume the sampling scale is M, divide the value interval [0, 1] equally into M intervals, and randomly generate a sampling value from each equally spaced interval, and only generate one sampling value for each interval; the sampling value can be expressed as :
其中，Um为m个区间的抽样值；U为0或1的随机数；Among them, Um is the sampling value of m intervals; U is a random number of 0 or 1;
得到M个抽样值后，利用反函数可得到Xn的第m个抽样值Xm为：After obtaining M sample values, the mth sample value Xm of Xn can be obtained by using the inverse function:
变量Xn抽样结束后，得到M个抽样值；当N个变量全部抽样结束后，即构成一个N×M阶的初始样本矩阵Xs；After the sampling ofvariables
步骤S4-2、排序Step S4-2, sorting
获得初始样本矩阵Xs后，采用Cholesky分解法对采样矩阵重新排序，降低矩阵内元素的相关性，提高抽样精度，最终获得排序后的N×M阶样本矩阵XN；After obtaining the initial sample matrix Xs, the Cholesky decomposition method is used to reorder the sampling matrix to reduce the correlation of elements within the matrix and improve the sampling accuracy, and finally obtain the sorted N×M order sample matrix XN;
样本矩阵XN中行元素代表调度周期内某时间区间风/光出力抽样值，列元素代表一个调度周期内各时间区间风光出力的抽样值，即形成含M个风/光随机预测场景的场景集以及各自场景出现的概率；The row elements in the sample matrix The probability of occurrence of each scenario;
采用K-均值聚类算法对数量削减后的风光随机场景集进行相关性聚类分析。The K-means clustering algorithm was used to perform correlation clustering analysis on the reduced scenery random scene set.
进一步，所述步骤S4中K-均值算法包括以下步骤：首先在某一数据集中随机选取K个样本作为初始聚类中心；然后计算所有样本点与聚类中心的距离，按照距离最近原则，将样本点划分到相应的簇中；在每次迭代过程中重新计算得到新的聚类中心：当聚类中心不再变化或者达到指定条件时，输出聚类结果；Further, the K-means algorithm in step S4 includes the following steps: first, randomly select K samples from a certain data set as the initial cluster center; then calculate the distance between all sample points and the cluster center, and according to the principle of closest distance, The sample points are divided into corresponding clusters; a new cluster center is recalculated during each iteration: when the cluster center no longer changes or reaches the specified condition, the clustering result is output;
设数据集D(x1，2，...，xn)中含n个样本，随机选取K个初始聚类中心后，则D中任意两个样本间的距离为：Assume that the data set D (x1, 2, ..., xn) contains n samples, and after randomly selecting K initial clustering centers, the distance between any two samples in D is:
其中，d(xj,xk)—样本点与间的距离，距离越小表示两个样本越相似；Among them, d(xj ,xk )—the distance between sample points. The smaller the distance, the more similar the two samples are;
当初始聚类中心选定后，算法迭代过程将自动运行，而选择不同初始聚类中心对聚类过程稳定性和聚类结果有较大影响；When the initial clustering center is selected, the algorithm iteration process will run automatically, and choosing different initial clustering centers has a greater impact on the stability of the clustering process and the clustering results;
计算样本间的误差平方和来优化初始聚类中心；数据集D中样本的误差平方和SD为：Calculate the sum of squared errors between samples to optimize the initial cluster center; the sum of squared errors SD of the samples in data set D is:
其中，cl为第l个初始聚类中心；SD越小表示样本点越密集。Among them, cl is the l-th initial clustering center; the smaller SD means the denser the sample points.
根据式(34)计算数据集D中样本的误差平方和SD，并选取其中最小值作为新的初始聚类中心，将上式迭代K次后获得K个优化初始聚类中心；得到新的聚类中心后对风光数据集进行聚类优化计算，将N×N个风光随机场景进一步聚类优化为N个风光典型预测场景。Calculate the sum of squared errors SD of the samples in data set D according to equation (34), and select the minimum value as the new initial clustering center. After iterating the above equation K times, K optimized initial clustering centers are obtained; a new clustering center is obtained. After clustering the cluster center, the scenery data set is subjected to clustering optimization calculation, and the N×N random scenery scenes are further clustered and optimized into N typical scenery prediction scenes.
本发明的一种虚拟电厂优化调度方法具有以下优点：A virtual power plant optimization dispatching method of the present invention has the following advantages:
1.采用随机优化处理风光不确定性问题，通过拉丁超立方抽样生成大量随机风光场景，并在充分考虑风光相关性和分布随机特性的基础上，K-均值聚类算法对随机场景进行降维优化，获得风电、太阳直接辐照度典型预测场景；1. Stochastic optimization is used to deal with the problem of scenery uncertainty, and a large number of random scenery scenes are generated through Latin hypercube sampling. On the basis of fully considering the correlation of scenery and the random characteristics of distribution, the K-means clustering algorithm is used to reduce the dimensionality of the random scenes. Optimize and obtain typical prediction scenarios of wind power and direct solar irradiance;
2.采用条件风险价值作为风险度量的指标，以运行收益最大化和风险损失最小化为优化目标，建立基于条件风险价值点(CVaR)风险控制的多电源虚拟电厂机组动态聚合优化模型；2. Use conditional value at risk as an indicator of risk measurement, take maximization of operating income and minimization of risk losses as optimization goals, and establish a dynamic aggregation optimization model of multi-power virtual power plant units based on conditional value at risk (CVaR) risk control;
3.针对不同运行场景求解目标函数，获得各机组最优出力计划，验证所提调度策略对提升系统经济性、促进风电并网消纳的有效性和可行性。3. Solve the objective function for different operating scenarios, obtain the optimal output plan of each unit, and verify the effectiveness and feasibility of the proposed dispatch strategy in improving system economy and promoting wind power grid connection and consumption.
附图说明Description of drawings
图1为本发明实施例中风光点时间分布。Figure 1 shows the time distribution of light points in an embodiment of the present invention.
图2为本发明实施例中不同L值下的系统总成本。Figure 2 shows the total system cost under different L values in the embodiment of the present invention.
图3为本发明实施例中负荷需求及可再生能源能力。Figure 3 shows the load demand and renewable energy capacity in the embodiment of the present invention.
图4为本发明实施例中VPP日出力计划(L＝0.50)。Figure 4 shows the VPP daily output plan (L=0.50) in the embodiment of the present invention.
图5为本发明一种虚拟电厂优化调度方法的系统流程图。Figure 5 is a system flow chart of a virtual power plant optimization dispatching method according to the present invention.
具体实施方式Detailed ways
为了更好地了解本发明的目的、结构及功能，下面结合附图，对本发明一种虚拟电厂优化调度方法做进一步详细的描述。In order to better understand the purpose, structure and function of the present invention, a virtual power plant optimization dispatching method of the present invention will be described in further detail below in conjunction with the accompanying drawings.
本发明提出在日前从现有的多电源机组中选择可调容量合适的机组组建虚拟电厂，就能有效规避风险损失，使虚拟电厂的收益最大化。本专利采用条件风险价值作为风险度量的指标，以运行收益最大化和风险损失最小化为优化目标，建立基于条件风险价值点(CVaR)风险控制的多电源虚拟电厂机组动态聚合优化模型。首先，采用场景技术来模拟虚拟电厂中源荷的不确定性；然后，在此基础上研究了虚拟电厂管理者的风险偏好对虚拟电厂机组选择的影响，以及环境惩罚成本和购电电价对虚拟电厂机组选择的影响；最后，通过算例仿真验证了该模型的正确性。结果表明，当虚拟电厂管理者选择合适的风险偏好时，虚拟电厂机组的动态聚合模型可有效降低虚拟电厂的风险损失以及提高虚拟电厂供能的稳定性。The present invention proposes that by selecting units with appropriate adjustable capacity from existing multi-power supply units to form a virtual power plant, risks and losses can be effectively avoided and the benefits of the virtual power plant can be maximized. This patent uses conditional value at risk as an indicator of risk measurement, and takes maximization of operating income and minimization of risk losses as optimization goals to establish a dynamic aggregation optimization model of multi-power virtual power plant units based on conditional value at risk (CVaR) risk control. First, scenario technology is used to simulate the uncertainty of the source load in the virtual power plant; then, on this basis, the impact of the virtual power plant manager's risk preference on the virtual power plant unit selection is studied, as well as the impact of environmental penalty costs and power purchase prices on the virtual power plant. The influence of the selection of power plant units; finally, the correctness of the model is verified through case simulation. The results show that when the virtual power plant manager chooses the appropriate risk preference, the dynamic aggregation model of the virtual power plant unit can effectively reduce the risk loss of the virtual power plant and improve the stability of the virtual power plant's energy supply.
一种虚拟电厂优化调度方法，包括以下步骤：An optimized dispatching method for a virtual power plant, including the following steps:
步骤一：虚拟电厂(virtual power plant，VPP)机组建模Step 1: Virtual power plant (VPP) unit modeling
1)风电机组1) Wind turbines
自然风风速的不确定性导致了风电机组输出功率的不确定性，风电机组的输出功率为：The uncertainty of the natural wind speed leads to the uncertainty of the output power of the wind turbine. The output power of the wind turbine is:
式中，PWPP(t)为风电机组在t时刻的输出功率；rt为在t时刻的自然风风速；ri为风电机组的切入风速；ro为风电机组的切出风速；rq为风电机组的额定风速；pq为风电机组的额定输出功率。In the formula, PWPP (t) is the output power of the wind turbine at time t; rt is the natural wind speed at time t; ri is the cut-in wind speed of the wind turbine; ro is the cut-out wind speed of the wind turbine; rq is the rated wind speed of the wind turbine; pq is the rated output power of the wind turbine.
2)光伏机组2) Photovoltaic unit
光照强度的不确定性导致了光伏机组输出功率的不确定性，光伏机组的输出功率为：The uncertainty of light intensity leads to the uncertainty of the output power of the photovoltaic unit. The output power of the photovoltaic unit is:
PPV(t)＝ηPVSPVθtPPV (t)＝ηPV SPV θt
式中，PPV(t)为光伏机组在t时刻的输出功率；ηPV为光伏板的转换效率；SPV为光伏板的面积；θt为在t时刻的光照强度。In the formula, PPV (t) is the output power of the photovoltaic unit at time t; etaPV is the conversion efficiency of the photovoltaic panel; SPV is the area of the photovoltaic panel; θt is the light intensity at time t.
3)微型燃气轮机3) Micro gas turbine
在VPP中通常会配置微型燃气轮机来平抑可再生能源出力的波动性，从而对外输出稳定的电能，和常规电厂行使一样的职能。其发电成本为：Micro gas turbines are usually configured in VPPs to smooth out the volatility of renewable energy output, thereby outputting stable electric energy and performing the same functions as conventional power plants. Its power generation cost is:
CMGT(t)为微型燃气轮机在t时段的发电成本；PMGT,t为微型燃气轮机在t时段的输出功率；a为微型燃气轮机发电成本的二次项系数；b为微型燃气轮机发电成本的一次项系数；c为微型燃气轮机发电成本的常数项。CMGT (t) is the power generation cost of micro gas turbine in period t; PMGT,t is the output power of micro gas turbine in period t; a is the quadratic term coefficient of micro gas turbine power generation cost; b is the first term of micro gas turbine power generation cost Coefficient; c is the constant term of the micro gas turbine power generation cost.
4)储能系统4) Energy storage system
储能系统也能够平抑可再生能源出力的波动性，只是受价格的限制，不能替代微燃机在VPP中的作用。储能系统还能起到削峰填谷的作用，使VPP获得更高的收益。Energy storage systems can also smooth out the volatility of renewable energy output, but are limited by price and cannot replace the role of micro-gas turbines in VPP. The energy storage system can also play a role in peak shaving and valley filling, allowing VPP to obtain higher returns.
储能系统的充放电双向特性如下：The two-way charging and discharging characteristics of the energy storage system are as follows:
-PES,max≤PES,e,t≤PES,man-PES,max ≤PES,e,t ≤PES,man
其中，PES,e,t为储能系统在t时段的充放电功率；PES,man为储能系统充放电功率的最大值。Among them, PES,e,t is the charge and discharge power of the energy storage system in period t; PES,man is the maximum value of the charge and discharge power of the energy storage system.
步骤二：VPP的收益与成本建模Step 2: Modeling the benefits and costs of VPP
1)虚拟电厂的购售电收益1) Revenue from electricity purchase and sale of virtual power plants
其中，εo(t)和εt(t)分别为虚拟电厂在t时刻的售电价格和购电价格；PVPP,o(t)和PVPP,t(t)分别为虚拟电厂在t时刻的售电量和购电量。Among them, εo (t) and εt (t) are the electricity sales price and power purchase price of the virtual power plant at time t respectively; PVPP,o (t) and PVPP,t (t) are the electricity sales price and purchase price of the virtual power plant at time t respectively. Electricity sales and purchases at any time.
2)虚拟电厂的环境惩罚成本2) Environmental penalty costs of virtual power plants
其中，xOR,n为数值取0或1的变量，在值为1时，n机组开启，在值为0时，n机组关闭；ep(m)为机组在功率P下m类污染物的排放量；f(m)为m类排放物的罚款金额；POR,n(t)为t时刻机组的功率。Among them, xOR, n is a variable with a value of 0 or 1. When the value is 1, the n unit is turned on, and when the value is 0, the n unit is turned off; ep (m) is the m type of pollutants under the power P of the unit. The amount of emissions; f(m) is the fine amount for type m emissions; POR,n (t) is the power of the unit at time t.
3)虚拟电厂所选机组的租赁成本3) Rental cost of the selected units of the virtual power plant
其中，xPV,i、xWPP,j、xOR,k、xES,e分别为光伏机组i、风电机组j、微燃机组k、第e台储能系统的0/1变量，该机组被虚拟电厂租赁取值为1，该机组未被虚拟电厂租赁取值为0；PES,e,ec(t)、PES,e,ed(t)分别为第e台储能系统在时刻的充电、放电功率；PIL(t)电厂中可中断负荷在t时刻的功率；CIL为可中断负荷单位电量的补偿成本；LPPV,i、LPWPP,j、OCES,e,op、LPES,e分别为光伏机组i、风电机组j、微燃机组k、第e台储能系统的单位输出功率的租赁费用。Among them, xPV,i , xWPP,j , xOR,k , xES,e are the 0/1 variables of the photovoltaic unit i, the wind turbine unit j, the micro-combustion unit k, and the e-th energy storage system respectively. The value of being leased by a virtual power plant is 1, and the value of the unit not being leased by a virtual power plant is 0; PES,e,ec (t) and PES,e,ed (t) are respectively the e-th energy storage system at time charging and discharging power; PIL (t) the power of the interruptible load in the power plant at time t; CIL is the compensation cost per unit of power of the interruptible load; LPPV,i , LPWPP,j ,OC ES,e,op , LPES,e are the rental costs per unit output power of photovoltaic unit i, wind turbine unit j, micro-combustion unit k, and e-th energy storage system respectively.
4)虚拟电厂所选机组的运行维护成本4) Operation and maintenance costs of the selected units of the virtual power plant
其中，OCMGT,k,op、OCES,e,op分别为微燃机组k、第e台储能系统的单位输出功率的运行费用。Among them, OCMGT,k,op and OCES,e,op are the operating costs per unit output power of the micro-combustion unit k and the e-th energy storage system respectively.
5)虚拟电厂的条件风险价值(conditional value-at-risk，CVaR)5) Conditional value-at-risk (CVaR) of virtual power plant
其中，v为边界值；b为提前给定的置信度，取0.5；dv为决策向量；Lf为损失函数，其值取虚拟电厂的收益负值。Among them, v is the boundary value; b is the confidence level given in advance, which is 0.5; dv is the decision vector; Lf is the loss function, and its value is the negative value of the virtual power plant's income.
步骤三：虚拟电厂VPP的运行约束条件Step 3: Operating constraints of the virtual power plant VPP
1)微燃机运行约束1) Micro-gas turbine operation constraints
PMGT,i,min≤PMGT,i(t)≤PMGT,i,maxPMGT,i,min ≤PMGT,i (t)≤PMGT,i,max
其中，PMGT,i(t)、PMGT,i(t-1)分别为微燃机组i在t时刻和t-1时刻的功率；PMGT,i,min、PMGT,i,max分别为微燃机组i的最小、最大输出功率；分别为微燃机组i的向下、向上爬坡功率。Among them, PMGT,i (t) and PMGT,i (t-1) are the power of micro-combustion unit i at time t and time t-1 respectively; PMGT,i,min and PMGT,i,max respectively are the minimum and maximum output power of micro-combustion unit i; are the downward and upward climbing powers of micro-combustion unit i respectively.
2)储能系统的荷电状态和充、放电约束2) State of charge and charging and discharging constraints of the energy storage system
(1-Cd,max)Si,man≤Si(t)≤Si,man(1-Cd,max )Si,man ≤Si (t)≤Si,man
0≤PES,i,ec(t)≤cec(t)PES,i,ec,max0≤PES,i,ec (t)≤cec (t)PES,i,ec,max
0≤PES,i,ed(t)≤ced(t)PES,i,ed,max0≤PES,i,ed (t)≤ced (t)PES,i,ed,max
cec(t)+ced(t)≤1cec (t)+ced (t)≤1
其中，Si(t)、Si(t-1)分别为第i台储能系统在t时刻、t-1时刻的荷电状态；γec、γed分别为储能系统的充电、放电效率系数；Si,man为第i台储能系统荷电状态的上限值；Cd,max为储能系统的最大放电深度；PES,i,ec(t)、PES,i,ed(t)分别为第i台储能系统在t时刻的充电、放电功率；PES,i,ec,max、PES,i,ed,max分别为第i台储能系统的最大允许充电、最大允许放电功率：cec(t)、ced(t)分别为储能系统在t时刻是否处于充电、放电的状态值，是则取1，否则取0，两者不可以同时为1。Among them, Si (t) and Si (t-1) are the state of charge of the i-th energy storage system at time t and time t-1 respectively; γec and γed are the charging and discharging of the energy storage system respectively. Efficiency coefficient; Si,man is the upper limit of the state of charge of the i-th energy storage system; Cd,max is the maximum discharge depth of the energy storage system; PES,i,ec (t), PES,i, ed (t) are the charging and discharging power of the i-th energy storage system at time t respectively; PES,i,ec,max and PES,i,ed,max are respectively the maximum allowable charging of the i-th energy storage system. , Maximum allowable discharge power: cec (t), ced (t) are respectively whether the energy storage system is in the state of charging and discharging at time t. If so, it takes 1, otherwise it takes 0. Both cannot be 1 at the same time. .
3)可中断负荷约束3) Interruptible load constraints
0≤PIL(t)≤sfILPL(t)0≤PIL (t)≤sfIL PL (t)
其中，PL(t)为虚拟电厂中的电力负荷；sfIL为虚拟电厂中电力负荷中的可中断负荷的比例系数。Among them, PL (t) is the electric load in the virtual power plant; sfIL is the proportional coefficient of the interruptible load in the electric load in the virtual power plant.
4)功率平衡约束4) Power balance constraints
5)备用容量约束5) Spare capacity constraints
为了克服运行优化过程中不确定因素的影响，需预先设置备用容量约束，具体的约束公式为：In order to overcome the influence of uncertain factors during the operation optimization process, the reserve capacity constraint needs to be set in advance. The specific constraint formula is:
其中，R+(t)为VPP在t时刻所需要的上旋转备用容量；R+(t)为VPP在t时刻所需要的下旋转备用容量。Among them, R+ (t) is the up-spinning reserve capacity required by VPP at time t; R+ (t) is the down-spinning reserve capacity required by VPP at time t.
6)系统风险约束6) System risk constraints
风险系数过大的调度方案对于调度人员参考价值较小，因此根据调度人员对于风险性的要求，令风险系数小于某一阈值：A scheduling plan with an excessive risk coefficient has less reference value for dispatchers. Therefore, according to the dispatcher's risk requirements, the risk coefficient is made less than a certain threshold:
Rf(P)≤δRf (P)≤δ
其中Rf(P)为风险系数，是一个概率，发生每一种情况的概率(备用容量不足)调度方案的风险程度通过风险系数进行表示。风险系数Rf(P)定义如下：Among them, Rf (P) is the risk coefficient, which is a probability. The probability of each situation occurring (insufficient spare capacity) and the degree of risk of the scheduling plan are expressed by the risk coefficient. The risk coefficient Rf (P) is defined as follows:
Rf(P)＝max{max(Ps,t,Pn,t)|t＝1,2,…,24}Rf (P)＝max{max(Ps,t ,Pn,t )|t＝1,2,…,24}
Ppl,t＝Pal,t-Pw,tPpl,t =Pal,t -Pw,t
Si,t＝min(Pi,t,max-Pi,t,uiT10)Si,t =min(Pi,t,max -Pi,t ,ui T10 )
Ni,t＝min(Pi,t-Pi,t,max,diT10)Ni,t =min(Pi,t -Pi,t,max ,di T10 )
其中，Rf(P)表示调度方案在所有调度时段正或负旋转备用不足概率最大值。Si,t、Ni,t别表示机组i在时刻t可提供的正负旋转备用容量；Ppl,t、Pal,t、Pw,t分别表示系统时刻t的净负荷、实际负荷、风电场总出力；ui、di分别表示机组i的分钟级上下爬坡速率；T10表示旋转备用响应时间，取值10min；E(Ppl,t)表示Ppl,t的期望值。Among them, Rf (P) represents the maximum value of the positive or negative spinning reserve shortage probability of the scheduling scheme in all scheduling periods. Si,t and Ni,t respectively represent the positive and negative spinning reserve capacity that unit i can provide at time t; Ppl,t , Pal,t and Pw,t respectively represent the net load and actual load of the system at time t , the total output of the wind farm; ui and di respectively represent the minute-level up and down ramp rate of unit i; T10 represents the spinning reserve response time, which takes a value of 10 minutes; E(Ppl,t ) represents the expected value of Ppl,t .
步骤四：场景选择Step 4: Scene selection
风光随机场景生成Scenery random scene generation
风险价值的计算方法有随机模拟法和历史数据模拟法等，由于风速、光照强度以及负荷的概率分布难以准确获得所以采用随机模拟法来计算风险价值。The risk value calculation methods include random simulation method and historical data simulation method. Since the probability distribution of wind speed, light intensity and load is difficult to obtain accurately, the random simulation method is used to calculate the risk value.
根据研究分析，本发明认为风电和太阳直接辐照度(direct normal irradiance，DNI)的概率分布分别服从Weibull分布、Beta分布，其概率密度分布函数分别为：According to research and analysis, the present invention believes that the probability distributions of wind power and solar direct irradiance (direct normal irradiance, DNI) obey Weibull distribution and Beta distribution respectively, and their probability density distribution functions are respectively:
式中：Γ为Gamma函数；α和β为形状系数；r为太阳辐照度；rmax为太阳辐照度的最大值；P为实时风电功率；k为形状参数；c为分布系数。In the formula: Γ is the Gamma function; α and β are shape coefficients; r is the solar irradiance; rmax is the maximum value of the solar irradiance; P is the real-time wind power power; k is the shape parameter; c is the distribution coefficient.
根据风光概率密度分布，利用抽样算法获得大量概率预测场景。与其他不同的抽样算法相比，拉丁超立方抽样(Latin hypercube sampling，LHS)具有抽样精度高、抽样规模小以及抽样效率高的特点，能显著改善对随机变量均值和方差的估计效果，稳健性好。LHS主要包括区间抽样和样本相关性排序：According to the scenery probability density distribution, a sampling algorithm is used to obtain a large number of probability prediction scenarios. Compared with other different sampling algorithms, Latin hypercube sampling (LHS) has the characteristics of high sampling accuracy, small sampling scale and high sampling efficiency. It can significantly improve the estimation effect of the mean and variance of random variables, and its robustness good. LHS mainly includes interval sampling and sample correlation ranking:
1)抽样1) Sampling
假设N个服从某概率分布的独立随机变量Xn，的概率累计函数Zn为：Assuming N independent random variables Xn obeying a certain probability distribution, the probability accumulation function Zn is:
Zn＝fn(Xn),n＝1,2,…,NZn =fn (Xn ), n = 1, 2,...,N
设抽样规模为M，将取值区间[0，1]均分为M个区间，并从每个等距区间中随机生成一个抽样值，每个区间只生成一个抽样值。抽样值可表示为：Assume the sampling scale is M, divide the value interval [0, 1] into M intervals, and randomly generate a sample value from each equally spaced interval, and only generate one sample value for each interval. The sampled value can be expressed as:
其中，Um为m个区间的抽样值；U为0或1的随机数。Among them, Um is the sampling value of m intervals; U is a random number of 0 or 1.
得到M个抽样值后，利用反函数可得到Xn的第m个抽样值Xm为：After obtaining M sample values, the mth sample value Xm of Xn can be obtained by using the inverse function:
变量Xn抽样结束后，得到M个抽样值。当N个变量全部抽样结束后，即构成一个N×M阶的初始样本矩阵Xs。After the sampling of variable Xn is completed, M sample values are obtained. When all N variables are sampled, an initial sample matrix Xs of order N×M is formed.
2)排序2) Sort
获得初始样本矩阵Xs后，采用Cholesky分解法对采样矩阵重新排序，降低矩阵内元素的相关性，提高抽样精度，最终获得排序后的N×M阶样本矩阵XN。After obtaining the initial sample matrix Xs, the Cholesky decomposition method is used to reorder the sampling matrix to reduce the correlation of elements within the matrix and improve the sampling accuracy, and finally obtain the sorted N×M order sample matrix XN.
样本矩阵XN中行元素代表调度周期内某时间区间风/光出力抽样值，列元素代表一个调度周期内各时间区间风光出力的抽样值，即形成含M个风/光随机预测场景的场景集以及各自场景出现的概率。The row elements in the sample matrix The probability of occurrence of each scenario.
为有效体现风光分布相关性，进一步优化预测结果，本发明采用K-均值聚类算法对数量削减后的风光随机场景集进行相关性聚类分析。In order to effectively reflect the correlation of scenery distribution and further optimize the prediction results, the present invention uses K-means clustering algorithm to perform correlation clustering analysis on the reduced scenery random scene set.
K-均值算法的基本思想为：首先在某一数据集中随机选取K个样本作为初始聚类中心；然后计算所有样本点与聚类中心的距离，按照距离最近原则，将样本点划分到相应的簇中；在每次迭代过程中重新计算得到新的聚类中心：当聚类中心不再变化或者达到指定条件时，输出聚类结果。The basic idea of the K-means algorithm is: first randomly select K samples from a certain data set as the initial cluster center; then calculate the distance between all sample points and the cluster center, and divide the sample points into corresponding clusters according to the principle of closest distance. in the cluster; recalculate the new clustering center during each iteration: when the clustering center no longer changes or reaches the specified condition, the clustering result is output.
设数据集D(x1，2，...，xn)中含n个样本，随机选取K个初始聚类中心后，则D中任意两个样本间的距离为：Assume that the data set D (x1, 2, ..., xn) contains n samples, and after randomly selecting K initial clustering centers, the distance between any two samples in D is:
其中，d(xj,xk)—样本点与间的距离，距离越小表示两个样本越相似。Among them, d(xj ,xk )—the distance between sample points. The smaller the distance, the more similar the two samples are.
当初始聚类中心选定后，算法迭代过程将自动运行，而选择不同初始聚类中心对聚类过程稳定性和聚类结果有较大影响。When the initial clustering center is selected, the algorithm iteration process will run automatically, and choosing different initial clustering centers has a greater impact on the stability of the clustering process and the clustering results.
为避免算法陷入局部聚类最优结束，本发明通过计算样本间的误差平方和来优化初始聚类中心。数据集D中样本的误差平方和SD为：In order to prevent the algorithm from falling into the local clustering optimal end, the present invention optimizes the initial clustering center by calculating the sum of square errors between samples. The sum of squared errors SD of the samples in data set D is:
其中，cl为第l个初始聚类中心；SD越小表示样本点越密集。Among them, cl is the l-th initial clustering center; the smaller SD means the denser the sample points.
根据上式计算数据集D中样本的误差平方和SD，并选取其中最小值作为新的初始聚类中心，将上式迭代K次后获得K个优化初始聚类中心。得到新的聚类中心后对风光数据集进行聚类优化计算，将N×N个风光随机场景进一步聚类优化为N个风光典型预测场景。Calculate the sum of squared errors SD of the samples in data set D according to the above formula, and select the minimum value as the new initial cluster center. After iterating the above formula K times, K optimized initial cluster centers are obtained. After obtaining the new clustering center, perform clustering optimization calculation on the scenery data set, and further cluster and optimize the N×N random scenery scenes into N typical scenery prediction scenarios.
综合考虑预测精度与计算效率，风电场景通过LHS抽样1000次，经场景削减和相关性优化后获得5个风光典型预测场景，各场景概率见表1。为分析随机优化生成的典型风光场景精度，将优化后所生成的典型风光预测场景、常规风光预测基准场景和调度周期内实际风光场景进行对比分析，如图1所示。Taking into account the prediction accuracy and calculation efficiency, the wind power scenarios were sampled 1000 times by LHS. After scenario reduction and correlation optimization, 5 typical wind and solar prediction scenarios were obtained. The probabilities of each scenario are shown in Table 1. In order to analyze the accuracy of typical scenery scenes generated by random optimization, the typical scenery prediction scenes generated after optimization, the conventional scenery prediction benchmark scenes and the actual scenery scenes within the scheduling cycle were compared and analyzed, as shown in Figure 1.
表1风光场景出现概率Table 1 Occurrence probability of scenery scenes
步骤五：确认目标函数：Step 5: Confirm the objective function:
其中，ηi为场景i发生的概率；L为权重系数，其表现VPP管理者的风险偏好，L≥0。当L取较小的值时，管理者的选择比较激进，希望获取较高的收益，但同时会面对较大的风险损失；当取较大的值时，管理者选择比较保守，获取的收益较低，但同时会面对的风险损失较小。Among them, etai is the probability of scenario i occurring; L is the weight coefficient, which expresses the risk preference of VPP managers, L≥0. When L takes a smaller value, the manager's choice is more aggressive, hoping to obtain higher returns, but at the same time he will face greater risk losses; when it takes a larger value, the manager's choice is more conservative and obtains The returns are lower, but at the same time the risks and losses you face are smaller.
步骤六：结果分析Step 6: Result Analysis
1)VPP管理者选择不同的风险偏好系数，VPP的机组组合优化结果如下图所示。可以明显看出，随着风险偏好系数L的减小，VPP的管理者会偏向于风险更高的选择，即VPP的机组组合规划越来越偏向于激进，条件风险价值越来越大。随着VPP的总成本不断降低，而CVaR值不断增加，所面临的风险也随之上升。针对不同的L值，VPP会获得不同的收益以及面对不同的风险损失，进而得到不同的VPP机组组合结果。当L值较小时，VPP管理者的选择相对激进，优化结果为选择接入大量的可再生能源机组来和没有使用风险系数的方法进行比价。1) VPP managers choose different risk preference coefficients, and the VPP unit portfolio optimization results are shown in the figure below. It can be clearly seen that as the risk preference coefficient L decreases, VPP managers will prefer higher-risk choices, that is, VPP's unit portfolio planning will become more and more radical, and the conditional risk value will become larger and larger. As the total cost of VPP continues to decrease and the CVaR value continues to increase, the risks faced also increase. For different L values, VPP will obtain different benefits and face different risk losses, and then obtain different VPP unit combination results. When the value of L is small, the VPP manager's choice is relatively radical, and the optimization result is to choose to connect a large number of renewable energy units to compare prices with methods that do not use risk factors.
2)VPP出力计划分析2) Analysis of VPP output plan
VPP作为一个发电厂，可以根据VPP内的机组出力拟定日前调度计划。在所建模型中，当VPP管理者确定风险偏好系数后，就可以得到最优的机组选择组合，再根据这个组合的机组出力，给出VPP日前出力计划。As a power plant, VPP can develop day-ahead dispatch plans based on the output of the units within the VPP. In the built model, when the VPP manager determines the risk preference coefficient, the optimal unit selection combination can be obtained, and then the VPP day-ahead output plan is given based on the unit output of this combination.
表2 VPP中机组额定容量数据单位：MWhTable 2 Unit rated capacity data unit in VPP: MWh
选取一个夏季典型日作为仿真对象，假设VPP管理者的风险偏好系数L值为0.50，以此选择最优的机组组合，组合中包括2台风电机组(W2、W3)、4台光伏机组、1台微燃机(M3)、2个储能系统(E1、E4)。负荷需求以及可再生能源出力如图3所示，VPP的日前出力计划如图4所示，分时段分析如下：Select a typical day in summer as the simulation object, assuming that the risk preference coefficient L value of the VPP manager is 0.50, and select the optimal unit combination. The combination includes 2 wind turbine units (W2, W3), 4 photovoltaic units, 1 A micro-gas turbine (M3) and two energy storage systems (E1, E4). The load demand and renewable energy output are shown in Figure 3. The day-ahead output plan of VPP is shown in Figure 4. The period-by-period analysis is as follows:
0:00一5:00：电价较低，可再生能源出力较少，只能满足小部分的负荷需求；VPP优先调度M3进行放电，以此来满足大部分的负荷需求；由于电价比较低以及微燃机的环境成本比较高，选择购入部分电量并调度可中断负荷，以此来满足剩余的负荷需求，并对E1、E4进行充电。0:00 to 5:00: The price of electricity is low, and the output of renewable energy is small, which can only meet a small part of the load demand; VPP prioritizes scheduling M3 for discharge to meet most of the load demand; due to the low price of electricity and The environmental cost of micro-gas turbines is relatively high. You choose to purchase part of the power and schedule the interruptible load to meet the remaining load demand and charge E1 and E4.
5:00一12:00：电价较高，可再生能源出力较多，可以满足大部分的负荷需求；VPP优先调度充电后的E1、E4进行放电，以此来满足部分的负荷需求；由于电价比较高，选择少购入或不购入电量，直接调度M3以及可中断负荷，以此来满足剩余的负荷需求。5:00 to 12:00: The electricity price is high, and renewable energy output is large, which can meet most of the load demand; VPP prioritizes charging E1 and E4 to discharge to meet part of the load demand; due to the electricity price Relatively high, choose to purchase less or no electricity, and directly dispatch M3 and interruptible loads to meet the remaining load demand.
12:00一19:00：电价较高，可再生能源出力多，可以基本满足负荷的需求；由于电价比较高，选择少购入或不购入电量；VPP调度M3进行放电以及可中断负荷就可以满足剩余的负荷需求，由于剩余的负荷需求少，M3和可中断负荷的出力处于较低水平就可以完全满足，并对E1、E4进行充电。12:00 to 19:00: The electricity price is high, and renewable energy output is large, which can basically meet the load demand; due to the relatively high electricity price, choose to purchase less or no electricity; VPP schedules M3 for discharge and interruptible load The remaining load demand can be met. Since the remaining load demand is small, the output of M3 and interruptible load can be fully satisfied at a low level, and E1 and E4 can be charged.
19:00一24:00：电价较低，可再生能源出力较少，只能满足小部分的负荷需求；VPP优先调度M3进行放电，以此来满足大部分的负荷需求；由于电价比较低，选择购入部分的电量，并调度充电后的E1、E4进行放电以及可中断负荷，以此来满足剩余的负荷需求。19:00 to 24:00: The electricity price is low and renewable energy output is small, which can only meet a small part of the load demand; VPP prioritizes M3 for discharge to meet most of the load demand; due to the low electricity price, Choose to purchase part of the electricity, and schedule the charged E1 and E4 to discharge and interrupt the load to meet the remaining load demand.
可以理解，本发明是通过一些实施例进行描述的，本领域技术人员知悉的，在不脱离本发明的精神和范围的情况下，可以对这些特征和实施例进行各种改变或等效替换。另外，在本发明的教导下，可以对这些特征和实施例进行修改以适应具体的情况及材料而不会脱离本发明的精神和范围。因此，本发明不受此处所公开的具体实施例的限制，所有落入本申请的权利要求范围内的实施例都属于本发明所保护的范围内。It is understood that the present invention has been described through some embodiments. Those skilled in the art know that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, the features and embodiments may be modified to adapt a particular situation and material to the teachings of the invention without departing from the spirit and scope of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed here, and all embodiments falling within the scope of the claims of the present application are within the scope of protection of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311665833.5ACN117674292A (en) | 2023-12-06 | 2023-12-06 | An optimized dispatching method for virtual power plants |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311665833.5ACN117674292A (en) | 2023-12-06 | 2023-12-06 | An optimized dispatching method for virtual power plants |
| Publication Number | Publication Date |
|---|---|
| CN117674292Atrue CN117674292A (en) | 2024-03-08 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311665833.5APendingCN117674292A (en) | 2023-12-06 | 2023-12-06 | An optimized dispatching method for virtual power plants |
| Country | Link |
|---|---|
| CN (1) | CN117674292A (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118473020A (en)* | 2024-07-09 | 2024-08-09 | 国网浙江省电力有限公司淳安县供电公司 | Virtual power plant configuration method and system |
| CN118523312A (en)* | 2024-07-18 | 2024-08-20 | 国网山东省电力公司烟台供电公司 | Active distributed resource optimization scheduling method based on virtual power plant |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118473020A (en)* | 2024-07-09 | 2024-08-09 | 国网浙江省电力有限公司淳安县供电公司 | Virtual power plant configuration method and system |
| CN118523312A (en)* | 2024-07-18 | 2024-08-20 | 国网山东省电力公司烟台供电公司 | Active distributed resource optimization scheduling method based on virtual power plant |
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Two-stage energy management for networked microgrids with high renewable penetration | |
| CN112803499B (en) | Wind, light and water multi-energy complementary capacity optimal configuration method with power/electric quantity compensation cooperation | |
| Walker et al. | Design of structured control policy for shared energy storage in residential community: A stochastic optimization approach | |
| CN112994115B (en) | New energy capacity configuration method based on WGAN scene simulation and time sequence production simulation | |
| CN107301472A (en) | Distributed photovoltaic planing method based on scene analysis method and voltage-regulation strategy | |
| CN107134810B (en) | Independent micro-energy-grid energy storage system optimal configuration solving method | |
| CN103544655A (en) | Layered optimization method of regional distribution network comprising micro-grid | |
| CN108233430B (en) | An AC-DC Hybrid Microgrid Optimization Method Considering System Energy Volatility | |
| CN113326467B (en) | Multi-objective optimization method, storage medium and optimization system for multi-station integrated energy system based on multiple uncertainties | |
| CN117674292A (en) | An optimized dispatching method for virtual power plants | |
| CN111934360A (en) | Virtual power plant-energy storage system energy collaborative optimization regulation and control method based on model predictive control | |
| CN110350527A (en) | A kind of increment power distribution network dual-layer optimization configuration method containing distributed generation resource | |
| CN111293718B (en) | AC/DC hybrid micro-grid partition two-layer optimization operation method based on scene analysis | |
| CN115765044B (en) | Combined operation and risk analysis method and system for wind, light and water power system | |
| CN113725904B (en) | Grid transformation methods, devices and equipment considering the decommissioning of aging generating units | |
| CN116896072A (en) | Optimal control methods and devices for offshore wind power and hydrogen production systems | |
| CN117977538A (en) | A method, system, device and medium for evaluating power system flexibility | |
| Zahmoun et al. | Optimal operation scheduling of a combined wind-hydro system for peak load shaving | |
| CN109546647B (en) | A safety and stability assessment method for power systems with wind-solar water storage | |
| CN119109063A (en) | A virtual power plant optimal scheduling method considering uncertain factors | |
| CN113536581A (en) | Energy storage system multi-state reliability modeling method considering operation strategy | |
| CN110120682B (en) | Power supply optimization scheduling method for tower barrel elevator with minimum lost air volume | |
| Gao et al. | Impacts of energy storage on reliability of power systems with WTGs | |
| Deng et al. | Analysis of Renewable Energy Accommodation Capability of Shanxi Power Grid Based on Operation Simulation Method | |
| Yao et al. | Design of short-term dispatch strategy to maximize income of a wind power-energy storage generating station |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |