CN110737198B - Predictive control method for large coal-fired power plant CO2 capture system based on BP neural network - Google Patents
Predictive control method for large coal-fired power plant CO2 capture system based on BP neural networkDownload PDFInfo
- Publication number
- CN110737198B CN110737198BCN201910952786.XACN201910952786ACN110737198BCN 110737198 BCN110737198 BCN 110737198BCN 201910952786 ACN201910952786 ACN 201910952786ACN 110737198 BCN110737198 BCN 110737198B
- Authority
- CN
- China
- Prior art keywords
- fired power
- variable
- power plant
- neural network
- coal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
Translated fromChinese
Description
Translated fromChinese技术领域technical field
本发明涉及热工自动控制领域，尤其是一种基于BP神经网络的大型燃煤电站CO2捕集系统预测控制方法，属于控制、调节的技术领域。The invention relates to the field of thermal automatic control, in particular to a predictive control method for a CO2 capture system of a large coal-fired power station based on a BP neural network, belonging to the technical field of control and adjustment.
背景技术Background technique
火电机组是当前CO2等温室气体最主要的排放源，因此研究火电机组CO2捕集技术是实现温室气体减排、落实巴黎协定的重要手段。以MEA为吸附溶剂的燃烧后CO2捕集技术是当前CO2捕集的主流技术。Thermal power units are currently the most important source of greenhouse gas emissions such as CO2 , so research on CO2 capture technology for thermal power units is an important means to reduce greenhouse gas emissions and implement the Paris Agreement. The post-combustionCO2 capture technology using MEA as the adsorption solvent is currently the mainstream technology forCO2 capture.
火电机组与燃烧后CO2捕集系统相互影响。根据电网负荷指令，火电机组需要参与负荷调峰，尾部烟气因此会随机组负荷产生波动，烟气波动会随之影响下游CO2捕集系统，对捕集率、再沸器温度等关键变量产生较大影响。同时，大型燃煤电站CO2捕集系统具有大惯性、强耦合、非线性的特性，常规PID控制系统由于是单回路控制器，难以有效处理CO2捕集系统各变量之间的耦合特性，同时也难以处理系统的大延迟和输出约束。预测控制系统是一种多变量的控制器，能够有效处理大惯性、强耦合的特性。The thermal power unit interacts with the post-combustionCO2 capture system. According to the load order of the power grid, thermal power units need to participate in load peak regulation, so the flue gas at the tail will fluctuate according to the load of the group, and flue gas fluctuations will subsequently affect the downstreamCO2 capture system, affecting key variables such as capture rate and reboiler temperature have a greater impact. At the same time, theCO2 capture system of a large coal-fired power station has the characteristics of large inertia, strong coupling, and nonlinearity. Because the conventional PID control system is a single-loop controller, it is difficult to effectively deal with the coupling characteristics between the variables of theCO2 capture system. It is also difficult to handle the large latency and output constraints of the system. The predictive control system is a multivariable controller that can effectively deal with the characteristics of large inertia and strong coupling.
然而，预测控制器的控制效果主要取决于预测模型的精度，CO2捕集系统在不同烟气和不同捕集率工况下的系统特性变化大，常规的线性状态空间预测模型难以有效描述大型燃煤电站CO2捕集系统复杂的非线性特征，影响控制效果。However, the control effect of the predictive controller mainly depends on the accuracy of the predictive model. The system characteristics of theCO2 capture system vary greatly under the conditions of different flue gas and different capture rates, and it is difficult for the conventional linear state-space predictive model to effectively describe large-scale The complex nonlinear characteristics ofCO2 capture system in coal-fired power plants affect the control effect.
发明内容Contents of the invention
本发明的发明目的是针对上述背景技术的不足，提供了基于BP神经网络的大型燃煤电站CO2捕集系统预测控制方法，利用BP神经网络工具箱建立能够准确描述大范围变负荷时CO2捕集系统未来输出特性的预测模型，实现了预测控制器的精准调节，极大的提高了控制效果，解决了常规PID控制器不能处理大延迟、约束和强耦合特性的技术问题。The purpose of the present invention is to address the deficiencies of the above-mentioned background technology and provide a predictive control method for large-scale coal-fired power plant CO2 capture systems based on BP neural network, using the BP neural network toolbox to establish a CO2 capture system that can accurately describe large-scale variable loads The prediction model of the future output characteristics of the capture system realizes the precise adjustment of the predictive controller, greatly improves the control effect, and solves the technical problems that conventional PID controllers cannot handle large delays, constraints, and strong coupling characteristics.
本发明为实现上述发明目的采用如下技术方案：The present invention adopts following technical scheme for realizing above-mentioned purpose of the invention:
基于BP神经网络的大型燃煤电站CO2捕集系统预测控制方法，通过包括目标值设置单元、神经网络预测模型、优化求解模块、大型燃煤电站CO2捕集系统模型、第一延迟单元、第二延迟单元和第三延迟单元的系统实现；优化求解模块有两路输入，分别是目标值设置单元的输出yr(k)和神经网络预测模型的输出 y(k+i|k),i＝0,1,…P-1，P为预测时域，优化求解模块的输出为当前时刻最优控制变量u(k)；大型燃煤电站CO2捕集整体系统的输入变量u(k)与输出变量y(k)分别通过第一延迟单元和第二延迟单元得到延迟变量u(k-1)与y(k-1)；扰动变量d(k) 通过第三延迟单元得到延迟变量d(k-1)，第一延迟单元、第二延迟单元和第三延迟单元输出变量u(k-1)、y(k-1)及d(k-1)与优化求解模块输出的u(k)及扰动d(k)作为神经网络预测模型输入，计算出CO2捕集系统预测输出y(k+i|k)。A predictive control method for large-scale coal-fired power plantCO2 capture system based on BP neural network, including target value setting unit, neural network prediction model, optimization solution module, large-scale coal-fired power plantCO2 capture system model, first delay unit, The system implementation of the second delay unit and the third delay unit; the optimization solution module has two inputs, which are the output yr(k) of the target value setting unit and the output y(k+i|k) of the neural network prediction model, i =0,1,...P-1, P is the prediction time domain, the output of the optimization solution module is the optimal control variable u(k) at the current moment; the input variable u(k) of the CO2 capture system of large coal-fired power plants and the output variable y(k) pass through the first delay unit and the second delay unit to obtain the delay variables u(k-1) and y(k-1); the disturbance variable d(k) obtains the delay variable d through the third delay unit (k-1), the output variables u(k-1), y(k-1) and d(k-1) of the first delay unit, the second delay unit and the third delay unit and the output of the optimization solution module ( k) and disturbance d(k) are used as the input of the neural network prediction model, and the CO2 capture system prediction output y(k+i|k) is calculated.
相应地，一种基于BP神经网络的大型燃煤电站CO2捕集系统预测控制方法，包括如下步骤：Correspondingly, a method for predictive control of a large coal-fired power stationCO2 capture system based on BP neural network includes the following steps:
(1)选取CO2捕集率及再沸器温度为大型燃煤电站CO2捕集系统的被控变量y(k)，选取贫液流量及再沸器抽汽流量为相对应的控制变量u(k)，选取烟气流量为主要扰动变量d(k)；(1) Select the CO2 capture rate and reboiler temperature as the controlled variables y(k) of the CO2 capture system in large coal-fired power plants, and select the lean liquid flow rate and reboiler extraction steam flow rate as the corresponding control variables u(k), select the flue gas flow rate as the main disturbance variable d(k);
(2)在开环情况下，改变入口烟气流量、贫液流量和再沸器抽汽流量，模拟白噪声信号，进行开环激励试验；设置采样周期Ts，获取不同烟气、捕集率负荷下大型燃煤电站CO2捕集系统模型的输入、输出数据；(2) In the open-loop condition, change the inlet flue gas flow rate, lean liquid flow rate and reboiler extraction steam flow rate, simulate the white noise signal, and conduct an open-loop excitation test; set the sampling period Ts to obtain different flue gas, trapping The input and output data of the large coal-fired power plant CO2 capture system model under normal load;
(3)将入口烟气流量和贫液流量和再沸器抽汽流量作为输入变量，将捕集率和再沸器温度作为输出变量；将当前时刻的输入u(k)、扰动d(k)、上一时刻的输入u(k-1)、扰动d(k-1)以及上一时刻的输出y(k-1)作为训练输入，利用BP神经网络进行离线训练，建立大型燃煤电站CO2捕集系统的预测模型，如公式(1)：(3) The inlet flue gas flow rate, lean liquid flow rate and reboiler extraction flow rate are taken as input variables, and the capture rate and reboiler temperature are taken as output variables; the current input u(k), disturbance d(k ), the input u(k-1) at the previous moment, the disturbance d(k-1) and the output y(k-1) at the previous moment are used as training input, and the BP neural network is used for offline training to establish a large coal-fired power station The prediction model ofCO2 capture system, such as formula (1):
y(k)＝f(u(k-1),d(k-1),u(k),d(k),y(k-1)) (1)；y(k)=f(u(k-1), d(k-1), u(k), d(k), y(k-1)) (1);
(4)将离线神经网络模型作为预测模型，获得未来P步时刻内CO2捕集系统的预测输出Y(k+P|k)，其中， Y(k+P|k)＝[y(k|k) y(k+1|k) … y(k+P-1|k)]T；(4) Use the offline neural network model as the prediction model to obtain the prediction output Y(k+P|k) of the CO2 capture system in the future P steps, where Y(k+P|k)=[y(k |k) y(k+1|k) … y(k+P-1|k)]T ;
(5)设置控制器相关参数，包括预测时域P、控制时域M、输出误差权矩阵Q、控制权矩阵R。采样周期Ts的选取一般要符合香农采样定理，可以用经验规则T95/Ts＝5～15来选取，其中，T95为过渡过程上升到95％的调节时间；预测时域P应尽量包含对象的真实动态部分；M取3～5；(5) Set controller-related parameters, including prediction time domain P, control time domain M, output error weight matrix Q, and control weight matrix R. The selection of the sampling period Ts generally conforms to the Shannon sampling theorem, and can be selected by the empirical rule T95 /Ts = 5~15, where T95 is the adjustment time for the transition process to rise to 95%; the prediction time domain P should be as far as possible Contains the real dynamic part of the object; M takes 3 to 5;
(6)设置控制器性能指标，如公式(2)所示：(6) Set the performance index of the controller, as shown in formula (2):
其中，Yr＝[yr(k) yr(k) … yr(k)]T为CO2捕集系统输出量给定值，ΔU＝[Δu(k)Δu(k+1) … Δu(k+M-1)]T为未来M时刻内输入量的差值；Among them, Yr＝[yr(k) yr(k) … yr(k)]T is the given value of the output of theCO2 capture system, ΔU＝[Δu(k)Δu(k+1) … Δu(k+ M-1)]T is the difference of the input amount in the future M time;
(7)采用非线性规划fmincon求解器求解性能指标，采用暖启动的方式将每一时刻计算出来的最佳输入量偏差△u(k)作为下一时刻的初始值，提高求解精度；(7) Use the nonlinear programming fmincon solver to solve the performance index, and use the warm start method to use the optimal input deviation △u(k) calculated at each moment as the initial value at the next moment to improve the solution accuracy;
(8)计算当前时刻贫液流量和再沸器抽汽的最佳控制量u(k)＝u(k-1)+Δu(k)；(8) Calculating the optimal control quantity u(k)=u(k-1)+Δu(k) of lean liquid flow and reboiler extraction at the current moment;
(9)输出最佳控制量u(k)，采集CO2捕集系统的输出y(k)，其后在每个采样周期内，重复执行第(4)步到第(9)步。(9) Output the optimal control quantity u(k), collect the output y(k) of the CO2 capture system, and then repeat steps (4) to (9) in each sampling period.
本发明采用上述技术方案，具有以下有益效果：本发明采用BP神经网络工具箱，建立了能够准确描述大范围变负荷时CO2捕集系统未来输出特性的正向模型，将BP神经网络模型作为预测控制器的预测模型解决了常规线性模型预测精度差的技术问题，实现了预测控制器在大规模变负荷情况下的精准调节，提高动态调节品质，通过模型预测控制方法解决了常规PID控制器不能处理大延迟、输入输出约束和强耦合特性的技术问题。The present invention adopts the above-mentioned technical scheme and has the following beneficial effects: the present invention adopts the BP neural network toolbox to establish a forward model capable of accurately describing the future output characteristics of theCO2 capture system when the load varies in a large range, and uses the BP neural network model as The prediction model of the predictive controller solves the technical problem of poor prediction accuracy of the conventional linear model, realizes the precise adjustment of the predictive controller in the case of large-scale variable loads, improves the quality of dynamic adjustment, and solves the problem of the conventional PID controller through the model predictive control method. It cannot handle the technical problems of large delays, input-output constraints, and strong coupling characteristics.
附图说明Description of drawings
图1为本发明控制系统的结构示意图。Fig. 1 is a schematic structural diagram of the control system of the present invention.
图2为本发明大型燃煤电站CO2捕集系统捕集CO2的流程示意图。Fig. 2 is a schematic flow chart of the CO2 capture system of the large-scale coal-fired power plant CO2 capture system of the present invention.
图3(a)为本发明与传统线性预测控制方法在捕集率给定值阶跃变化时捕集率控制效果的对比示意图。Fig. 3(a) is a schematic diagram comparing the control effect of the capture rate between the present invention and the traditional linear predictive control method when the set value of the capture rate changes stepwise.
图3(b)为本发明与传统线性预测控制方法在捕集率给定值阶跃变化时再沸器温度控制效果的对比示意图。Fig. 3(b) is a schematic diagram comparing the reboiler temperature control effect of the present invention and the traditional linear predictive control method when the set value of the capture rate is changed stepwise.
图4(a)本发明与传统线性预测控制方法在捕集率给定值阶跃变化时贫液流量控制效果的对比示意图。Fig. 4(a) is a schematic diagram comparing the lean liquid flow control effect between the present invention and the traditional linear predictive control method when the set value of the collection rate changes stepwise.
图4(b)本发明与传统线性预测控制方法在捕集率给定值阶跃变化时再沸器抽汽流量控制效果的对比示意图。Fig. 4(b) is a schematic diagram of comparison between the present invention and the traditional linear predictive control method when the set value of the trapping rate changes stepwise.
具体实施方式Detailed ways
下面结合附图对发明的技术方案进行详细说明。The technical solution of the invention will be described in detail below in conjunction with the accompanying drawings.
本发明基于BP神经网络的大型燃煤电站CO2捕集系统预测控制方法通过如图1所示的控制系统实现，该系统包括：目标值设置单元、神经网络预测模型、优化求解模块、第一延迟单元、第二延迟单元和第三延迟单元；优化求解模块有两路输入，分别是目标值设置单元的输出yr(k)和神经网络预测模型的输出 y(k+i|k),i＝0,1,…P-1，P为预测时域，优化求解模块的输出为当前时刻最优控制变量u(k)；大型燃煤电站CO2捕集整体系统的输入变量u(k)与输出变量y(k)分别通过第一延迟单元和第二延迟单元得到延迟变量u(k-1)与y(k-1)；扰动变量d(k) 通过第三延迟单元得到延迟变量d(k-1)，第一延迟单元、第二延迟单元和第三延迟单元输出变量u(k-1)、y(k-1)及d(k-1)与优化求解模块输出u(k)及扰动d(k)作为神经网络预测模型输入，计算出CO2捕集系统预测输出y(k+i|k)。The present invention based on BP neural network predictive control method of large coal-fired power stationCO capture system is realized through the control system shown in Figure 1, the system includes: target value setting unit, neural network predictive model, optimization solution module, first The delay unit, the second delay unit and the third delay unit; the optimization solution module has two inputs, which are the output yr(k) of the target value setting unit and the output y(k+i|k) of the neural network prediction model, i =0,1,...P-1, P is the prediction time domain, the output of the optimization solution module is the optimal control variable u(k) at the current moment; the input variable u(k) of the CO2 capture system of large coal-fired power plants and the output variable y(k) pass through the first delay unit and the second delay unit to obtain the delay variables u(k-1) and y(k-1); the disturbance variable d(k) obtains the delay variable d through the third delay unit (k-1), the first delay unit, the second delay unit and the third delay unit output variables u(k-1), y(k-1) and d(k-1) and the optimization solution module output u(k ) and disturbance d(k) are used as the input of the neural network prediction model, and the CO2 capture system prediction output y(k+i|k) is calculated.
如图2所示，大型燃煤电站CO2捕集系统，包含：CO2捕集率、再沸器温度和贫液流量及再沸器抽汽流量、烟气流量等主要变量。基于BP神经网络的大型燃煤电站CO2捕集系统预测控制方法，包括如下步骤：As shown in Figure 2, theCO2 capture system of a large coal-fired power station includes:CO2 capture rate, reboiler temperature, lean liquid flow rate, reboiler extraction steam flow rate, flue gas flow rate and other main variables. A predictive control method for large coal-fired power plantCO2 capture system based on BP neural network, including the following steps:
(1)选取CO2捕集率及再沸器温度为大型燃煤电站CO2捕集系统的被控变量y(k)，选取贫液流量及再沸器抽汽流量为相对应的控制变量u(k)，选取烟气流量为主要扰动变量d(k)；(1) Select the CO2 capture rate and reboiler temperature as the controlled variables y(k) of the CO2 capture system in large coal-fired power plants, and select the lean liquid flow rate and reboiler extraction steam flow rate as the corresponding control variables u(k), select the flue gas flow rate as the main disturbance variable d(k);
(2)在开环情况下，改变入口烟气流量、贫液流量和再沸器抽汽流量，模拟白噪声信号，进行开环激励试验；设置采样周期Ts＝30s，获取不同烟气、捕集率负荷下大型燃煤电站CO2捕集系统模型的输入、输出数据；(2) In the open-loop situation, change the inlet flue gas flow rate, lean liquid flow rate and reboiler extraction steam flow rate, simulate the white noise signal, and carry out the open-loop excitation test; set the sampling period Ts = 30s, and obtain different flue gas, Input and output data of large coal-fired power plantCO2 capture system model under capture rate load;
(3)将入口烟气流量和贫液流量和再沸器抽汽流量作为输入变量，将捕集率和再沸器温度作为输出变量；将贫液流量、再沸器抽汽流量和烟气流量当前时刻的数据u(k)、d(k)和前一时刻的数据u(k-1)、d(k-1)以及CO2捕集率和再沸器温度前一时刻的数据y(k-1)作为神经网络输入，以CO2捕集率和再沸器温度当前时刻的数据y(k)作为神经网络输出，利用BP神经网络工具箱建立燃煤电站CO2捕集系统的预测模型，该神经网络含有两层隐藏层，神经元个数分别为14和5，训练函数为traingdm，如公式(1)：(3) The inlet flue gas flow rate, lean liquid flow rate and reboiler extraction steam flow rate are used as input variables, and the capture rate and reboiler temperature are used as output variables; the lean liquid flow rate, reboiler extraction steam flow rate and flue gas Flow data u(k), d(k) at the current moment and data u(k-1), d(k-1) at the previous moment, as well asCO2 capture rate and reboiler temperature data y at the previous moment (k-1) is used as the input of the neural network, and the data y(k) of the CO2 capture rate and reboiler temperature at the current moment are used as the output of the neural network, and the BP neural network toolbox is used to establish the CO2 capture system of the coal-fired power station Forecasting model, the neural network contains two hidden layers, the number of neurons is 14 and 5 respectively, and the training function is traindm, such as formula (1):
y(k)＝f(u(k-1),d(k-1),u(k),d(k),y(k-1)) (1)；y(k)=f(u(k-1), d(k-1), u(k), d(k), y(k-1)) (1);
(4)将离线神经网络模型作为预测模型，获得未来P步时刻内CO2捕集系统的预测输出Y(k+P|k)，其中, Y(k+P|k)＝[y(k|k) y(k+1|k) … y(k+P-1|k)]T；(4) Use the offline neural network model as the prediction model to obtain the prediction output Y(k+P|k) of the CO2 capture system in the future P steps, where Y(k+P|k)=[y(k |k) y(k+1|k) … y(k+P-1|k)]T ;
(5)设置控制器相关参数。包括预测时域P，控制时域M，输出误差权矩阵Q，控制权矩阵R。Ts的选取一般要符合香农采样定理，可以用经验规则 T95/Ts＝5～15来选取，其中，T95为过渡过程上升到95％的调节时间；预测时域P 应尽量包含对象的真实动态部分；M取3～5；
(6)设置控制器性能指标，如公式(2)所示：(6) Set the performance index of the controller, as shown in formula (2):
其中，Yr＝[yr(k) yr(k) … yr(k)]T为CO2捕集系统输出量给定值，ΔU＝[Δu(k)Δu(k+1) … Δu(k+M-1)]T为未来M时刻内输入量的差值；Among them, Yr＝[yr(k) yr(k) … yr(k)]T is the given value of the output of theCO2 capture system, ΔU＝[Δu(k)Δu(k+1) … Δu(k+ M-1)]T is the difference of the input amount in the future M time;
(7)采用非线性规划fmincon求解器求解性能指标，采用暖启动的方式，将每一时刻计算出来的最佳输入量偏差△u(k)作为下一时刻的初始值，提高求解精度；(7) Use the nonlinear programming fmincon solver to solve the performance index, and use the warm start method to use the optimal input deviation △u(k) calculated at each moment as the initial value at the next moment to improve the solution accuracy;
(8)计算当前时刻贫液流量和再沸器抽汽的最佳控制量 u(k)＝u(k-1)+Δu(k)；(8) Calculate the lean liquid flow rate and the optimal control quantity u(k)=u(k-1)+Δu(k) for reboiler extraction at the current moment;
(9)输出最佳控制量u(k)，采集CO2捕集系统的输出y(k)。其后在每个采样周期内，重复执行第(4)步到第(9)步。(9) Output the optimal control quantity u(k), and collect the output y(k) of the CO2 capture system. Thereafter, in each sampling period, steps (4) to (9) are repeatedly executed.
本申请基于BP神经网络的预测控制方法控制效果与传统状态空间预测控制效果的对比如图3(a)、图3(b)、图4(a)、图4(b)所示。在初始稳态工况为烟气流量550kg/s,u1＝444.284kg/s、u2＝112.732kg/s、y1＝70％、y2＝392k时，在 600秒时，输出目标值分别变化为90％、392k，运行一段时间后，在6600秒输出目标值又变化为55％、392k，整个仿真过程中，烟气流量保持在550kg/s不变。系统总共运行12600秒，为方便观察比较，以30秒为采样周期进行取点、绘图。由图3(a)-图4(b)所示可知，基于BP神经网络的预测控制器控制效果更好，波动小，响应速度快；同时，由于神经网络模型的准确预测，本申请涉及的控制器无稳态误差；然而，基于线性状态空间的预测控制预测精度低，在没有反馈矫正的情况下均呈现较大的稳态误差。The comparison between the control effect of the predictive control method based on BP neural network in this application and the effect of traditional state space predictive control is shown in Fig. 3(a), Fig. 3(b), Fig. 4(a) and Fig. 4(b). When the initial steady-state working condition is flue gas flow rate 550kg/s, u1 =444.284kg/s, u2 =112.732kg/s, y1 =70%, y2 =392k, at 600 seconds, output the target value Change to 90% and 392k respectively. After running for a period of time, the output target value changes to 55% and 392k at 6600 seconds. During the whole simulation process, the flue gas flow rate remains unchanged at 550kg/s. The system runs for a total of 12600 seconds. For the convenience of observation and comparison, the sampling period is 30 seconds for taking points and drawing. As can be seen from Fig. 3(a)-Fig. 4(b), the predictive controller based on BP neural network has better control effect, less fluctuation and fast response speed; at the same time, due to the accurate prediction of the neural network model, the application involves The controller has no steady-state error; however, predictive control based on linear state space has low predictive accuracy and exhibits large steady-state error without feedback correction.
本申请把大型燃煤电站CO2捕集系统作为一个二输入二输出的多变量对象，采用基于BP神经网络的预测控制技术，选取贫液流量及再沸器抽汽流量为控制变量，CO2捕集率及再沸器温度为被控变量。利用BP神经网络工具箱，建立燃煤电站CO2捕集系统神经网络模型，从而可以精准预测捕集系统未来时刻的输出特性，提高控制品质。This application regards the CO2 capture system of a large coal-fired power station as a multi-variable object with two inputs and two outputs, adopts the predictive control technology based on BP neural network, and selects lean liquid flow and reboiler extraction flow as control variables, CO2 The capture rate and reboiler temperature are the controlled variables. Using the BP neural network toolbox, the neural network model of the CO2 capture system of coal-fired power plants can be established, so that the output characteristics of the capture system can be accurately predicted in the future and the control quality can be improved.
Claims (4)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910952786.XACN110737198B (en) | 2019-10-09 | 2019-10-09 | Predictive control method for large coal-fired power plant CO2 capture system based on BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910952786.XACN110737198B (en) | 2019-10-09 | 2019-10-09 | Predictive control method for large coal-fired power plant CO2 capture system based on BP neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110737198A CN110737198A (en) | 2020-01-31 |
| CN110737198Btrue CN110737198B (en) | 2022-11-18 |
Family
ID=69269848
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910952786.XAActiveCN110737198B (en) | 2019-10-09 | 2019-10-09 | Predictive control method for large coal-fired power plant CO2 capture system based on BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110737198B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113393905B (en)* | 2021-06-03 | 2024-03-22 | 东南大学 | Chemical absorption CO 2 Dynamic robust soft measurement system and method for trapping system |
| CN113359451A (en)* | 2021-06-10 | 2021-09-07 | 江苏舜高智能科技有限公司 | CO based on large coal-fired power plant2Collecting system economical efficiency prediction control method |
| CN116679572B (en)* | 2023-08-03 | 2023-09-29 | 北京绿能碳宝科技发展有限公司 | Carbon dioxide trapping self-learning method based on deep Q learning network |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2011261169A1 (en)* | 2010-06-02 | 2013-01-10 | Kc8 Capture Technologies Ltd | Optimising objectives of a plant and a carbon dioxide capture facility |
| CN106842955A (en)* | 2017-03-15 | 2017-06-13 | 东南大学 | CO after burning with exhaust gas volumn Disturbance Rejection2Trapping system forecast Control Algorithm |
| CN107450325A (en)* | 2017-09-06 | 2017-12-08 | 东南大学 | CO after one kind burning2The Multi model Predictive Controllers of trapping system |
| CN108958031A (en)* | 2018-07-19 | 2018-12-07 | 东南大学 | CO after burning2Trap coal generating system predictive coordinated control method |
| CN110026068A (en)* | 2019-04-08 | 2019-07-19 | 东南大学 | A kind of large-scale coal fired power plant CO based on Neural network inverse control2Trapping system and feed forward control method |
| CN110286593A (en)* | 2019-07-02 | 2019-09-27 | 东南大学 | A control method for post-chemisorption combustion CO2 capture system based on multi-objective predictive control |
- 2019
- 2019-10-09CNCN201910952786.XApatent/CN110737198B/enactiveActive
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2011261169A1 (en)* | 2010-06-02 | 2013-01-10 | Kc8 Capture Technologies Ltd | Optimising objectives of a plant and a carbon dioxide capture facility |
| CN106842955A (en)* | 2017-03-15 | 2017-06-13 | 东南大学 | CO after burning with exhaust gas volumn Disturbance Rejection2Trapping system forecast Control Algorithm |
| CN107450325A (en)* | 2017-09-06 | 2017-12-08 | 东南大学 | CO after one kind burning2The Multi model Predictive Controllers of trapping system |
| CN108958031A (en)* | 2018-07-19 | 2018-12-07 | 东南大学 | CO after burning2Trap coal generating system predictive coordinated control method |
| CN110026068A (en)* | 2019-04-08 | 2019-07-19 | 东南大学 | A kind of large-scale coal fired power plant CO based on Neural network inverse control2Trapping system and feed forward control method |
| CN110286593A (en)* | 2019-07-02 | 2019-09-27 | 东南大学 | A control method for post-chemisorption combustion CO2 capture system based on multi-objective predictive control |
Non-Patent Citations (2)
| Title |
|---|
| 《化学吸附燃烧后CO2捕集系统前馈优化控制》;唐炜洁等;《工程热物理学报》;20190930;第40卷(第９期);第1969-1975页* |
| 《基于神经网络的电厂烟气二氧化碳捕集过程建模》;揭超等;《计算机与应用化学》;20150628;第32卷(第6期);第757-760页* |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110737198A (en) | 2020-01-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107450325B (en) | A multi-model predictive control method for post-combustion CO2 capture system | |
| CN110737198B (en) | Predictive control method for large coal-fired power plant CO2 capture system based on BP neural network | |
| CN112286043B (en) | PID parameter setting method based on controlled object step response characteristic data | |
| CN109343349B (en) | SCR flue gas denitration optimal control system and method based on ammonia injection amount compensator | |
| CN110026068B (en) | A CO2 capture system and feedforward control method for large coal-fired power station based on neural network inverse control | |
| CN106842955B (en) | Predictive control method for post-combustion CO2 capture system with flue gas volume disturbance suppression | |
| CN110908351B (en) | A Disturbance Suppression Predictive Control Method for SCR Denitrification System Fused with Support Vector Machine | |
| CN110764419B (en) | CO of large coal-fired power plant2Capture global scheduling and predictive control system and method | |
| CN105045233B (en) | The Optimization Design of PID controller based on time metric in Power Plant Thermal system | |
| CN102841540A (en) | MMPC-based supercritical unit coordination and control method | |
| CN109143872A (en) | A kind of continuous stirred tank reactor course control method for use based on event triggering GDHP | |
| CN105180136A (en) | Thermal-power-plant boiler main steam temperature control method based on fractional order proportional-integral (PI) dynamic matrix | |
| CN101709863B (en) | Hybrid control method for furnace pressure system of coal-fired boiler | |
| CN109032117A (en) | Single loop control system method of evaluating performance based on arma modeling | |
| CN114609986A (en) | Cement decomposing furnace denitration regulation and control optimization system and method based on predictive control | |
| CN106610587A (en) | Temperature multi-model prediction function control method and device | |
| CN201476905U (en) | Neural network PID temperature controlled thermocouple automatic verification system | |
| CN110286593B (en) | A control method for post-combustion carbon dioxide capture system based on multi-objective predictive control | |
| CN117875167B (en) | A method, device and equipment for optimizing process parameters of flue gas CO2 capture device | |
| CN118760016A (en) | A new laboratory remote monitoring and operation maintenance management system based on the Internet of Things | |
| CN118935746A (en) | Temperature control method and device for gas water heater, gas water heater and medium | |
| CN108062021B (en) | Parameter self-tuning method based on partial derivative information for SISO full-format model-free controller | |
| CN116880208A (en) | Thermal power generating unit coordination control method based on stepped generalized predictive control | |
| CN115155310A (en) | Ammonia spraying accurate optimization method for SCR denitration system | |
| CN108154231A (en) | Methods of self-tuning of the MISO full format Non-Model Controller based on systematic error |
Legal Events
| 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |