Characteristic Modeling And Optimal Control Simulation Research Of Reheated Steam Temperature Based On Neural Network

As one of the key parts of the control system of thermal power units,reheat steam temperature is generally adjusted by combining the baffle on the flue gas side(or the swing angle of the burner)with the water spray temperature reduction on the steam side.Due to the strong inertia,delay and nonlinear characteristics of flue gas side regulation,the flue gas baffle(or burner swing angle)automatic operation rate is low,the adjustment effect is poor,and the operation can not be well coordinated with the water spray temperature reduction,resulting in excessive amount of reheat water spray,reducing the operation economy of the unit.In addition,the unit participates in AGC deep peak regulation and frequent load changing operation,which increases the control difficulty of reheat steam temperature,making it difficult for conventional PID control to achieve the desired control effect.So the research of advanced reheat steam temperature optimization control schemes is of great significance to the safe and economic operation of thermal power units.Therefore,taking a 600 MW supercritical unit as the research object,the modeling and optimization control of reheat steam temperature characteristics based on neural network are carried out,and the simulation and verification of the model and control scheme are carried out with thermal power unit simulator.Based on the detailed analysis of the characteristics and influencing factors of reheat steam temperature,the prediction model of reheat steam temperature is established by using BP neural network with different order time delay for inputs and output feedback.By comparing the on-line prediction effects of different order time delay models under load disturbance,a better characteristic model structure of reheat steam temperature is established.A simplified particle swarm optimization(s PSO)algorithm is used to optimize the initial weights and thresholds of the neural network,and then the off-line training of the model is completed.The steam temperature characteristic model is verified by on-line simulation,and the results show that the prediction model has high accuracy.Based on the above neural network prediction model,combined with s PSO algorithm,a reheat steam temperature predictive optimal control scheme based on neural network is proposed.The reheat steam temperature predictive optimization real-time control algorithm is compiled by MATLAB,and the detailed optimal control simulation experiment is carried out in the full-range thermal power unit simulator of supercritical unit.The experimental results show that the reheated steam temperature predictive optimal control scheme proposed in this paper can make the reheated steam temperature track the set value quickly,obviously improve the lag of steam temperature response,and effectively improve the control quality of the system.At the same time,on the premise of ensuring the reheat steam temperature stability,the optimized control strategy realizes the coordinated control of flue gas baffle and water spray,which can effectively reduce the use frequency and water spray volume of water spray temperature control,improve the regulation performance of reheat steam temperature control system,and improve the operation economy of the unit.