Research On Dynamic Process Modeling Of Electric Boiler Based On Neural Network

The configuration of the electrode boiler in the thermal power plant contribute to the thermal electrolysis coupling of the thermal power unit,which can improve the peak shaving capability of the unit and the flexibility of the system operation,thereby reducing wind power abandonment.There is a strong coupling and nonlinear relationship among the parameters of the electrode boiler.Neural networks have good complex dynamic system modeling capabilities and are increasingly used.Therefore,the neural network modeling research of electrode boilers is of great significance.In this paper,the structural principle and control flow of the electric boiler system were analyzed in detail for the 15 MW electrode boiler.Based on a deep understanding of the different structural principles,learning algorithms and identification structures of artificial neural networks,the two structures of time delay BP neural network and Elman neural network were compared,and the gradient descent method,Levenberg-Marquardt algorithm and Bayesian regularization were analyzed.Then,a first-order delayed neural network prediction model for the electric power and main steam pressure of the electrode boiler were established.Secondly,aiming at the shortcomings of particle swarm optimization algorithm in neural network order optimization,a restarting strategy was introduced,and an improved particle swarm optimization algorithm was proposed.This method was used to optimize the delay order of neural network parameters,and then the optimal electric boiler prediction model was obtained.By comparing the dynamic data under various disturbances,it is verified that the neural network model has higher prediction accuracy.Finally,the MATLAB software was used to make the electrode boiler and the 350 MW unit simulation unit communicate in real time and a combined model was established.Through simulation experiments,it is verified that there is a power conservation relationship between the combined models,and it is concluded that the electric power of the electrode boiler can effectively and quickly affect the power consumption of the unit.The work done in this paper provides a theoretical basis for the coordinated control of the electrode type boiler in the deep peak peaking of the source side of the thermal power plant,which is beneficial to the engineering application of the electrode boiler in the thermal power plant.