Development Of Coal-fired Units Running Optimization Software Based On RBF Neural Network

Under the background of the scientification and datamation developed,the demand of the electricity is becoming higher and higher in all walks of life.In China,thermal power is the main form of power generation,while coal as the main way of thermal pow-er generation has always been a core position in the power industry at home and abroad.In order to improve the combustion efficiency of boiler,reduce the amount of NO_X emission and run the combustion system of boiler stably,the optimization of power plant boiler is becoming the primary task at present.This paper selects the 660MW supercriti-cal unit boiler of a domestic power plant as the research object,and studies the optimiza-tion of reducing discharge and increasing efficiency of coal fired units.Based on the his-torical data of operation and combined with RBF neural network,a network model was built to predict NO_X emission and thermal efficiency of power plant boilers which used boiler operating parameters as input,NO_X emission and boiler thermal efficiency as out-put.In the process of modeling,optimizing the centers which are the hidden nodes of RBF neural network by utilizing particle swarm algorithm with weight(PSO)to enhance the network's capability of nonlinear approximation and improve the convergence speed of the model.Then testing the accuracy of the improved network model by test sam-ple.And experiment shows that the established network model can represent the combus-tion characteristics of boiler better.Based on the above model,the optimized model is established which selects the NO_X emission and boiler efficiency as the goal,the given coal quantity,once throttle opening,twice throttle opening and burn throttle opening as the optimization variables.And using genetic algorithm and multi-objective optimized function to optimize.The al-gorithm only needs two parameters which are given population size and the number of iterations.With an operation to obtain a set of Pareto optimal solutions,the optimal solu-tions of the model generated finally.This paper uses C#programming language to complete the software development and system optimization operation of boiler,including the design and description of the function modules,the system mainly realizes the data acquisition,parameter monitoring,parameter prediction and parameter optimization function,access to data is mainly ob-tained through real-time database system in SIS and CEMS.The real-time monitoring of main parameters of boiler will realize the visualization function of the current operation and change trend.Through the established optimization model,we can give the adjust-ment plan of parameter optimization for the current working condition,guide the staff to adjust,and realize the optimization function of the boiler optimization system.The sys-tem not only meets the environmental protection department's low emission requirements for NO_X emission index,improves the combustion efficiency of power plant boiler,but also realizes intelligent management of boiler operation,which has a positive impact on China's environmental protection and electric power enterprises.