Predictive Controller Design For Coke Oven Blowing Cooler System Based On SVR

Coke industry is a pillar industry in China,its process is complex and multi-interference,production links are linked.One of the important links is the pressure control of gas collecting pipe,the quality of pressure control directly affects the working state of the whole Coke oven,and has a direct relationship with the quality of Coke,crude gas transmission.As an important part of gas collecting system of coke oven,it is difficult to obtain accurate mathematical model by traditional modeling method because of strong coupling,strong interference and nonlinearity.Aiming at the above problems,it is very important to find the modeling method for cold drum system and design advanced control algorithm.After reading lots of related literatures and investigating on the spot,the operation state of drum cooling system is divided into three working conditions according to the gas output,namely maintenance and insulation,normal production and coking coal.A model based on Support vector regression(SVR)can directly reflect the internal connection of coking furnace cold drum system.The system model needed for model predictive control(MPC)is obtained by SVR identification.In order to optimize the forecasting controller of coking furnace cold drum based on SVR model,an on-line rolling optimization controller based on APSO-SVR algorithm was designed by adaptive weight specific optimization(APSO)algorithm.The problems of system model identification,optimization control design,feedback correction,selection of algorithm parameters are solved.Finally,programming on the MATLAB platform.