Research On Compound Control Of Coke Oven Gas Collector Pressure System Based On Multi Kernel Function SVM

Industrial production needs a lot of resources.As the most commonly used resource,coal resources occupy the highest proportion in the consumption of resources.As the main way of clean transformation of coal resources,coking has far-reaching significance for the development of national economy and the improvement of ecological environment.Gas collector pressure system is an important part of resource transformation process.There are a series of difficulties that can hardly be controlled by conventional way.Because the gas collector pressure system is a multivariable nonlinear coupling system with large interference.It is necessary to understand its model characteristics so that it can be controlled steadily.In this paper,the theory of system identification is introduced,and multi kernel support vector machine(MKSVM)algorithm is used to identify the system structure.By introducing the statistical learning theory,the empirical risk minimization principle and the structural risk minimization principle are analyzed,and the support vector machine(SVM)algorithm is introduced from the latter.By analyzing the effect of kernel function on the accuracy of SVM identification,a weighted combination of the global kernel function and the local kernel function is selected,and the final analysis error is used to select the optimal weighted model.Based on the inverse system theory,the inverse model of the collector pressure system based on SVM is identified and added to the original system to decouple the system.The inverse model control based on SVM can decouple and realize the linearization transformation of the original system.In order to deal with the phenomenon of mismatch or large disturbance,a compound handover control mode based on PID control is designed.The switching between controllers is carried out according to the switching criterion,and the system is quickly separated from high voltage or low voltage under large interference.Thus the anti-interference and robustness of the system are improved.Finally,the MATLAB platform is used to simulate the designed scheme.The results show that the control scheme can decouple the original system and achieve the linearized tracking between input and output of the system.The compound switching control based on SVM can suppress interference and has good robustness,which verifies the feasibility of the control scheme designed in this paper.