Regression predictive method of coke oven global collector pressure based on support vector machine

Coke oven collector pressure system is a complex control system with multi-variable, uncertain, nonlinear and strong coupling. Firstly, on the basis of a brief description of the technological process of coke oven collector pressure, a regression prediction model of the coke oven gas mainly collector pressure based on support vector machine (SVM) method is proposed. The simulation results of the actual pressures and the predicted show that the proposed method has higher forecast accuracy, which provides bases on the real-time decoupling control and the set-point optimization based on the operation mode strategy of the coke oven collector pressure system.