Direct Blowing Pulverizing System Modeling Based on IPLS-SVM

In order to solve the problem that the output of ball mill pulverizing system is difficult to directly measured in thermal power plant with double inlet and double outlet ball mill pulverizing system which is a large delay, strong nonlinear system. It introduces the pruning method to improve the incremental least square support vector machine´s sparsity that based on the incremental least square support vector machine algorithm. The LS-SVM model is simplifies by not only deleting some too big or too small training samples at the same time, but also deleting the large change rate data. It avoids the influence of bad samples on model, and simplifies the LS-SVM model. Incremental pruning least squares support vector machine algorithm (IPLS-SVM) is used for the soft measurement model of direct blowing pulverizing steel ball coal mill with double inlet and double outlet. Compared the modified model for simulation, simulation results show that the speed of convergence is faster, more suitable for online learning.