Soft sensor of vertical mill material layer based on LS-SVM

This note describes a soft sensor model for measuring indirectly material layer in the vertical mill grinding process. It is based on the method of least squares support vector machine. By process mechanism analysis, secondary variables are determined. Some optimal parameters are testified through cross validation, and then we do model simulation analysis. The test results display that the soft senor model achieves the on-line prediction to resolve the problem that material layer is difficult to measure directly. When it is applied to the actual production process, it is beneficial to the stability and continuous of vertical mill, and it could guide the operator to change parameters of the equipment timely to further optimize the vertical mill control system.