Chaotic Modeling of Permanent Magnet Synchronous Motor Based on Multiple Kernel Symmetric Least Squares Support Vector Machines

In this paper, a multiple kernel symmetric least squares support vector machine (MKSLSSVM) regression modeling method is proposed for the case that chaotic characteristics are displayed in the permanent magnet synchronous motors (PMSM) under certain circumstances and the exact chaotic model is difficult to obtain. A symmetric constraint condition is added to the least squares support vector machine (LSSVM) model to construct the symmetric LSSVM (SLSSVM). Then, SLSSVM is integrated with multiple kernel learning technique to form a novel equivalent kernel, which is composed of linear combination of multi basic kernels. This novel equivalent kernel can be employed for the chaotic modeling of PMSM. Simulation results show that, compared with LSSVM, the proposed scheme can reduce the effect of modeling error caused by selecting of kernel function and enhance the chaos modeling accuracy.