SVM modeling on slow time varying system and online correction studying

The problem of identification and online correction for slow time-varying system has become a hotspot. Based on the traditional modeling algorithm and their shortages, in this paper, the intelligence algorithm AO-SVM is studied intensively. Because of the defects of AO-SVM, such as the sampling deletion strategy and instability of the algorithm, the improved AO-SVM which is combined with ZD-SVM is proposed. The simulation results show that the improved algorithm ensures the accuracy of modeling and avoids the data saturation.