Research on Weighted Least Square Optimization Algorithm of Nonlinear System Analysis

Abstract

In the actual control system, the controlled object is usually nonlinear. Therefore, nonlinear time series analysis has been always received considered attention from many scholars. The ability to accurately model is critical for nonlinear time series analysis. However, the challenge in system modeling is how to estimate the model parameter. Unlike the traditional parameter estimation method with the defects of low precision, poor convergence and long time optimization, this paper proposes a new optimization algorithm of nonlinear time series which combine the advantages of fast convergence near the minimum value and being able to converg for any initial value. During the calculation, the first-order derivative should be solved while the inverse matrix is not necessary. The simulation results show that the proposed method not only ensures the convergence of iteration planning but also improves the convergent speed. It can be applied the nonlinear time series analysis and provides the powerful guarantee for accurate trends prediction.