A Nonlinear System State Estimation Method Based on Adaptive Fusion of Multiple Kernel Functions

With the development of industry, physical model controlled object tends to be complicated and unknown. It is particularly important estimate state variables a nonlinear system when This paper proposes estimation method based on adaptive fusion multiple kernel functions improve accuracy estimation. First, dynamic neural network used build model, where function node constructed by weighted linear combination local global functions. Then, weight are put together form an augmented vector, which can estimated in real time using high-degree cubature Kalman filter. The filter performs weights according specific samples, makes approximate follows value. Finally, simulation results verify feasibility effectiveness proposed algorithm.