Aiming at dynamic system identification problem, a method of nonlinear dynamic system identification by output feedback recurrent fuzzy neural network was proposed by using a reconfigurable field-programmable gate array (FPGA). The output feedback recurrent fuzzy neural network can realize high-precision and high-speed identification in nonlinear systems, but neural network complex parallelism required a longer time, its application was limited in real-time control and online system identification. According to the parallelism, the FPGA implementation was superior in neural network. Therefore, FPGA logic function was defined by using high-〖JP〗performance FPGA card of NI company and LabVIEW graphical programming. The output feedback recurrent fuzzy neural network with dynamic back-propagation algorithm and the online identification of the rudder position servo system were realized. The results of the experiment confirm that the identification cycle of system can achieve in the millisecond range based on FPGA neural network. It provides a new approach to the application of nonlinear dynamic system identification.
参考文献
相似文献
引证文献
引用本文
王晓华,王有松.基于FPGA的神经网络在非线性实时系统辨识中的应用[J].机床与液压,2022,50(20):174-178. WANG Xiaohua, WANG Yousong. Neural Network Application in the Nonlinear Real-Time System Identification Based on FPGA[J]. Machine Tool & Hydraulics,2022,50(20):174-178