基于神经网络的动态系统逆模型辨识及闭环控制(英文)

Inverse Identification and Closed-Loop Control of Dynamic Systems Using Neural Networks

本文提出一种动态线性或非线性系统的神经网络逆模型辨识结构．并引出两种PID与神经网络逆模型相结合的自适应控制方案．神经网络模型采用基于U－D分解卡尔曼滤波学习算法（UDK）的动态前向多层同．仿真结果表明了所述辨识方案的有效性及特点．

A structure for inverse identification of dynamic linear or non-linear system usingneural network is presented. Two types of closed-loop control schemes that combine the neuralnetwork inverse model with PID are proposed. The dynamic feed forward multilayer network for identification and control is trained by a novel learning algorithm based on U-D factorizationKalman filter (UDK). The potentials of the proposed structure and schemes are demonstrated bysimulation studies.