【摘要】 研究如何构造神经网络（ＮＮ） α阶时延逆系统，并将其用于ＭＩＭＯ 强耦合非线性离散时间系统的线性化解耦。ＮＮα阶时延逆系统由单个静态神经网络和若干时延因子组成，将其与原系统复合可形成具有最小阶时延的伪线性解耦系统。该方法对被控系统数学模型的先验知识要求很少，不但能有效地实现对原系统的线性化解耦，且结构简单、易于工程实现更多还原

【Abstract】 This paper focuses on constructing neural network(NN) α-th order time-delay inverse and using it to decouple and linearize MIMO nonlinear discrete-time system. The NN α-th order inverse of the MIMO nonlinear system is composed of a static neural network and a number of time-delay factors(z\+ -1 ). When it is simply cascaded with the original nonlinear MIMO system, a number of independent pseudo-linear SISO systems with the minimum-order time-delays are obtained. Simulation results show that the proposed decoupling method is very effective even for model unknown MIMO system and is easy to be put into enginearing.更多还原