论文部分内容阅读
给出用神经网络（ Ｎ Ｎ）α阶积分逆系统实现连续非线性 Ｍ Ｉ Ｍ Ｏ 系统线性化解耦的方法。 Ｎ Ｎα阶积分逆系统由一个静态神经网络加若干积分器构成，将其串联在原系统之前，原系统则解耦成若干个相互无关的 Ｓ Ｉ Ｓ Ｏ 伪线性积分系统。理论分析与仿真结果表明，对于精确模型未知的较一般的非线性 Ｍ Ｉ Ｍ Ｏ 系统，所给出的方法均能实现有效的线性化解耦，且结构简单，易于工程实现。
The method to realize the linearization decoupling of continuous nonlinear M I M O system by using the neural network (N N) α order integral inverse system is given. N Nα-order integral inverse system consists of a static neural network and several integrators, which are connected in series before the original system. The original system is decoupled into several unrelated S I S O pseudo-linear integral systems. The theoretical analysis and simulation results show that the proposed method can achieve effective linearization decoupling for the more general nonlinear M I M O system with unknown exact model, and the structure is simple and easy to implement.