摘要
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。
A new dynamic neural network construction named NN α -th order integral inverse system is presented to decouple nonlinear MIMO continuous-time system with linearization. The NN α -th order integral inverse system consists of a single static NN and several integrators. Connecting it directly with the original nonlinear MIMO system, one can decouple and linearize the nonlinear coupled system into a number of independent pseudo linear SISO systems. Both the theoretical analysis and simulation results show that this decoupling method is effectiveness for a general class of nonlinear MIMOcontinuoussystemeven if the system′s mathematical model is unknown. Besides, construction of NN α- th order integral inverse system is simple for easy implementation.
出处
《控制与决策》
EI
CSCD
北大核心
1999年第5期403-406,412,共5页
Control and Decision
基金
国家自然科学基金
江苏省自然科学基金
关键词
MIMO
非线性系统
线性化
解耦
连续时间系统
MIMO, nonlinear system, linearization, decoupling, neural network, inverse system