神经网络-标称系统混合模型的离散变结构控制

提出将神经网络和标称系统混合建模方法引入到柔性结构主动控制当中，在混合模型的基础上，利用离散变结构控制（VSC）对柔性结构振动进行控制．离散变结构控制的滑模面是以标称系统为基础，由最优化二次型价值函数确定，并通过黎卡提方程求解．利用标称模型和神经网络混合建模方法来减小系统的不确定性，达到减弱变结构控制在实际控制系统中的抖动问题．神经网络采用多层前馈网络（MFNN），来对不确定部分进行建模．仿真结果表明系统振动受到了有效的控制，说明提出的神经网络变结构控制（NNVSC）方法非常有效。

Discrete variable structure control for flexible structure vibration reduction based on hybrid modelling of neural network

In this paper,the hybrid modelling of neural network is introduced into the active control of the flexible structure vibration,which is controlled by discrete Variable Structure Control (VSC) method based on the hybrid model.The the discrete sliding mode surface is determined by the quadratic optimal cost function and solved from the algebraic Riccati e quation based on the nominal system.To reduce the chatter coming from the large uncertainty of the system,the hybrid model of neural network and nominal system is adopted.The Neural Network (NN) method that is multilayer feed_forward neural network (MFNN) is adopted to model the uncertainty part.VSC is determined by the hybrid model of the nominal model and NN.The simulation shows that the vibration of the structure has been reduced clearly,which indicates the Neural Network Variable Structure Control (NNVSC) scheme is very effective for the large uncertainty system.