基于模型跟随的神经网络非线性重构控制

Model-Following Non-Linear Reconfigurable Control Based on RBF Neural Networks

针对歼击机在结构故障下的动力学方程 ,提出了一种基于径向基函数 (RBF)神经网络的模型跟随非线性重构控制策略。该方法不必精确已知系统故障的位置及其损伤程度 ,可直接对故障系统实施重构控制 ,使其输出能精确跟踪期望参考模型的输出。该方法在模型跟随重构控制的基础上 ,引入了神经网络控制器 ,以补偿故障引起的非线性因素的影响。理论分析和仿真验证表明 ,所提方法可保证闭环系统具有良好的重构性能和很强的鲁棒性 ,且算法高效简单 ,易于计算机在线控制。

A new type of non-linear reconfigurable control strategy based on model-following method using radial basis function (RBF) neural networks for the fight aircraft structural damage is presented in this paper. This method can make the outputs of impaired system tracking those of reference model accurately without knowing the location and damage degree of failure. It is formed on the foundation of model-following adaptive control and a RBF neural networks is used to compensate non-linear dynamics caused by failure. Theoretical analysis and simulation results reveal that this method has good reconfigurable performance and robustness. Furthermore, this control strategy is very simple and easy for computer to handle on-line.\;