论文部分内容阅读
提出了一种关于变工作点非线性系统故障诊断研究的新方法，将此类非线性系统用变参数线性系统表示，其中模型参数为可测量工作点及故障的函数。基于基函数神经网络，运用带“遗忘因子”递推最小二乘法估计系统模型参数。此外引入“参考工作点”这一新概念。预先训练出系统在各工作点健康运行时相对于参考工作点运行时的模型参数变化量，当系统在任意工作点运行且出现故障时，可将系统工作点变动和故障发生这两者造成的影响区分开来，然后根据故障诊断决策规则确定故障种类。最后，在某位置伺服系统故障诊断研究中证实了这种方法的有效性。
A new method for fault diagnosis of nonlinear system with variable working point is proposed. The nonlinear system is represented by a variable parameter linear system, in which the model parameters are functions of measurable working point and fault. Based on basis function neural network, the parameters of system model are estimated by recursive least square with “forgetting factor”. In addition, the introduction of a “reference point” this new concept. Pre-train the system in the healthy operation of the operating point relative to the reference point of operation when the model parameters change, when the system at any point of operation and failure, the system can change the operating point and the failure of both Distinguish the impact, and then determine the type of fault according to fault diagnosis decision rules. Finally, the effectiveness of this method is verified in a fault diagnosis study of a position servo system.