论文部分内容阅读
鉴于对线性系统的故障诊断方法比较成熟,而对非线性系统的故障诊断还有很大不足的状况,将非线性系统在多个工作点进行分段线性化。在非线性系统中借鉴线性故障诊断方法,对每个工作点的模型进行线性故障诊断设计,取得工作点密度和故障诊断性能的均衡,同时使用动态神经网络对各个工作点的工作参数进行拟合。该方法将一组改进的鲁棒观测器与一个为其确定参数的动态神经网络相结合,对非线性系统进行故障诊断。用典型的3水箱模型验证了这个合成方法,验证结果显示,该方法有很好的全局工作点适应性和对干扰的鲁棒性。
In view of the maturity of the fault diagnosis of linear systems, and the failure diagnosis of nonlinear systems, the nonlinear system is piecewise linearized at multiple working points. In the nonlinear system, the linear fault diagnosis method is used to diagnose and design the linear fault of each working point. The working point density and fault diagnosis performance are balanced, and the dynamic neural network is used to fit the working parameters of each working point . In this method, a set of improved robust observers is combined with a dynamic neural network to determine the parameters for fault diagnosis of nonlinear systems. The typical three-tank model is used to verify this synthetic method. The verification results show that this method has good global adaptability and robustness against interference.