【摘要】 提出了以系统脉冲响应序列作为闭环系统模型的建模方法,并且给出了这种建模方法的计算方法以及模型输出的计算方法。这种闭环系统建模方法保留了脉冲响应序列中的高频成分,避免了从脉冲响应序列向传递函数变换过程中丢失脉冲响应函数高频信息的缺点;同时,也避免了神经网络每次学习结果获得的网络模型不确定的缺点。采用该方法建立的闭环系统模型精度非常高并且可以应用在故障诊断的系统中。文章最后以数值仿真和实际带随机干扰的系统为例,证明该方法是有效的。更多还原

【Abstract】 The paper presents a new modeling method applied to fault diagnosis for constant linear closed-loop system by taking the impulse response series as the system model, and provides the calculation process of the method and output of model. The high frequency part of the pulse series, in the method, is reversed so as not to lose the frequency information of the pulse series in its transfer function. On the other hand, the method can also avoid the disadvantage that the learning results of neural network are uncertain every time. In the last part, the application with random disturbance of digital simulation and practical system shows that the modeling method is high accurate and suitable to be applied in fault diagnosis area.更多还原