Fault Diagnosis Method Based on Indiscernibility and Dynamic Kernel Principal Component Analysis

In order to solve the problem of complex industrial system dynamic, non-linear detection accuracy and calculation load, the fault diagnosis method is proposed based on indiscernibility and dynamic kernel principal component analysis(IDKPCA). Reduce the amount of data by using the degree of indifferentiability. Extension through observing the screening of the new matrix to construct augmented matrix, and the matrix using kernel principal component analysis (KPCA) extracting nonlinear spatial correlation characteristics of variable data, finally detected by monitoring statistics system fault, with the method of the contribution to identify the fault variables. This method improves the traditional dynamic methods, and can give full consideration to the nonlinear and dynamic in the process of industrial, more precise description of the industrial process features, more accurate monitoring of complex industrial system fault, and accurately identify the fault variables. The improved algorithm reduces the leakage rate and false alarm rate, improves the diagnostic reliability, and can detect the minor faults in the production process in time. The method is applied to the fault diagnosis of wind turbines. By comparing with KPCA method, better fault detection results are obtained.