Fault diagnosis method of hydraulic system based on multi-source information fusion and fractal dimension

Considering nonlinear characteristics of the signals measured by multiple sensors in a hydraulic system, a novel fault diagnosis method of the hydraulic system based on multi-source information fusion and fractal dimension is proposed which mainly includes multi-source information’s feature extraction and sample construction methods. First, the sensor signals dealt with wavelet transform are decomposed into a finite number of intrinsic mode functions (IMFs) by ensemble empirical mode decomposition, and the top five IMFs with more characteristic information are determined. Then, a Grassberger-Procaccia-based algorithm is proposed to calculate the fractal dimension of the IMFs under different embedding dimensions. After that, considering the change laws of fractal dimensions with embedding dimensions, third-order tensor samples of the sensor signals are constructed. At last, fault pattern recognition of the system is realized by using k-nearest neighbor algorithm and fuzzy clustering method, respectively. Simulation results indicate that the proposed method can improve the average accuracy of fault diagnosis by about 10%, which provides theoretical and technical support for the multi-sensor monitoring and diagnosis of complex nonlinear systems.