Intelligent fault diagnosis technology based on hybrid algorithm

In order to diagnose the mechanical fault of complex system conveniently and accurately, a hybrid intelligent fault diagnosis technique based on hybrid algorithm is established by collecting the different signals of the equipment. For vibration signals, wavelet packet decomposition and rough sets reduction algorithm is used to extract feature vector of signal and reduce this vector into effective decision-making table. For the signals of temperature, pressure and other data, the data of each sensor are fused to form the feature vector, and the Back Propagation(BP) neural network optimized by genetic algorithm is used to train and pattern recognition. Finally, the Dempster-Shafer (D-S) decision fusion method is adopted to fuse the diagnostic results of two kinds of signals. This composite fault diagnosis model has a high diagnostic accuracy and precision. For ease of use, a signal processing platform was designed based on MATLAB language and using Graphical User Interface (GUI).