Research on Vibration Signal Diagnosis Platform System Based on Computer Fuzzy Logic

An on-line vibration detection method for power transformers is introduced. This design includes sensor module, real-time processing module, PC module. The sensor module collects the vibration, voltage, load voltage and other signals of the motor and transmits them to the processing unit on site. The real-time processing module transmits the processed information to the PC for display, storage, analysis and other functions. A real-time monitoring system for power transformer vibration based on virtual instrument /Compact RIO is developed. A new signal detection algorithm is established by combining wavelet energy information entropy theory with SVM. A frequency domain analysis algorithm based on wavelet analysis is proposed. The normalized feature vector is the input signal of support vector machine. Then a support vector machine (SVM) multi-classifier model based on multi-class learning is proposed. Experimental results show that the SVM classifier proposed in this paper has high classification accuracy under multiple working conditions of power transformer.