Study on multi-fractal fault diagnosis based on EMD fusion in hydraulic engineering

Abstract

The vibration signal analysis of the hydraulic turbine unit aims at extracting the characteristic information of the unit vibration. The effective signal processing and information extraction are the key to state monitoring and fault diagnosis of the hydraulic turbine unit. In this paper, the vibration fault diagnosis model is established, which combines EMD, multi-fractal spectrum and modified BP neural network; the vibration signal waveform is identified and purified with EMD to obtain approximation coefficient of various fault signals; the characteristic vector of the vibration fault is acquired with the multi-fractal spectrum algorithm, which is classified and identified as input vector of BP neural network. The signal characteristics are extracted through the waveform, the diagnosis and identification are carried out in combination of the multi-fractal spectrum to provide a new method for fault diagnosis of the hydraulic turbine unit. After the application test, the results show that the method can improve the intelligence and humanization of diagnosis, enhance the man–machine interaction, and produce satisfactory identification result.