Study On Wind Turbine Gearbox Vibration Signal Analysis Based On FastICA Algorithm

Quality problems emerging behind the rapid growth in the installed capacity of windturbine has become the focus of the industry worries.With the accumulation of running time,comprehensive failure rate in gearbox is the highest among the hydraulic, electronic controland mechanical transmission systems of the wind turbines installed in the wind farm. As theacquisition monitoring signal of wind turbine gearbox is complex, an ideal signal analysismethod is need, in order to ensure timely and effective monitoring the state of the turbinegearbox and making accurate fault diagnosis.FastICA algorithm based on the negative entropy belongs to one of the independentcomponent analysis.This algorithm based on the signal statistical characteristics to recoverthe source signals without any a priori knowledge, neither depend on the signal frequency,norneed priori knowledge, has fast convergence rate, the separated signal loss is small and easierto do further analysis. Is likely to be the ideal analysis and processing tool for the windturbine gearbox signals.This paper studies on the signal of the wind turbine gearbox system.Range from thechoosing and the placement of the sensor of gearbox to the collecting of the vibration signalof the running unit and writing MATLAB separation program to separate the signals.Thispaper analyzed the strengths and weaknesses of the current common signal processingmethod carefully, combined with the analysis and processing characteristics and actualrequirements, proposed the application of independent component analysis algorithm basedon the FastICA in the analysis of wind turbine gearbox signals,make validation to thisapplication in SIMULINK simulation and actual1000KW wind turbine gearbox environment.This paper described the placement and installation of gearbox sensor, the structure offront-end devices (The JK07460Wind Turbines online fault diagnosis system of Tang ChiTechnology) and the process to acquire signals.This paper emphasized the principle andprocess of separation algorithm based on the FastICA to separate signal and make a simpleassessment to the experimental results.Got two sets of data belong to the1000KW wind turbine gearbox input shaft and output shaft through the separation procedure to the collectedshock and vibration signal,the experimental results and the expected results are consistent.Eventually come to this algorithm applied to the analysis of wind turbine gearbox signal isreasonable and promising.