The Wind Turbine Vibration Signal Analysis Based On The LMD

With the widely application of the wind turbine, its security gets more and moreattention. In order to ensure safe, reliable and efficient operation of the wind turbine,fault diagnosis technology has been deep researched and some importantdevelopment has been used. In the operation process, the vibration signal is verycomplex and non-stationary with the form of multiple components. In recentlydecade years, most analysis method, such as Fourier Transform, Short-time FourierTransform, Winger Distribution and Wavelet Transform method has some limitations.Comparing and analyzing these common time-frequency analyses, this paperproposes an adaptive analysis, Local mean decomposition (LMD). LMD methodcombined with order analysis applied to the fault diagnosis of the wind turbine.Studying the wind turbine fault feature extraction method based on LMD, combingwith the Order Analysis to analysis the rotating machine fault signal.The mainresearch content is as follow:LMD theory and some basic concepts are expounded, and can be used foranalyzing the wind turbine. Studying the LMD algorithm and solving the problemssuch as selecting the sliding step. Against sliding step selection influence on LMDdecomposition, this paper proposes an adaptive method of automatically determinethe sliding step, and against the noise problem, this paper proposes a de-noiseprocessing. To extract the characteristics of the wind turbine vibration signal, usingthe improved LMD method decomposes the vibration signal.The direct method and energy operator demodulation method are compared.The experimental validating the direct method is an effective method of extractingthe instantaneous frequency of the rotating machinery of the wind turbine.Put forward the order fault diagnosis method based on the estimation ofinstantaneous frequency. Through the simulation estimates the order analysis method is an effective method for vibration analysis. LMD and the direct method are used todetermine the sampling time of the constant incremental angle, obtaining theconstant angle increment sampling signal and conducting the order analysis.Using the bearing vibration data validation the order analysis based oninstantaneous frequency estimation of LMD is effective.