For Wind Turbine Gearbox Bearing Fault Diagnosis Technology Research

Gear box is an important part of wind turbine drive system, gear box of the rolling bearing is one of the higher failure rate of transmission chain parts. Bearing diagnosis not only influences the normal operation of the unit, and even affected the safe and stable operation of the power supply side of the power grid. So make fast diagnosis for the wind turbine gearbox bearing fault has important realistic significance and practical value. Detailed analysis of the basic parts of a wind turbine structure, failure mechanism, characteristics and characteristic frequency. For wind turbine mechanical drive system of common mechanical failure problems, such as gear box broken teeth, pitting corrosion, wear, eccentricity, bearing inner ring and outer ring, rolling body damage failure problems. For nonlinear and non-stationary noise fault vibration signal characteristics, using the Empirical Mode Decomposition (EMD) method of noise vibration signal de-noising processing, and then extract the fault feature, research and analysis of the failure type. The fault diagnosis technology, modern time-frequency analysis technology and theory of rotating machinery fault knowledge, type of doubly-fed wind turbine drive system comprehensive analysis and research of mechanical vibration fault diagnosis. Finally using intelligent Support Vector Machines (SVM) classification algorithm for generating sets of gear and rolling bearing fault diagnosis analysis and the classification research, failure data of the measured experiment verify the feasibility and accuracy of fault diagnosis algorithm. For wind turbine so online diagnosis to provide a train of thought and method. The research content and conclusions are as follows:(1) In this dissertation, the mechanical fault diagnosis theory analysis and research of wind turbine gearboxes rolling bearing mechanical vibration fault mechanism, analysis of the characteristic frequency of the parts failure, understand the different failure of failure characteristics, fault diagnosis and classification for subsequent experiments provide a theoretical basis.(2) The fault data obtained through the experiment, using wavelet decomposition and EMD method for noise reduction processing, spectrum, envelope spectrum analysis, the fault characteristics of spectrum analysis, verify with the improved EMD feasibility and superiority of the noise reduction method.(3) To select Feng-Feng failure data, the effective value, variance and kurtosis index as its eigenvector. By training SVM algorithm of fault feature vector, to form the basic fault diagnosis model, then the grid search method with three parameters optimization algorithms. Genetic Algorithm(GA), Particle Swarm Optimization (PSO) algorithm for SVM diagnosis model for parameter optimization, to obtain the gearbox precise localization of rolling bearing fault points and fault type of effective identification, simulation results show that grid search method although some calculation speed is relatively fast, but the fault type classification accuracy is lov Genetic algorithm is easy to fall into local optimum and computing speed is relatively slow, the inefficiency of classification. Particle swarm optimization algorithm is the highest classification accuracy, computing time between the front two methods.