The Fault Diagnosis Research Of The Frequency Convetrer In The Wind Power System Based On Intelligent Algorithm

With a rising installed capacity of wind farms year by year, design and manufacture problem of the wind generator is not difficult. However, maintenance and repair of the failure is an inevitable promblem after these wind power equipment in operation a period of time. Therefore, how to improve the wind power system reliability and maintain this already installed unit’s normal operation is becoming an important topic for scientific and technical workers.The function of full-scale power converter is used for keeping the frequency, stabilizing the phase, controling voltage changes with the wave of the power grid and cooperating with master control system for realizing largest wind energy utilization, if the converter failure has not handled in time, it will reduce the quality of power supply, affect the whole wind generator power generation, and even endanger the grid. The IGBT of the full-scale power converter bear the change of the temperature and the surrounding environment influence during working time, so the IGBT lead to converter failure easily. If we can locate the fault quickly, point out the fault type accurate and give maintenance timely when the power device IGBT have weak fault symptoms, these actions can reduce the grid of failure, improve the reliability of wind power generation system.This thesis mainly for the converter fault diagnosis of direct-driven permanent magnet synchronous wind power systems of IGBT when appear weak fault, extract the DC bus voltage fault waveform feature by wavelet packet transform, then using intelligent optimization algorithm for support vector machine classification model optimization, fast and effective diagnosed IGBT the fault occurred position and fault type, this is novel fault diagnosis method for online wind power monitoring system. The main works and conclusions in this paper are as following:(1) This thesis built the simulation model of direct-driven permanent magnet synchronous wind power systems by the PSCAD simulation software, mimic different open-circuit fault and short-circuit fault, then, extract the data of DC bus voltage as the failure information;(2) Extract the feature vector by wavelet packet decomposition method after get the fault waveform, the model of support vector machine classifier is trained by fault feature vector; (3) In order to obtain the precise location for fault point IGBT and effective identification for converter failure types, genetic algorithm, particle swarm optimization and improved particle swarm optimization are used to optimal the parameters of support vector machine diagnosis model, then, collecting new fault data prove the validity of the model. The simulation results show:the improved version of the particle swarm algorithm diagnosis effect is good.