Fault Diagnosis Study Of The Frequency Converter In The Wind Power System

As our country has put forward a plan to reduce carbon release, most provinces of our country are constructing windfarms on a large scale.The proportion of installed wind power in the power system of our country is becoming more and more.After all these windfarms are beult, there will be some faults with them for many reasons and it will cause loses to wind farm. Therefor, a fault diagnosis system of wind turbine's converter is studied in this paper on the basis of xinjiang's windfarm.Xinjiang has constructed large scale windfarms earlier than other provinces, and there are many wind power system manufacturer in Xinjiang, so it is a typical example of the object about fault diagnosis.The structure of wind turbine and converter are studied, and the reasons of converter's faults are analysed in this paper, then comes the feasibility study of using NN in fault diagnosis. The software of PSIM and MATLAB are used to build wind turbine converter's model. They produce waves that contain informations about fault style and it's location.After abstracting data from these waves and normalizing, they can be used to train NN.Perceptron,BP and SOM are used respectively in fault diagnosis.From the studies of this paper,it can be known that perceptron can not fulfill the need of wind turbin converter fault diagnosis,BP can fulfill it's need but isn't flexible enough,and SOM can enlarge training data at any time and it does not need fault style's definition before training. SOM also can save time and has good precision so it best fulfill windfarm's construction condition.This paper combine traditional fault diagnosis technology of power electronics' device with the fault diagnosis technology of xinjiang wind turbine's converter in a creative way ,and it can have some active effect on using wind power largely.