Design And Implementation Of Wind Turbine Condition Monitoring And Fault Diagnosis System

As a clean and green energy source, wind power is receiving more and more attention innowadays society. The economic efficiency of wind power generation is the major concern ofindex and factors in wind farms. During the operation of wind farm, the cost in dailymaintenance and fault repair accounts for large proposition of generation company’sexpenditure. Wind power generation unit is a complex plant which is consisted of mechanicand electric devices. A malfunction in some part of unit would directly affect the normaloperation of whole system. The uncertainty in maintenance could cause huge waste in humanand material resources. The condition monitor and fault diagnosis system has been applied innew type of wind turbine, which is still unavailable for those early produced units which mayface the frequent malfunction.This dissertation designs a system which incorporates the functions of data acquisition,wireless transmission and fault prediction for condition monitor and fault diagnosis in highlyrisked parts of wind turbine. The designed system can rapidly and precisely locatemalfunction parts, and predict the tendency of working condition of inside parts for a periodof future time. This system samples the vibration signal of gear box and current signal ofgenerator by using the Piezoelectric Acceleration Transducer and Hall current sensor. Thesampled signals have been amplified by charge amplifier, and transferred to an industrycomputer by data acquisition card. A data base is built in industry computer for data readingand storing. The method of multi-antenna technology and enhancement of transmission andreceiving gain is used to establish wide-area Wi-Fi network, which enables the wirelesscommunication between wind turbine and Master System. The acquired data in MasterSystem is analyzed and diagnosed by applying the Wavelet Package Analysis andFouriertransform, and extracted the hidden fault feature vectors. Finally, the location and degree offault parts are acquired by using BP neural network for feature vectors analysis. Meanwhile,BP neural network can predict the future operating status of devices in wind turbine for themaintenance and repair plan of wind firm, which improves the pertinence and effectiveness ofwind firm operation.This system has been installed and tested in Honghaiwan wind firm. The operationresults demonstrate the designed system can effectively monitor the current working conditionof wind generation unit, and precisely diagnose the common faulty of generator and gear box.