Research On Wind Turbine Fault Diagnosis Technology Based On Deep Learning Algorithm

With the continuous change of energy structure,wind energy,solar energy,ocean energy and other clean energy have attracted more and more attention.Wind power is developing rapidly in the field of new energy,and gradually favored by the energy industry.With its green,pollution-free,renewable and other characteristics,it is listed as the main development object of new energy.However,due to the working environment of wind turbines in remote areas,affected by local temperature,humidity and other uncertain factors,and suffering from rain,snow,wind,frost damage all year round,the operating environment is complex,resulting in a high failure rate of wind power generation equipment.The wind turbine system is huge,the maintenance cost is large,and it is difficult to find the early failure of parts in time.Therefore,the condition monitoring and fault diagnosis of wind turbine can reduce the failure rate and improve the economic benefits of wind power plant.To sum up,this paper puts forward the research of wind turbine fault diagnosis technology based on deep learning algorithm.The main method is to judge the operation state of wind turbine based on multi-scale wavelet decomposition model,and finally use deep confidence network algorithm to achieve fault classification.The main research work of this paper is as follows(1)The basic structure and operation principle of wind turbine are discussed.According to the operation state of wind turbine,the fault mechanism of different parts of wind turbine is explored.Combined with the research status of fault diagnosis at home and abroad,the common methods of fault diagnosis for different parts of wind power generation equipment are described.Finally,according to the characteristics of the vibration data collected by the vibration acceleration sensor of the wind turbine,the multi-scale wavelet decomposition algorithm is proposed to detect the operation status of the wind turbine,and the deep confidence network algorithm is used to realize the fault classification.(2)Wavelet de-noising algorithm is applied to de noise the vibration data of wind turbine,and the pure original vibration signal is extracted.Then,the multi-scale wavelet decomposition algorithm is used to process the vibration data after wavelet de-noising.The normal vibration signal and the vibration signal to be detected are decomposed by multi-scale wavelet respectively.The normal vibration signal diagram of the same part after decomposition is compared with the vibration signal diagram to be detected,If the deviation between the vibration signal diagram to be detected and the normal vibration signal diagram is large,and the deviation tends to increase,it is proved that the wind turbine has a fault,and the wind turbine needs to be shut down for inspection immediately,so as to avoid increasing the severity of the fault and causing greater economic losses.(3)When judging the fault of wind turbine,it is necessary to diagnose the type of fault.A perfect deep confidence network model is constructed,and the multi-scale wavelet decomposition algorithm is input to judge the vibration data with faults.Finally,the types of faults are determined according to the probability model of the generated results.It can accurately determine the location of the fault,and reduce the incidence of major faults for wind power plants.(4)Finally,the effectiveness of multi-scale wavelet decomposition algorithm and deep confidence network algorithm for fault detection and diagnosis is verified by simulation experiments.Using the above fault diagnosis model,the early weak fault vibration signal can be found in time,and the fault can be handled in time,which provides the basis for the fault diagnosis of wind turbine.