Research On Fault Diagnosis Of Electrical Part Of Different Types Of Wind Turbines With Different Capacity

After the 21 st century,the wind power industry has mushroomed in China.After decades of development,the installed capacity and development speed of wind power are unprecedented in China.However,with the increase in the operating years of wind power equipment,a series of rickets such as aging equipment,frequent failures,and improved maintenance costs have brought many inconveniences to the healthy development of China’s wind power industry.Frequent failures are common problems faced by all wind power generation sites,and the high maintenance costs caused by fault repairs are not uncommon.In order to improve the operating efficiency of the fan and reduce the operation and maintenance cost,this paper uses the SCADA operating data of the wind turbine to study the fault diagnosis and prediction technology of the generator,converter and pitch system in the electric part of the wind turbine.Firstly,In order to eliminate the dimension influence between the wind turbine generator SCADA operation data and to eliminate the abnormal data under the fault operation state,the standardization method and the abnormal point elimination method are proposed.Then,based on the SCADA operation data of wind turbines as the data source,the DFIG bearing temperature prediction model of the improved gray neural network is established,and the possibility of generator failure is predicted by the development trend of temperature.Finally,using MATLAB to simulate the phase-to-phase short-circuit fault on the stator-rotor side of the doubly-fed generator,the short-circuit current between the phases of the doubly-fed generator in the wind turbine and the phases under different short-circuit fault types are obtained.The relationship between the short circuit voltage.The feasibility of using the current and voltage as training samples is determined.The actual current and voltage are used as training samples for BP neural network.the classification model of phase-to-phase short-circuit fault of wind turbine generator-based doubly-fed generator based on BP neural network is obtained.Taking the SCADA operation data of wind turbine as the training sample,a converter IGBT fault classification model for decision tree is established.In order to make the classification fault classification model of the decision tree higher in classification accuracy and simplify the classification rules,a C4.5 algorithm different from ID3 is selected.At the same time,in order to improve the processing speed of the decision tree to the IGBT fault classification of the converter,the decision tree is pruned,and in the case analysis,a better classification effect is achieved.In order to make up for the shortcomings of the traditional pitch system monitoring model,the fault monitoring model of the variable-function nuclear main wind turbine generator pitch system with variable wind conditions is established.Among them,the division of wind speed conditions mainly refers to the idea of dynamic transition to the micro-element,which greatly improves the fault monitoring accuracy of the pitch system and reduces the possibility of false warning.