In view of the problem of fault detection under the condition of doubly-fed wind turbine access to the weak AC grid at present,a fault recognition and detection algorithm under the framework of customized deep learning network was proposed.Firstly,the actual system data was used and normalized to zero-mean standard data,and then the data set was trained by customized deep learning network to form a pre-training network.At the same time,abnormal data was simulated and input to the previously pre-trained network to output prediction data.The threshold was calculated according to the prediction data and simulated abnormal data.When the absolute value of the difference between the prediction data and the simulated abnormal data exceeds the threshold,the system data was judged to be abnormal.The simulation results showed that the proposed method can solve the fault recognition problem more accurately and meet the requirements of fault detection in a weak grid containing doubly-fed wind turbines.
关键词
自定义深度学习网络/双馈风机/弱电网/故障识别
Key words
customized deep learning network/doubly-fed wind turbine/weak grid/fault recognition
维普
万方数据