A Novel Fault Detection Technology for Power System Relay Protection Based on Convolutional Recurrent Neural Network
With the advancement of the construction of new power systems,the capacity of distributed new energy has significantly in-creased,and more and more power electronic devices are connected to the power grid,resulting in essential changes in fault characteris-tics during short circuits.Traditional relay protection fault detection technology is difficult to apply.To improve the accuracy of new power system protection,a novel power system relay protection fault detection technology based on convolutional recurrent neural network is proposed.Firstly,wavelet transform is used to extract the time-frequency characteristics of fault currents as input data for the network.Then,a convolutional recurrent neural network algorithm is designed that includes convolutional neural networks and bidirec-tional gated recurrent units.The convolutional neural network is used to extract features from the input fault data,while the bidirectional gated recurrent units mine dynamic sequence relationships in the features.Finally,Softmax classifier is used for fault classification and the process of fault detection is completed.The experiment shows that the fault detection rate of the proposed method can reach 99.6%in the new power system,which is 8.2%and 9.7%higher than the common convolutional neural network and feedforward neural net-work,respectively,which proves that the proposed method has higher fault detection accuracy.
new power systemsdistributed new energyrelay protectionfault detectionconvolutional recurrent neural network
万方数据
维普
