Fault Waveform Construction Method for Detecting the Intelligent Switch based on GAN and CNN

The data source for detecting the Intelligent Switches is limited, and the data con not reflect actual fault condition, this paper proposes a fault waveform construction method for detecting the intelligent switch. By extracting the features in time and frequency domain of the fault waveform, a multi-dimensional fault feature matrix is constructed, and the fault waveform construction model is obtained through convolutional neural network(CNN) and generative adversarial network(GAN) training. In the model, the GAN model is used to construct fault waveform and the constructed waveforms are classified to various fault conditions by CNN network. Accurate fault waveform can be generated through the model. Finally, the constructed waveform is tested through experiments.