Research on Power Quality Composite Disturbance Recognition Based on Deep Learning

With the integration of renewable energy sources such as wind power generation and photovoltaic power generation, many power quality disturbance events are caused. In order to improve the recognition accuracy of power quality disturbance, a new method of PQDs recognition based on convolutional neural network and attention mechanism based bidirectional long short-term memory network is proposed. The spatial features of disturbance signals are automatically extracted by CNN, and the internal dynamic transformation rules of local features of PQDs are learned by BiLSTM hidden layer, so as to extract the time features of PQDs. The features extracted by the BiLSTM hidden layer are used as the input of the attention layer. The attention mechanism is used to distinguish the extracted features by weighting. Dig deep time correlation. The CNN-ABiLSTM model can independently find the time window of the disturbance signal, and only analyze the data sequence in the disturbance signal window to improve the performance of the algorithm. The results show that the CNNABiLSTM model has a classification accuracy of 98.23% for disturbance signals under 20dB noise. The CNN-ABiLSTM network has strong anti-noise performance and good adaptability.