Non-invasive Load Monitoring Based on CNN-LSTM-Attention

With the large-scale deployment of smart meters at the receiving end of power, the idea of using non-invasive load monitoring (NILM) to realize intelligent electricity consumption, energy saving and emission reduction has attracted the attention of researchers. NILM technology can be used as one of the auxiliary services provided by power grid to users, which can improve users' awareness of energy saving and improve the accuracy of load forecasting. Traditional NILM technology has some problems, such as low resolution and slow decomposition speed. This paper proposes a deep learning model of CNN-LSTM-Attention model, which introduces Attention mechanism in the traditional CNN-LSTM model. Finally, the model is evaluated by using REDD data set, which proves that the model has good performance in the decomposition speed and precision of specific electrical equipment.