A novel feature extraction method for the development of nonintrusive load monitoring system based on BP-ANN

The novel feature extraction method for nonintrusive load monitoring (NILM) systems was proposed in this paper. In order to monitoring the status of each load, a sensor is installed for each load traditionally. The status of the load is transmitted to the main controller such that the status of each load can be monitored. On the other hand, the NILM is able to detect the status of loads by analyzing the current signals that is picked-up by a current sensor installed at the main electrical panel. Traditional NILM methods used real power, reactive power, harmonic contents of power signatures and transient energy to determine the status of appliances. These methods are very complex and require a lot of computation. In this paper, a novel method that integrates artificial intelligent recognition technique and load current acquisition method for NILM is proposed. The proposed method uses timedomain information. This approach is different from traditional NILM methods. The proposed method is able to detect the energization and de-energization of loads by applying back-propagation neural networks (BP-ANNs). The overall correct rate for this method is above 98.75%. This result shows that the proposed method is able to determine the operation status of loads with proper robustness.