Household Appliance Identification Based on a Novel Load Signature Processing Framework

Non-intrusive load identification is a promising candidate in providing energy consumption information of individual appliances, which can improve energy-using habits and save energy. The load signature has a huge impact on the performance of non-intrusive load identification systems. In this paper, a new household appliance load identification model based on a novel load signature processing framework is proposed. Based on the proposed framework, the original VI trajectory of the household appliance is mapped to the binary image. The LeNet trained on the MNIST dataset is utilized to extract the deep features from the binary VI image. After that, the ReliefF algorithm is adopted to select the most important information from the deep features. The SVM is utilized to identify household appliances based on the obtained load signature. Finally, the experimental results based on the measured dataset and the PLAID dataset indicate that the proposed load signature processing framework is effective in improving the household appliance identification accuracy.