Modern development of an Adaptive Non-Intrusive Appliance Load Monitoring system in electricity energy conservation

Traditionally, electricity energy consumption of a residence can be monitored through the watt-hour meter installed at the panel. However, the power consumption of individual appliances cannot be identified directly. To track and control the energy consumption of each appliance, many meters are needed. This approach results in high financial costs. To reduce the installation costs, the Non-Intrusive Appliance Load Monitoring (NIALM) approach was proposed recently. In this paper, a novel Adaptive Non-Intrusive Appliance Load Monitoring (ANIALM) system that integrates appliance energizing and de-energizing transient feature extraction methods with soft-computing techniques is developed to keep track of the energy consumption of each appliance. The energizing and de-energizing transient responses of appliances can be captured through the analyses of ANIALM. Two recognizers, k-Nearest Neighbor Rule (k-NNR) and Back-Propagation Artificial Neural Network (BP-ANN), are used to identify different types of appliances and their operation status under different single-load and multiple-load operation scenarios. The Artificial Immune Algorithm (AIA) with the Fisher criterion is employed to adaptively adjust the feature parameters in order to improve the identification performance of recognizers when a new type of appliance is added for monitoring. From the experimental results obtained in different actual environments, the proposed ANIALM system is confirmed that it is able to identify the operation status of appliances. Also, although the generalization of both recognizers is similar and excellent, the k-NNR recognizer used by the ANIALM system is preferred from the aspects of recognizers’ identification mechanism and training performance due to its simplicity in computation and implementation.