A Semi-supervised Method of Non-intrusive Load Monitoring with Answer-Question Technique

摘要

Nowadays, there is not an easy-to-implement system to help residential customers in monitoring and displaying electricity consumption because such a system needs installation of measuring instruments at every interested electric appliance. This difficulty can be overcome by one point or Non-Intrusive Load Monitoring (NILM). This paper presents a semi-supervised method of NILM with the answer-question technique. The proposed method uses selected machine learning techniques for clustering electricity consumption of various electric appliances. In particular, an improved K-Means technique with optimal number of cluster determination and feature selection methods is presented. Moreover, an answer-question technique is employed to improve clustering results. Subsequently, some experimental results are given to show effectiveness and accuracy of the proposed method. Consequently, the proposed method can help in uncovering details of electricity, meanwhile reducing cost of installing measuring instruments.