Multi-objective non-intrusive load disaggregation based on appliances characteristics in smart homes
Wen Fan, Qing Liu, Ali Ahmadpour, Saeed Gholami Farkoush
Abstract
The Load Disaggregation (LD) is an optimizing problem. The actual operation states of the appliances would not serve as an optimal solution for a single-objective function, due to various noises as well as frequency interferences from the adjacent systems. In this paper, the LD model with a multi-objective function combines the appliance features at both macro and micro levels. This model contributes a good representation of the appliances from several viewpoints. Recognizing numerous appliances is carried out through five objective functions using apparent, active, and reactive powers, currents, and harmonics as the loads characteristic. The suggested problem is solved utilizing Moth-Flame Optimization (MFO) algorithm with several objectives for LD. Besides, it prevents tuning the weighted parameters and does not ignore the conflict among the objectives. In addition, the Factorial Hidden Markov Model (FHMM) is used to define the allowable modes of the appliances for the next second. It could be resorted to an objective–rank project to cope with the restraint on the number of appliances functioning concurrently. The efficiency of the suggested method for LD is shown by experimental outcomes and is compared with other methods. The results of various combinations of appliances features are evaluated by various evaluation metrics. It is presented that in more features, the results are more accurate. The results show that the accuracy of the proposed method is at least 20 % more than others.