Non-intrusive load disaggregation of smart home appliances using the IPPO algorithm and FHM model

At present, producing the lowest power requires prognosticating the load operations in the home automation systems. Several electrical household devices with various performances work in smart homes. These variances are defined as the positions of the electrical devices. In the current study, a practical mixed method is suggested for load disaggregation of the electrical devices. A highly precise model of the Factorial Hidden Markov Model (FHMM) is applied for electrical device positions simulating. In the proposed method, the present position of every electrical device is obtained and available in the dataset. Then, the distinct suitable positions for the following instant are supplied. To obtain the best approximation of positions, it is reported to the Improved Prey?Predator Optimization Algorithm (IPPOA). Additionally, three restraints are applied in IPPOA for better adjusting the position matrix. The first one is that every electrical device must possess one position at any stage. Second, regarding the continuous active electrical household devices is necessary. Third, the FHMM should be implemented for load model generation. In terms of the third restraint, the assessed databanks and the computation period are notably decreased owing to the utilization of the FHMM. The quickness and precision of the answers for useful data of six smart homes are evaluated against other techniques to confirm the efficiency of the proposed technique.

Automated summary

The study proposes a practical mixed method for load disaggregation of electrical devices in smart homes, utilizing the highly precise Factorial Hidden Markov Model. The Improved Prey-Predator Optimization Algorithm is applied, along with three constraints for better adjustment of the position matrix, resulting in decreased evaluation databanks and computing periods. The efficiency of the proposed technique is confirmed through the evaluation of speed and accuracy of useful data from six smart homes against other methods.