Towards efficient human-machine interaction for home energy management with seasonal scheduling using deep fuzzy neural optimizer

Abstract

Maintaining the records of domestic consumers' electricity consumption patterns is very complex task for the utilities, especially for extracting the meaningful information to maintain their demand and supply. Due to the increase in population, large amount of valuable data from the domestic sector is extracted by the smart meters and it becomes a vulnerable issue to tackle this information in recent era. In this work, we have proposed the fuzzy deep neural optimizer to optimize the cost and power demand of the stochastic behavior of the domestic consumers. For optimization process, this optimizer considers three control parameters: energy consumption, time of the day, and price and two performance parameters: cost and peak reduction. The dataset used for this optimization process is of two seasons: summer and winter season and it is obtained from Pecan Street Incorporation site. Takagi Sugeno fuzzy inference system is applied for the computation of the rules, which are formulated using the Membership Functions (MFs) of the aforementioned parameters. The nature of the MFs is chosen as Gaussian MFs to continuously monitoring the consumers' behaviors at different time intervals. Simulations are performed to show the robustness of the proposed optimizer in terms of energy efficiency and cost optimization up to 8 kWh and 1$ for the summer season and 12.5 kWh and 4$ for winter season. The proposed optimizer outperforms the previous scheme with remarkable results and highly recommended for the future systems where consumers are growing tremendously.