A Novel Demand-Side Management Strategy for Residential Load with an Integrated Deep-learning Based Forecasting Engine

Changing dynamics of power utilization has led to fluctuating load demand pattern. This erratic demand behavior makes it difficult for utilities to maintain supply-demand equilibrium. Therefore, the dependency of the utilities on efficient demand-side scheduling techniques is now on the rise. In this paper, a novel demand-side management (DSM) technique designed based on Teaching-Learning based optimization (TLBO) has been suggested. Firstly, a long-short-term memory-based prediction technique has been developed to obtain accurate load forecasting. The forecasted load curve is then substituted as input to the TLBO optimization to obtain an optimized demand response. This optimized demand response is fed into the load scheduling algorithm to achieve the scheduled load demand curve. TLBO optimization, together with the load scheduling algorithm, constitutes the proposed DSM methodology. The entire methodology has been tested on a data set containing half hourly load consumption value of 35 households for a period of 1 year. The simulation is performed on the MATLAB 2018a platform.