A two-stage deep convolutional model for demand response energy management system in IoT-enabled smart grid

Abstract

Home Energy Management System (HEMS) plays an integral role in SG that optimally schedules the appliances to achieve energy savings, cost reduction, Peak–Average Ratio (PAR) minimization via the Demand Side Management (DSM) system. Previous studies are developed to perform electricity demand scheduling for the efficient management of electricity in all the sectors. However, effective monitoring and energy management solutions are essential for the residential loads to control the operation of the appliances depending upon the consumption of power during high peak hours. For this purpose, we proposes an optimal load scheduling technique named Two Stage Deep Dilated Multi-Kernel Convolutional network (DDMKC)-Modified Elephant Herd optimization algorithm (MEHOA) approach to manages, shifts the load of the consumer and thereby lowers the electricity bill. In this model, the forecaster scheme utilizes the DR pricing information that accurately predict the future pricing signal to make optimal decision and achieve minimum degree of discomfort. Based on the forecasted future prices, the MEHOA algorithm schedule the power consumption pattern to the appliances to solve the problem residential load management issues of the consumers and thereby enhances the user comfort (UC), alleviate PAR, and lowers the payment of electricity bill. Experimental analysis is conducted in terms of several metrics like energy consumption, user comfort, and PAR measures and compared with other scheduling based optimization models. The proposed model achieves the percentage greater of 60%, 62%, and 65% for the DAP, RTP, and ToU DR pricing schemes respectively. The result shows that the proposed model effectively and efficiently.