New Customized Bidirectional Real-Time Pricing Mechanism for Demand Response in Predictive Home Energy Management System

The advancement of Internet of Things technologies enables precise control over home appliances, leading to a growing demand for demand response (DR) in residential areas. The electricity price mechanism has been an important means to implement residential DR, dynamically adjusting power demand to alleviate power mismatch in response to time-varying prices. However, despite various studies on pricing mechanisms, the unidirectional nature of these electricity pricing mechanisms greatly diminishes the willingness of end consumers to actively engage in housing DR as their electricity consumption behavior is not effectively reflected into their electricity pricing. To overcome these challenges, this study proposes a predictive home energy management system (PHEMS) by developing a new customized bidirectional real-time pricing (RTP) mechanism-based DR strategy along with an efficient price forecasting model. The proposed PHEMS first develops a new customized bidirectional RTP mechanism that enables the participation of end users in formulating hourly RTPs based on their hourly shifted power and household flexible appliance for strong DR participation. Second, a new deep-learning-based forecasting model, namely, unshared convolutional neural network-nested long short-term memory, is employed to address both the spatial–temporal variabilities of the real-time (RT) price and support global RT optimization. A rolling-horizon-based optimization strategy is then constructed to enable self-correction, ensuring robust and economic operation for residential end users. The experimental results demonstrate the superiority of the proposed PHEMS in terms of the forecasting accuracy, peak reduction, valley filling, and electricity cost reduction while ensuring user comfort.