Optimizing Energy Consumption in Smart Buildings through Reinforcement Learning-Based Demand Response Strategies

This research studies the feasibility of RL-based response strategies that can maximize energy consumption in a smart building. By the inclusion of recreational pondering and observational investigation, RL operators automatically calibrated HVAC setpoints lighting levels level machine usage to reduce energy costs without compromising human comfort. It is thus evident from the results that significant value energy used on savings, RL-based algorithms achieved an average cost reduction of 16% compared to pattern based techniques. And the reasonable limits of tenant consolation hold standard fluctuation in temperature at 0.4 °C and variances from normal lighting level up to ten percent are also allowed. The occurrence of imperative violation is also reduced as RL-based approaches exhibit a lower number in comparison to rule based control and passive planning methods.