Optimal control of renewable energy in buildings using the machine learning method

Abstract

In this paper, the optimal control of renewable energy in the green building using a powerful reinforcement learning control method is presented to minimize energy consumption and power losses in the distribution sector. Solar energy for heating, solar photovoltaic cells, and biomass boilers is considered the main renewable energy source. The control system is determined based on the energy sources usage of the building and what is stored for later use or injected into the environment to satisfy the requirement of renewable energy along with minimum use of other energies. The optimization is represented in the central management system in the building with a novel system which is called here Building Energy Enhancement Learning (BEEL). Optimized BEEL is maximized the numerical reward of the signal by the represented reinforcement learning algorithm and compare with a hybrid supporting vector-wavelet learning algorithm. The new method reveals the corresponding actions are based on the reward and have higher compliance than the other represented machine learning methods. Two unique features of reinforcement learning compared to artificial intelligence methods are trial and error and delay reward, which is done here with 99.98% accuracy.