Chapter Seven - Intelligent learning approaches for demand-side controller for BIPV-integrated buildings

Abstract

Buildings are the most significant contributors to energy consumption and greenhouse gas emissions worldwide. Building integrated photovoltaics (BIPV) represents an effective measure toward reducing the primary energy consumption and carbon emissions of building operations. Using efficient and advanced demand-side controllers to accurately forecast demand for heating, cooling, and lighting loads and the electricity that can be generated by BIPV is essential to improve building energy management and energy flexibility in buildings. Because of the diversity and complexity of weather conditions and the unpredictability of residential electricity consumption, the accurate matching of building energy demands and the power production capacity of BIPV has become more uncontrollable and challenging. With the rapid development of artificial intelligence (AI) techniques, a number of researchers have used various machine learning methods to predict the feasibility of BIPV power, lighting consumption, cooling and heating load demands and electricity consumption, as well as to demonstrate the effectiveness of machine learning in energy consumption control and load demand prediction. This paper reviews the current development and application status of BIPV, intelligent algorithms applicable to BIPV, and intelligent algorithms for demand-side control of building energy. Then, this paper analyzes the challenges, existing issues, and rationalized research recommendations for the current applications of intelligent learning methods for demand-side controllers for BIPV-integrated buildings. In addition, this paper statistically analyzes representative case studies of intelligent learning methods for demand-side controllers for BIPV-integrated buildings and identifies best practices for their applications. This paper can contribute to theoretical recommendations for developing efficient and advanced controller models and accurate algorithmic models for short-term building energy prediction, and consequently further promoting energy-flexible buildings.