面向区域尺度太阳能光伏板改造的双层智能决策模型

Abstract Photovoltaic (PV) panel retrofit is crucial for reducing building energy consumption and addressing climate change. However, buildings vary in characteristics, making efficient decision-making on PV panel retrofit plans at regional or urban scales a complex challenge. A two-layer “bottom-up” intelligent decision-making model for regional PV panel retrofit is proposed in this study, replacing the traditional “top-down” method. For the individual building, various machine learning algorithms were compared, with the random forest (RF) algorithm achieving 93% accuracy and an AUC of 0.80, while the Euclidean distance was used for similarity calculation.; For regional building stocks, combined with tacit knowledge, the NGSAII optimization algorithm considers retrofit probability, similarity, cost, and retrofit benefits as optimization objectives, producing Pareto optimal solutions. The feasibility of the proposed model was demonstrated using 307 buildings from 40 residential communities in Beijing, resulting in 17 optimal solutions. The results were compared and analyzed against the optimization model that did not consider tacit knowledge. The constructed model was found to account for elements that are difficult to quantify, such as technical difficulty, in addition to cost-effectiveness. This research enhances the “bottom-up” decision-making framework, providing decision-makers with support to maximize investment returns and promote sustainable regional development.