Multi-Objective optimization for design of an Agrophotovoltaic system under Non-Dominated sorting Genetic algorithm II

Abstract

Agrophotovoltaic (APV) systems, which were first proposed in 1982 by Goetzberger and Zastrow, can simultaneously produce both photovoltaic solar energy and crops. However, their structure, whereby photovoltaic (PV) panels are located above the crops, could negatively affect crop growth. Therefore, engineers need to identify an efficient structure for an APV system before beginning any construction project involving an APV. The goal of this study is to propose a multi-objective optimization approach with a fuzzy inference system to identify the optimal design of the APV system in terms of crop production and solar power generation. To this end, a NonDominated Sorting Genetic Algorithm-II (NSGA-II) is adopted with three objectives: (1) Maximization of solar energy production, (2) Maximization of crop production, and (3) Minimization of the construction cost of an APV system. In the proposed approach, linear regression and polynomial regression are used to estimate the electricity generation quantity, the yields of five crops (sesame, mung bean, red bean, corn, and soybean), and the overall construction cost. Field study data collected from the Jeollanam-do Agricultural Services and Extension Services in South Korea are used to develop the estimation models. After the Pareto Front is constructed via NSGA-II, the devised fuzzy inference system selects the best solution regarding the impact of the APV system on the farmers' economy. The proposed approach suggests that the final optimum design of the APV system according to the fuzzy inference system is seen in the cultivation of red beans with a shading ratio of 27.83% underneath the bifacial PV panel type. As a result, the proposed approach allows for the simultaneous consideration of multiple aspects, so it is expected to provide practical and sustainable solutions to farmers and engineers by supporting APV system design.