一种用于比较光伏跟踪系统的折衷解决方案：基于7E和不确定性分析并辅以机器学习算法

Abstract As the photovoltaic industry continues to expand, the low efficiency of fixed systems highlights the need for improved solutions, with tracking systems emerging as a viable option to enhance energy generation. Accordingly, this study proposes a novel framework based on compromise solution, in which four common solar tracking types including single-axis east–west tracking (SEWT), single-axis north–south tracking (SNST), single-axis azimuth tracking (SAZT), and dual-axis tracking (DAT) are evaluated from energy, exergy, economic, environmental, energoeconomic, exergoeconomic, and enviroeconomic along with risk assessment. To predict the uncertainty of tracking systems, the problem is solved 5000 times, and then predicted 50,000 years using the Random Forest Algorithm (R 2 ∼ 0.95), and the Monte Carlo simulation is employed to calculate the uncertainty of systems. The results showed the highest electricity generation (3.21 GWh/yr), exergy efficiency (20.5 %), environmental benefit (2016 ton/yr), exergoeconomic (72 kWh/$), and enviroeconomic (20162 $) belong to the DAT, while the lowest levelized cost of electricity (51 $/MWh) and payback period (14 years) are associated with the fixed structure (FS). Also, the probability of project failure is 18.6–34.8 %, with the lowest one related to FS. In general, the compromise solution showed that the priority of tracking systems is as follows: DAT > SAZT > SNST > FS > SEWT.