Work

A novel differential evolution method with a hierarchical decoder for the photovoltaic layout optimization problem

Year: 2024

Type: article

Abstract: As photovoltaic technology advances, the scale of systems expands. A typical photovoltaic power plant spans over 1 km 2 and includes more than 100,000 modules and 300 km of cabling, leading to substantial design complexities. Manual design processes can take weeks and often fail to guarantee cost-optimal solutions. Consequently, an efficient and optimal layout design method for photovoltaic plants is crucial, as it enhances design efficiency while optimizing overall costs. This design method involves the precise placement of equipment and strategic configuration of cable connections. Challenges in this process arise from the need to accommodate various site shapes, equipment sizes, topological connections, and design constraints specific to power stations. To address these challenges, we develop a differential-evolution-based photovoltaic layout optimization method. Firstly, we propose a hierarchical decoder which includes a grided layout process coupled with a clustering method to zone the site and position equipment, followed by the application of the A* algorithm to determine the efficient cable paths. This process is further refined using the differential evolution algorithm, tailored to meet the constraints of photovoltaic plant design while enhancing the solution search scope. Results from benchmark cases demonstrate that the proposed method is effective across small, medium, and large cases, outperforming the most advanced methods in 38 out of 45 cases. The effectiveness of the zoning, pathfinding, and differential evolution mechanisms within the method has been verified through ablation studies. Moreover, in real power plant cases, this method has reduced overall costs by 44% compared to those of manual designs. • A model of the photovoltaic layout optimization problem has been established. • A hierarchical approach for equipment layout and cable routing has been developed. • Differential evolution is employed for constraints compliance and optimal solution. • A new benchmark cases across small, medium and large scales have been generated.

Source: International Journal of Electrical Power & Energy Systems

DOI: 10.1016/j.ijepes.2024.110397

Authors: Yuanqing Yao, Yibo Wang, Hongjie Jia, Yunfei Mu

Institutions: Tianjin University, Chinese Academy of Sciences, Institute of Electrical Engineering

Language: English

Topic: Metaheuristic Optimization Algorithms Research

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

SDGs: Industry, innovation and infrastructure, Affordable and clean energy

Open Access status: gold

Estimated APC paid: $3,500