Backtracking search algorithm with reusing differential vectors for parameter identification of photovoltaic models. (1st November 2020)

• Record Type: Journal Article
• Authors: Zhang, Yiying; Huang, Caifeng; Jin, Zhigang
• Is Part Of: Energy conversion and management. Volume 223(2020)
• Publication Date: 2020-11-01
• Subjects: Photovoltaic model -- Backtracking search algorithm -- Reusing differential vectors -- Solar energy; Direct energy conversion -- Periodicals; Energy storage -- Periodicals; Energy transfer -- Periodicals; Énergie -- Conversion directe -- Périodiques; Direct energy conversion; Periodicals; 621.3105
• DOI: 10.1016/j.enconman.2020.113266
• Languages: English
• ISSNs: 0196-8904

Abstract

Highlights: Backtracking search algorithm with reusing differential vectors is proposed. The proposed method doesn't need any effort for fine tuning initial parameters. A reusing differential vectors mechanism is built to perform the search task. The proposed method is used for parameter identification of photovoltaic models. The proposed method outperforms the compared recently reported optimizers. Abstract: In order to simulate, control and optimize photovoltaic (PV) systems, how to accurately identify the unknown parameters of PV models is a major challenge. To overcome this challenge, this work reports a very simple but efficient optimization method called backtracking search algorithm with reusing differential vectors (BSARDVs). BSARDVs has a very simple structure and only needs the essential population size and stopping criterion for optimization. Mutation operator is employed to generate new individuals in the search process of backtracking search algorithm (BSA), which guides the search direction of population by the differential vectors between history population and current population. To enhance the global search ability of BSA, BSARDVs first archives some most promising difference vectors from history population and then reuses these differential vectors for generating next generation population. The performance of BSARDVs is investigated for parameter identification of three PV models, i.e. single diode model, double diode model and PV module model. Experimental results reveal BSARDVs can find the better solution than the compared algorithms on double diode model. In addition, for single diode model and PV module model, the solutions of BSARDVs are the same solutions with those of some compared algorithms while BSARDVs consumes less function evaluations than these algorithms. This proves the effectiveness of reusing differential vectors in BSA for parameter identification of PV models.