Toward unique electrical ladder network model synthesis of a transformer winding high-frequency modeling using K-means and metaheuristic-based method

• Published in: Compel Vol. 43; no. 1; pp. 247 - 266
• Main Authors: Chanane, Abdallah, Houassine, Hamza
• Format: Journal Article
• Language: English
• Published: Bradford Emerald Publishing Limited 01.03.2024; Emerald Group Publishing Limited
• Subjects: Accuracy; Algorithms; Clusters; Empirical analysis; Fault diagnosis; Frequency response; Heuristic methods; Mathematical models; Methodology; Optimization; Optimization algorithms; Parameter estimation; Parameter identification; Penalty function; Resonant frequencies; Transformers; Uniqueness; Winding; K-means algorithm; Frequency response analysis; Electrical ladder network synthesis; Flexible constraint coyote optimization algorithm; Diagnosis; Metaheuristic algorithm; Transformer winding
• ISSN: 0332-1649; 2054-5606; 0332-1649
• DOI: 10.1108/COMPEL-05-2023-0207

Purpose Although, numerous optimization algorithms have been devoted to construct an electrical ladder network model (ELNM), they suffer from some frail points such as insufficient accuracy as well as the majority of them are unconstrained, which result in optimal solutions that violate certain security operational constraints. For this purpose, this paper aims to propose a flexible-constraint coyote optimization algorithm; the novelty lies in these points: penalty function is introduced in the objective function to discard any unfeasible solution, an advanced constraint handling technique and empirical relationship between the physical estimated parameters and their natural frequencies. Design/methodology/approach Frequency response analysis (FRA) is very significant for transformer winding diagnosis. Interpreting results of a transformer winding FRA is quite challenging. This paper proposes a new methodology to synthesize a nearly unique ELNM physically and electrically coupled for power transformer winding, basing on K-means and metaheuristic algorithm. To this end, the K-means method is used to cluster the setting of control variables, including the self-mutual inductances/capacitances, and the resistances parameters. Afterward, metaheuristic algorithm is applied to determine the cluster centers with high precision and efficiency. Findings FRA is performed on a power transformer winding model. Basing on the proposed methodology, the prior knowledge in selecting the initial guess and search space is avoided and the global solution is ensured. The performance of the abovementioned methodology is compared using evaluation expressions to verify its feasibility and accuracy. Originality/value The proposed method could be generalized for diagnosis of faults in power transformer winding.