Optimization configuration of flexible interconnection device DG based on improved E-C-Kmeans and GA under SOP support

•A hybrid optimization model integrating improved E-C-Kmeans clustering and path-tracing GA was proposed for DG allocation under SOP support.•The E-C-Kmeans algorithm introduces entropy-based feature weighting and density-aware mechanisms, improving clustering robustness for volatile wind–solar data...

Full description

Saved in:
Bibliographic Details
Published in:Systems and soft computing Vol. 7; p. 200414
Main Authors: Yang, Zhichun, Yang, Lei, He, Junxiong, Yang, Fan, Min, Huaidong, Zhang, Nan, Su, Jing
Format: Journal Article
Language:English
Published: Elsevier B.V 01.12.2025
Elsevier
Subjects:
Distributed power supply
Feature selection weighted k-means clustering algorithm
Flexible interconnection devices
Genetic algorithm
Optimization configuration
Distributed power supply
Feature selection weighted k-means clustering algorithm
Genetic algorithm
Flexible interconnection devices
Optimization configuration
ISSN:2772-9419, 2772-9419
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•A hybrid optimization model integrating improved E-C-Kmeans clustering and path-tracing GA was proposed for DG allocation under SOP support.•The E-C-Kmeans algorithm introduces entropy-based feature weighting and density-aware mechanisms, improving clustering robustness for volatile wind–solar data.•A multi-layer benefit coordination framework is established, balancing decision, scheduling, and execution objectives in distributed power systems.•Simulation results show the proposed model achieves higher resource utilization (87.7 %), fault recovery rate (82.6 %), and optimal individual count (183).•The model demonstrates strong adaptability under fluctuating load and weather conditions, providing stable and practical solutions for DG planning. The increasing popularity of flexible interconnected devices has made them a crucial component of distributed power systems. At present, the effective integration and optimized configuration of distributed power sources remain a key focus in power system research. To address this, the study introduces improvements to both the feature selection weighted K-means clustering algorithm and the genetic algorithm, applying them to data preprocessing and configuration optimization. Based on these, an integrated optimization configuration model is proposed. The experimental results demonstrated that the improved clustering algorithm achieved the highest accuracy of 96 %. Under varying wind speeds and light intensities, its minimum sum of squared clustering errors reached 8 and 10, respectively. The improved genetic algorithm generated up to 190 optimal individuals, with a highest accuracy of 91.2 % and a minimum computation time of 6.2 s. Simulation testing further showed that, under support from back-to-back voltage source converters, the decision-making layer achieved a maximum comprehensive income of 1.547 million yuan, an environmental impact of only 41.7 %. The scheduling layer reached the lowest operational cost of 1.627 million yuan. In conclusion, the proposed approach demonstrates significant improvements in distributed power utilization and system operating efficiency.
ISSN:2772-9419
2772-9419
DOI:10.1016/j.sasc.2025.200414