Carbon module division of substation based on feature extraction and simplification algorithm
YANG Fan
CHEN Fulei
FU Anyuan
TANG Yue
WANG Yongli
Abstract：This study addresses key challenges in the classification of carbon emission modules in substations,including incomplete feature representation,insufficient handling of non-linear data,and weak integration with low-carbon design strategies.To this end,an intelligent partitioning framework is proposed that integrates Life Cycle Assessment-based carbon accounting,deep feature encoding,and an improved density-based clustering algorithm.A five-dimensional feature encoding system is constructed,incorporating process type,material properties,spatial attribution,carbon emission intensity,and engineering functionality.An enhanced fast density peak clustering algorithm is employed,incorporating a dual calibration mechanism based on silhouette coefficient and Davies-Bouldin index.This method effectively addresses the joint optimization challenge of dimensionality reduction in high-dimensional features and the preservation of physical engineering semantics.Case studies demonstrate that,compared with traditional rule-based partitioning approaches,the proposed method significantly improves clustering quality in terms of both compactness and separability,and enables accurate identification of core carbon-intensive modules such as control cable laying,cable trays,and 220 kV main transformers.The approach provides methodological support for refined carbon footprint management and low-carbon design optimization in substations and offers scalability to broader applications in next-generation power systems.
Keywords：substationfeature extractionDPCcarbon module division
Publication Date：2025-11-28
Online Publishing Date：2025-12-18(First online date of this platform, not the publication date of the document)
Pages：7( 96-102 )