02071nas a2200181   4500000000100000008004100001653002400042653001600066653001900082653000800101100001800109700001500127700001600142700001500158700002600173245010600199520158400305       2026                            d10aSemantic enrichment10aMEP systems10aGraph matching10aBIM1 aZhuohao Zhang1 aWeiya Chen1 aJianhua Cai1 aHanbin Luo1 aMirosław Skibniewski00aHierarchical system decomposition and semantic enrichment of BIM for lifecycle support of MEP systems3 a<p>The effective lifecycle management of modular Mechanical, Electrical, and Plumbing (MEP) systems is often compromised by a significant data gap between the digital design in Building Information Modeling (BIM) and the as-built conditions of prefabricated components. This paper specifically addresses the challenge of defining and structuring functionally independent component groups, or \textquoterightblocks,\textquoteright within complex BIM models that lack explicit system-level semantics and suffer from disorganized data. To solve this, a framework was proposed for hierarchical MEP system decomposition and semantic enrichment and implemented through a hybrid topological-geometric matching method that first identifies all structurally identical component assemblies via a topological search and then verifies these candidates by comparing their morphological signatures. Experimental validation demonstrates the method\textquoterights high accuracy and scalability, achieving an F1-score of 0.98, near-linear time complexity, and above 99\% precision in large-scale applications, as verified by domain experts. This research provides a robust, bottom-up approach to automatically populate a structured system hierarchy, bridging the critical data gap between the design model and physical, prefabricated modules. By creating semantically enriched and well-structured models, this work establishes a foundational data layer that enables future advancements in automated quantity take-off, modular maintenance, and end-of-life management, such as component reuse.</p>