Huang, Chien-PuChien-PuHuangSHANG-HSIEN HSIEH2026-01-152026-01-152026-0414740346https://www.scopus.com/record/display.uri?eid=2-s2.0-105024981745&origin=resultslisthttps://scholars.lib.ntu.edu.tw/handle/123456789/735369Predictive maintenance (PdM) is essential for minimizing downtime, extending asset lifespan, and enabling data-driven decision-making in smart facility management. However, traditional PdM systems often face challenges related to fragmented data sources, limited semantic interoperability, and a lack of automation capabilities. This study introduces STRIDE (SemanTic Reasoning and Integration for Data-driven Engineering), a modular framework that integrates IoT sensor data, Building Information Modeling (BIM), and facility maintenance records into an ontology-based knowledge graph using Neo4j. STRIDE leverages semantic reasoning and low-code automation using Python and Power Automate to support real-time anomaly detection, task generation, and explainable workflow execution. To support intelligent maintenance automation, STRIDE is implemented as a five-layer architecture, comprising semantic ingestion, reasoning logic, workflow orchestration, visualization services, and stakeholder collaboration. Validated through a simulated HVAC scenario, STRIDE achieved an 82.5% task automation rate and a query response time of 0.74 s across 30,082 graph nodes. These results demonstrate the framework’s effectiveness in enabling scalable, transparent, and Digital Twin-ready predictive maintenance in complex facility environments.falseKnowledge graphOntology-driven integrationPredictive maintenanceSemantic automationSmart facility managementWorkflow orchestrationjournal article10.1016/j.aei.2025.1042402-s2.0-105024981745