Optimizing Construction Supply Chains in Smart Building Projects via a GIS–BIM Collaborative Platform and Machine Learning–Driven Delivery Forecasting

Abstract

Smart construction projects require a high degree of collaboration among multiple participants to ensure timely delivery of materials. However, traditional planning methods are often limited by dynamic site conditions and fragmented data. This study proposes a framework integrating geographic information system (GIS), building information modeling (BIM), and machine learning (ML), aiming to optimize end-to-end construction supply chain logistics. This platform integrates spatiotemporal data such as road networks, real-time traffic flows, 3D building models, and historical distribution records, and adopts a hybrid prediction mechanism combining random forest and long short-term memory (LSTM) networks. The cloud-based visual dashboard integrates 3D models with spatial scenes, synchronously pushing predictive alerts, allowing planners to simulate different scenarios and proactively avoid delays. A case study of a high-rise project in Shanghai (involving over 1,200 freight trips) shows that the integrated system achieved an on-time delivery rate of 88%, reduced the average delivery time by 23 minutes per trip, and reduced the prediction error from 48 minutes to 29 minutes. Participants reported a significant reduction in unused labor hours and emergency changes to orders. This achievement provides a scalable example of data-driven supply chain management in the field of smart construction. In the future, it will explore enhanced features such as reinforcement learning-based dynamic route planning and blockchain-backed smart contracts.