With the focus of smart construction project management, this paper presents a closed-loop digital twin framework under the integration of Building Information Modeling (BIM), Internet of Things (IoT), and data mining (DM) techniques. To be specific, IoT connects the physical and cyber world to capture real-time data for modeling and analyzing, and data mining methods incorporated in the virtual model aim to discover hidden knowledge in collected data. The proposed digital twin has been verified in a practical BIM-based project. Based on large inspection data from IoT devices, the 4D visualization and task-centered or worker-centered process model are built as the virtual model to simulate both the task execution and worker cooperation. Then, the high-fidelity virtual model is investigated by process mining and time series analysis. Results show that possible bottlenecks in the current process can be foreseen using the fuzzy miner, while the number of finished tasks in the next phase can be predicted by the multivariate autoregressive integrated moving average (ARIMAX) model. Consequently, tactic decision-making can realize to not only prevent possible failure in advance, but also arrange work and staffing reasonably to make the process adapt to changeable conditions. In short, the significance of this paper is to build a data-driven digital twin framework integrating with BIM, IoT, and data mining for advanced project management, which can facilitate data communication and exploration to better understand, predict, and optimize the physical construction operations. In future works, more complex cases with multiple data streams will be used to test the developed framework, and more detailed interpretations with the actual observations of construction activities will be given.

本文以智能建筑项目管理为重点，提出了一种在建筑信息模型（BIM），物联网（IoT）和数据挖掘（DM）技术集成下的闭环数字孪生框架。具体地说，物联网将物理世界和网络世界连接起来，以捕获实时数据以进行建模和分析，虚拟模型中包含的数据挖掘方法旨在发现收集到的数据中的隐藏知识。拟议中的数字双胞胎已经在基于BIM的实际项目中得到了验证。基于来自IoT设备的大量检查数据，将4D可视化和以任务为中心或以工作人员为中心的过程模型构建为虚拟模型，以模拟任务执行和工作人员协作。然后，通过过程挖掘和时间序列分析研究高保真虚拟模型。结果表明，可以使用模糊矿机预测当前过程中可能出现的瓶颈，而下一阶段已完成任务的数量可以通过多元自回归综合移动平均（ARIMAX）模型进行预测。因此，战术决策不仅可以提前防止可能发生的故障，而且可以合理安排工作和人员，以使流程适应多变的条件。简而言之，本文的意义在于构建一个与BIM，IoT和数据挖掘相集成的数据驱动的数字孪生框架，以进行高级项目管理，这可以促进数据通信和探索，从而更好地理解，预测和优化物理结构。操作。在以后的工作中，将使用具有多个数据流的更复杂的案例来测试开发的框架，