返工是阻碍装配式建造提质增效的重要因素之一, 但目前在项目实施过程中对返工风险的识别与管理主要依靠工程师的经验判定, 难以形成系统、高效的知识传递和增值。该文基于精益管理思想, 提出了装配式建造返工风险的转换-流动-价值(transformation-flow-value, TFV)理论模型和装配式建造信息流动模型, 构建了装配式建造过程返工风险智能识别本体模型。为实现完整的装配式建造返工风险识别过程, 该文建立了包含知识管理、信息处理和推理查询3个模块的总体设计框架。通过实例数据映射, 实现非结构化数据的语义化表达并将其储存在本体知识库中; 建立推理检索规则, 实现返工风险智能识别和风险信息的智能输出。以一个叠合板为例阐述了返工风险智能识别过程。研究结果表明, 精益管理和语义网技术的结合对推动装配式返工风险的信息化管理具有重要意义。

Objective: Prefabricated construction features advantages such as improving efficiency, shortening the construction period, and reducing costs. However, the factors represented by the rework process hinder its further development. Rework risk management in an actual project relies mainly on expert experience, making it difficult to efficiently and accurately transfer knowledge and add value. This paper establishes an intelligent ontology for rework risk identification by introducing lean management thinking into rework risk management for the prefabricated construction process. The goals are to identify potential rework risks in advance, block the rework flow between construction processes, and minimize non-value-added activities. Methods: This paper examines the workflow for identifying rework risks in the prefabricated construction process based on the evolutionary transformation-flow-value (TFV) theoretical model and the information flow model. Integrating the requirements and constraints of the design criteria, production, and construction, the models can demonstrate how an owner's expectations flow between lean design, lean supply, and lean assembly and how to query and share rework risk expertise to drive a project forward. Based on lean management ideas, this paper further constructs an intelligent ontological model for rework risk identification in the prefabricated construction process. To realize the advanced rework risk identification in the prefabricated construction process, this paper establishes an overall design framework including three modules: knowledge management, information processing, and reasoning queries. Building information is stored in a building information modeling (BIM) framework. Rework-risk-related knowledge is usually presented or stored in the form of expert experience, engineering practice, documented information, and project documentation. Using individual data mapping, these unstructured data are semantically expressed and stored in the ontological knowledge base as risk precursors. Based on inference and retrieval algorithms, the specifications related to prefabricated construction are written into the ontological knowledge base as risk determination conditions to achieve intelligent rework risk identification and risk information output for individuals in the prefabricated construction process. A laminated slab example is used to illustrate the rework risk identification process. Results: The process of identifying the rework risk for a laminated slab example demonstrates the following results: The semantic expression of heterogeneous data such as the BIM model, Internet of Things monitoring data, textual information, and document records provides a means for collecting precursory information regarding risk objects. By using relevant algorithms, project participants can eliminate cumbersome data burdens and capture potential product rework risk and optimization measures more quickly and accurately, which ensures that the rework risks in each process will not be overlooked or misunderstood, as well as eliminate the value outflow to maximize the owner's success. Conclusions: By building an ontological model for rework risk identification based on lean management, the flow of rework between processes may be blocked to some extent, thus reducing the waste caused by non-value-added behaviors. This paper's process of identifying rework risk combines lean management thinking and semantic web technology, which has significance in promoting the knowledge management of rework risks in prefabricated construction.