摘要
针对建筑工程质量管理过程中决策效率低,大量历史案例信息未有效利用的问题,提出一种深度学习支持的基于案例推理(Case-Based Reasoning,CBR)的工程质量问题防治模型,以实现质量问题防治相关知识、经验的有效重用.收集、分析相关历史案例数据,基于正则表达式与双向编码器表示-双向长短期记忆网络-条件随机场(BERT-BiLSTM-CRF)模型抽取案例特征属性信息,提高案例库构建的自动化程度.引入BERT算法改进CBR的相似性度量,通过案例检索匹配和修改,为目标案例推荐可参考的防治方案,并结合某实际案例对该防治模型进行验证.结果表明,该防治模型可以从案例库中识别相似案例,提高质量管理决策的效率和准确性.
Abstract
To address ineffective decision-making and utilization in the quality management process of construction projects in a large number of historical cases,a case-based reasoning(CBR)model supported by deep learning is proposed.Historical cases are collected and analyzed.Their feature attributes are extracted based on the regular expression and bidirectional encoder representations from the transformers-bidirectional long short-term memory-conditional random field(BERT-BiLSTM-CRF)model.This model can achieve a more automatic construction of the case database.The BERT algorithm is introduced to improve the similarity measure of CBR.A prevention and treatment approach is recommended for the target case through case retrieval matching and modification and is validated with a real-life case.The results show that this approach can recognize similar cases from the case database,enhancing the efficiency and accuracy of quality management decision-making.
NETL
NSTL
万方数据