基于无线传感与自适应稀疏时频分析的桥梁拉索时变索力识别方法

For a cable-supported bridge, the cable is the critical stress-carrying component. The cable tension forces is time-varying subjected to the moving vehicle loads and environmental effects, causing cumulative fatigue damage of a cable. The cumulative fatigue damage of cable is a great threat to the safety of bridge structures. Aiming to address the key issues of identifying time-varying cable tension forces, this proposal is to develop new approaches for identification of time-varying cable tension forces of bridges based on wireless sensor and adaptive sparse time-frequency analysis. First, the identification method of time-varying bridge cable tension forces with complete wireless sensing vibration data using matching pursuit algorithm-based adaptive sparse time-frequency analysis is proposed. Furthermore, the identification method of time-varying bridge cable tension forces with incomplete wireless sensing vibration data using l1 sparse optimization-based adaptive sparse time-frequency analysis is developed. Third, the adaptive group sparse time-frequency analysis is proposed to identify time-varying cable tension forces of multiple cables. Finally, the model experiments and field tests on a real bridge are to be carried out to validate the ability of the proposed approaches. This proposal is anticipated to develop new approaches with important scientific significance and practical value for SHM of bridges in the identification of time-varying cable tension forces of bridges with high resolution and accuracy, which will lay the foundation for safety evaluation and fatigue cumulative damage evaluation of bridge cables.

拉索是缆索类桥梁的关键受力构件，在车辆荷载和环境因素作用下，拉索索力是时变的，时变索力是引起拉索疲劳损伤的主要原因，拉索的累积疲劳损伤将严重威胁桥梁结构安全。面向桥梁拉索时变索力识别的难点科学问题，本项目利用智能无线传感技术、结合信号时频分析领域最新进展，研究基于自适应稀疏时频分析的桥梁拉索时变索力识别方法。首先，针对拉索振动无线传感监测数据，研究数据完备时基于匹配追踪算法的自适应稀疏时频分析时变索力识别方法；其次，研究数据不完备时基于l1范数稀疏优化的自适应稀疏时频分析时变索力识别方法；然后，研究基于自适应群稀疏时频分析的相邻拉索时变索力同步识别方法；最后，通过斜拉桥模型实验和金塘大桥现场测试验证方法的有效性。本项目将实现桥梁拉索时变索力高分辨率和高精度识别，为拉索极限状态安全评定和疲劳累积损伤评价奠定基础，丰富和发展结构健康监测数据分析与挖掘理论，具有重要的科学意义和实用价值。

本项目研究基于无线传感与自适应稀疏时频分析的桥梁拉索时变索力识别方法，考虑到实际监测数据的复杂性，提出了基于多信息卷积神经网络结构的健康监测异常数据诊断方法；在此基础上，建立了数据完备时基于匹配追踪算法的自适应稀疏时频分析时变索力识别方法，研究了数据不完备时自适应稀疏时频分析时变索力识别方法鲁棒性；提出了基于机器学习自适应稀疏时频分析的时变索力识别方法。并进一步并研究了基于结构响应稀疏性的结构模态参数识别的机器学习方法。项目累计发表论文19篇，其中SCI论文15篇，国际会议特邀报告1次，软件著作权4项，申请发明专利3项。参编中国工程建设协会标准《大跨度桥梁结构健康监测系统预警阈值标准》（T/CECS 529-2018），参编交通部《公路桥梁结构安全监测系统技术规程》（JT/T 1037-2016）外文版标准。研究成果获国家科技进步二等奖一项（排名5）。培养硕士研究生4人（已毕业），指导在读博士研究生2人，获研究生国家奖学金2人，博士生1人获CSC资助在美国普渡大学交流访问1年。参加与项目相关的国际和国内会议13人次，邀请国外学者讲学1人次，主办了首届国际结构健康监测竞赛（IPC-SHM 2020）。