稀疏表示与云计算融合的桥梁结构损伤正则化识别方法研究

Under the conditions of time-varying environments and moving heavy vehicle loads, bridge structural damages will be caused, accumulated and even will induce a fatal bridge accident at last. It is necessary to immediately monitor oprational condition of bridges, identify structural damages and evaluate safety of bridges. Based on sparse representation theory, cloud-computing technique, bridge damage sparsity in some domains and difference among structural deterioration performances in the spatio-temporal multiple scales, the dynamic system model of cracked bridges will be first established under the moving heavy vehicle loads and the time-varying environmental conditions, the macro correlated relationship between the bridge structural response massive data recorded in a SHM system and the MapReduce effects of structural damage indices from bridge responses will then be determined, and some new sparse regularization identification approachs will be finally proposed for bridge structural damages by combining the sparse representation theory and colud-computing technique. It aims at sparse representation mode and regularization solutions to structural damage of cracked bridges, development of macro relationship between the MapReduce effects of nonlinearly sequential parameters and the SHM massive data of bridge structural responses, and exploring the self-organizing evoluation mechanism for structural damage of bridges, respectively. The accuracy for regularization identification of bridge structural damages and the processing efficiency for SHM massive data will be improved and some technical supports provided for bridge SHM as well.

在时变环境因素与移动重车荷载作用下桥梁结构发生损伤并逐步积累扩大，甚至诱发桥梁事故，实时监控桥梁状态、及时识别桥梁结构损伤并评估其安全性的研究工作迫在眉睫。基于稀疏表示理论和云计算技术，考虑桥梁结构损伤多域稀疏性及结构劣化行为时-空多尺度差异性，建立环境与移动重车作用下裂纹桥梁结构时-空多尺度动力模型，宏观映射(Map)归约(Reduce)桥梁SHM系统海量实测数据与损伤稀疏特征相关关系，提出稀疏表示与云计算融合的桥梁结构损伤正则化识别新方法。重点研究裂纹桥梁结构损伤稀疏表示模式及其正则化求解方案，宏观确立结构SHM海量数据与非线性损伤序参量稀疏特征MapReduce相关关系，并揭示环境-车-桥梁集群结构损伤序参量演化规律与结构损伤质变的自组织演变机理。旨在提高时变环境与移动重车作用下桥梁结构损伤正则化识别精度和SHM海量数据处理效率，为桥梁SHM理论形成与发展提供科学依据与技术支持。

采用理论、数值仿真与实验技术相结合的方法，基于时变环境和桥梁移动重车荷载作用下桥梁结构动力劣化行为，通过融合稀疏表示理论与云计算技术研究了桥梁结构损伤识别(SDI)问题。针对桥梁结构损伤多域稀疏性及其结构劣化行为时-空多尺度差异性，建立了时变环境与移动重车作用下裂纹桥梁时变动力系统模型，通过宏观映射(Map)归约(Reduce)桥梁结构健康监测(B-SHM)实测数据与结构稀疏损伤相关关系，提出了多种基于结构损伤多域稀疏性表示、字典学习、加权策略、传递率函数、群智能与云计算融合的桥梁结构损伤正则化识别新方法。重点研究了裂纹桥梁结构损伤稀疏表示模式及其正则化求解方案，宏观确立了B-SHM实测数据与非线性损伤序参量稀疏特征MapReduce相关关系，提出并发展了多种移动车载稀疏正则化识别方法，揭示了环境-车-桥梁集群结构损伤序参量演化规律与结构损伤质变的自组织演变机理，增强了桥梁结构损伤可识性，提高了桥梁SDI正确率与B-SHM实测数据的处理效率，为桥梁结构安全评估与B-SHM理论发展与应用提供了源头技术支撑和有益借鉴。至目前，已发表论文52篇，包括31篇期刊文章（其中，SCI收录21篇（含中科院分区表一区Top期刊4篇）及EI收录29篇）以及21篇学术会议文章（其中EI收录11篇及ISTP收录3篇）；已申请发明专利9项（其中5项已授权）；已培养4位毕业博士和8位毕业硕士，目前尚有8位研究生在读。