基于响应扩展的大跨空间结构损伤识别理论研究

Thousands of large-span spatial structures (LSS) have been erected worldwide, of which the safety performance attracts extensive attention. Therefore, the damage identification for LSS has been always a hot research topic in recent years. Normally, a large-span spatial structure consists of hundreds of joints and members, meaning a huge size of the potential damages. The sensors are whereas limited and can supply only a few measured data, which can hardly satisfy the need of accurately identifying the damages in a real structure. In order to solve this technical bottleneck, this project systematically studies the theory for expanding the structural responses and identifying the structural damages based on the expanded structural responses. The novel methods for expanding the static and dynamic responses from a few tested freedoms to all freedoms are studied respectively, which can increase the usable testing data. The composition of the expansion errors is analyzed and the error-controlling methods are developed. For static loading, environmental excitation and step excitation, the corresponding identification strategies are developed, which utilize the expanded static responses, the expanded dynamic responses and the combination of the two kinds of expanded responses. By numerical analysis and model testing, the effectiveness of the response expansion methods and damage identification strategies are validated. The influence of the main factors, such as loading patterns, sensor layout, expansion errors, the sampling time and frequency of the dynamic responses, is also investigated. The study results can avoid identifying the closely space modes, improve both the structural testing efficiency and the damage identification accuracy, and establish the basis for the engineering practice of the damage identification of LSS.

目前国内外有数以万计的大跨空间结构正在服役，它们的安全性能备受关注，准确识别此类结构的损伤非常重要。由于大跨空间结构的节点和杆件数量众多，识别规模非常庞大，因此有限测点提供的实测信息不能满足此类结构损伤识别的需要。为解决该技术瓶颈，本项目系统研究大跨空间结构的响应扩展和基于扩展后响应的损伤识别理论。分别研究静、动力响应由少数测点扩展至全自由度的方法，分析其误差构成机理，发展相应的误差控制方法，以增加高质量的可用测试信息。引入扩展后的静、动力响应，针对静力加载、环境激励和阶跃激励，分别发展基于静、动力响应扩展和两种响应扩展相结合的损伤识别策略。通过数值模拟和模型试验，验证响应扩展方法和损伤识别策略的有效性，重点考察加载模式、测点布置、扩展误差及采样时长等因素对损伤识别规模与精度的影响。研究成果将避免识别密集模态、提高测试效率、改善损伤识别效果，为大跨空间结构损伤识别的工程应用奠定理论基础。

由于大跨空间结构的节点和杆件数量众多，识别规模非常庞大，因此有限测点提供的实测信息不能满足此类结构损伤识别的需要。为了增加可用测试信息，本项目研究了大跨空间结构由阶跃激励先后产生的静、动力响应扩展理论，并发展了基于静、动力响应扩展的损伤识别方法。具体而言，建立了基于贡献模态的静力位移扩展新方法，构建了静力位移间接测量理论，分析了静力位移扩展误差构成机理，提出了基于间接测量的静力位移扩展误差控制策略。相应地，建立了基于贡献模态的动力响应扩展新方法，分析了动力响应扩展误差机理，提出了可靠筛选高质量非测点的方法。进一步，利用扩展后的完备或非完备静、动力响应，分别给出了基于静力位移/柔度灵敏度的结构损伤识别静力方法，研究了基于动力响应扩展识别模态参数的损伤识别动力方法，提出了基于静、动力响应扩展与密集连接卷积神经网络的结构损伤识别方法。多种类型结构的研究结果验证了本项目所提出的静、动力响应扩展理论，扩展误差控制策略和高质量非测点筛选方法以及损伤识别方法的有效性。本项目的研究丰富了结构健康监测相关理论，为大跨空间结构高效实施阶跃激励作用下静、动力测试和有效识别结构损伤提供了一种新的思路。

内容获取失败，请点击重试