结构损伤识别的小波—人工免疫算法及优化研究

发布时间：2018-03-30 20:22

本文选题：小波分析　切入点：模态参量　出处：《长沙理工大学》2015年硕士论文

【摘要】：随着土建行业日新月异的发展,结构损伤诊断引起了人们的高度重视。对结构进行损伤检测和安全评估已成为研究的热点问题。小波分析被誉为“数学显微镜”,它具有空间局部化的性质,可以通过提取分析结构模态参量等信息,以信号图像中小波奇异点的方式表现结构的损伤位置。而人工免疫算法结合了传统人工智能算法和生物免疫系统的自适应能力两方面的优势,具备很强的信息处理能力及鲁棒性,运用其特性可以用于结构损伤程度的识别。为此,本文以结构损伤识别为研究内容,取得了如下主要成果:1.将小波分析优越的全局搜索能力与人工免疫算法强大的系统计算能力结合,提出了相应的损伤识别原理,建立了一种既能有效识别结构损伤位置,又能准确识别损伤程度的小波—人工免疫算法。2.为了更好的识别结构的损伤程度,论文采用欧式距离和Baldwin效应对标准免疫算法进行优化,改进了抗体种群多样性,突出了算法的记忆功能,新加入的算子使优化后的人工免疫算法收敛速度更快,计算效率更高,数值仿真计算验证了优化算法的有效性。3.论文通过有限元计算分析,并利用Matlab编制人工免疫算法程序,将小波分析和人工免疫算法相结合,小波系数图中模极大值点即为结构的损伤位置,将损伤位置作为人工免疫算法程序的已知量,那么该程序的输出参量即为结构的损伤程度。4.论文研究了简支梁分别含2处和3处裂缝,且裂缝深度不同的损伤识别问题。通过计算分析验证了本文所建立方法的正确性。在此基础上,进一步研究了连续梁分别含2处和3处裂缝,两层一跨框架结构分别含2处和3处裂缝,且裂缝深度不同的损伤识别问题。数值分析结果表明,本文所建立的小波—人工免疫算法既保留了小波分析能够快速准确地识别损伤位置的优点,又发挥了人工免疫算法精确计算损伤程度的优势。该方法可供结构损伤诊断的工程应用参考。
[Abstract]:With the rapid development of the civil engineering industry, Structural damage diagnosis has attracted great attention. Damage detection and safety assessment of structures have become a hot issue. Wavelet analysis is praised as a "mathematical microscope", which has the property of spatial localization. Information such as modal parameters of structures can be extracted and analyzed. The damage location of the structure is represented by signal image wavelet singularity, and the artificial immune algorithm combines the advantages of the traditional artificial intelligence algorithm and the adaptive ability of the biological immune system, and the artificial immune algorithm combines the advantages of the traditional artificial intelligence algorithm and the adaptive ability of the biological immune system. Because of its strong information processing ability and robustness, it can be used to identify the degree of structural damage. The main achievements are as follows: 1. Combining the superior global search ability of wavelet analysis with the powerful system computing ability of artificial immune algorithm, the corresponding damage identification principle is proposed, and a damage location identification method is established. In order to better identify the damage degree of the structure, the Euclidean distance and Baldwin effect are used to optimize the standard immune algorithm, which improves the diversity of antibody population. The memory function of the algorithm is highlighted, the new operator makes the improved artificial immune algorithm converge faster and the computational efficiency is higher. The numerical simulation proves the effectiveness of the optimization algorithm. 3. Using Matlab to program artificial immune algorithm, wavelet analysis and artificial immune algorithm are combined. The maximum value of modulus in wavelet coefficient graph is the damage position of structure, and the damage position is taken as the known quantity of artificial immune algorithm program. The output parameter of the program is the damage degree of the structure. 4. The paper studies that the simply supported beam contains 2 cracks and 3 cracks respectively. The method is proved to be correct by calculation and analysis. On the basis of this, two and three cracks in continuous beam are further studied. Two stories and one span frame structure contain two cracks and three cracks, and the crack depth is different. The numerical results show that, The wavelet artificial immune algorithm in this paper not only retains the advantage that wavelet analysis can quickly and accurately identify the damage location. The advantages of artificial immune algorithm for accurate calculation of damage degree are also brought into play. This method can be used as a reference for engineering application in structural damage diagnosis.
【学位授予单位】：长沙理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TU317

【相似文献】