【作者】 张娜；

【导师】 时培明；

【作者基本信息】 燕山大学 ， 精密仪器及机械， 2015， 硕士

【摘要】 大型承载钢结构在工程领域应用广泛,其安全性、可靠性和正常使用寿命越来越被人们所关注。大型承载钢结构在长期服役过程中,不断的承受载荷循环加载、环境腐蚀、材料老化等因素的影响,会不可避免的出现损伤,及时进行损伤诊断是保证其安全使用的第一步。大型承载钢结构的安全使用是工程运行的基础,若发生事故直接带来严重的经济损失,甚至造成人员伤亡。故进行大型承载钢结构损伤诊断,不仅可以避免灾难性事故的发生,而且可以为工程正常运行提供保障。目前大多数的损伤诊断方法仅限于在简单结构上的实现,对于大型复杂结构的损伤诊断还存在只能诊断单损伤、诊断时间长等各种各样的缺点。为实现对其进行多损伤位置和程度诊断,本文根据遗传算法的多峰值优化特性,开发了基于遗传算法的大型承载钢结构损伤诊断程序,综合考虑了单损伤和多损伤工况。首先,分析了结构损伤诊断所用参数,运用ANSYS参数化编程语言建立了简支梁和井架钢结构的数值模型,通过有限元模态分析得到结构完好状态下的动态特性参数,为后续的研究提供数值基础。其次,针对遗传算法收敛速度慢的情况,采用收敛速度快、避免种群早熟的自适应遗传算法进行损伤诊断,通过对比两种算法得出:对于简支梁结构损伤诊断,自适应遗传算法优于普通遗传算法;对于井架钢结构,普通遗传算法不能实现损伤诊断,自适应遗传算法可实现单损伤诊断但诊断结果不够精确。最后,根据仅应用遗传算法不能有效的实现大型承载钢结构的损伤诊断的结论,结合大型承载钢结构的结构特性,研究了基于结构静动力参数和遗传算法的多损伤分步诊断方法。利用单元应变模态和节点静力位移诊断多损伤位置;提取定位出的损伤单元作为遗传算法的编码变量,应用结构固有频率构造遗传算法目标函数,应用遗传算法诊断多损伤程度。通过在井架钢结构有限元模型上计算,实现了单损伤、多损伤的位置和程度诊断。更多还原

【Abstract】 Increasingly concern is expressed on the safety:reliability,and normal life of large bearing steel structures, which widely used in the engineenig field. Under the impact of factors such as contiouous losding of cyclic loading, environmental degradation,materian aging and so on, large bearing steel structures will be inevitablyb damaged though a long service To insure its safe use, damage detection in time should be put on the firsst step. The site use of large bearing steel structures is the basis of engineering operation, otherwise serious economic losses and even casualties will be caused during see/dents. Therefore, large beefing steel damage detection can not only. avoid the occurrence of catastrophic accidents, but provide protection for nomal operation.Presently most of damage detection methods are restricted to simple structues, and cause all softs of problems in large complex structure damage detection, including limitation to single injury, and consuming a lot of time.The articla conducted damage detection of largee bearing steel structures in genetic algorithm with multimodal optimizstion rectums, and the damage location and degree of single injury and multiple injuries are concerned synthetically.Firstly, paramaters used stmctural damage detection were analyand, and simply suppicted baam modal md deerick steel stuctore modal were bulit using the ANSYS parametric programming language.Dynamic characteristic paramaters of structure in good condition were obtainad through the finito alemmant modal analysis, which provides numerical basis for follow-up study.Seeondly,aceotding to previous resaarch results of slow cornvergence with genetic algorithm, adaptive ganatic algorithm which can fast convarge and avoid population prematurity was used.The imphnnmtation process of damaage detection of large bearing steel structures is varifiad on a simply supported beam model with two mathods mentioned above. Cemparative analysis of damage detection results are ss follows,for simply supported beam model, adaptiva genetic algorithm is suparior to the osdinary: for derrick steel structuces, osdinary eenetic sleorithm can’t detect the damaee, and adaotive genetic algorithm can detect single damage while the result is’ t accurate enough. Finally, damage detection methods using structural parameters and genetic algorithm were advised, because of the failure to conduct effectively large bearing steel structures damage detection with genetic algorithms. In this article, two step detection motheod for multiple injuries based on structural parameters and genetic algorithm is studied. Static structure parameters and dynamic parameters were extracted to locate damage. Damage detection procedure of large bearing steel structures based on genetic algorithm used to conduct damage degree detection on damage unites located. The damage location and degree of single injury and multiple injuries was achieved on numerical model of derrick steel structures with numerical model calculation.更多还原