节点文献

模式识别在工程结构振动损伤识别中的应用

The Application of Pattern Recognition to Engineering Structural Dynamic Damage Dentification

【作者】 袁立侠

【导师】 李霆;

【作者基本信息】 华北工学院 , 机械设计及理论, 2004, 硕士

【摘要】 工程结构与人类的生活息息相关,工程结构的质量问题不仅关系到人民的生命财产安全,而且还会带来社会财富的巨大损失。但事实上,工程结构往往存在各种各样的缺陷和损伤。进行结构损伤诊断,可以提高结构的可靠性与安全性,避免重大事故的发生,减少事故危害性,可以获得潜在的巨大的经济效益和社会效益。结构损伤检测的方法很多,振动测试技术由于易于实现诊断系统的自动化、小型化,且理论基础雄厚、分析测试设备完善、诊断结果准确可靠,加上振动诊断具有实时性、在线性、遥测性、可控性和提取信号的方便性等诸多优点而在工程诊断技术中占主导地位,在结构损伤诊断中具有广阔的应用前景。振动诊断的方法很多,常用的有三种:直接分析法,时序分析法和参数识别法。其中参数识别法直接从测量的输入输出信号识别模态参数或物理参数的变化情况,具有很大的方便性。基于模态分析的振动诊断技术自 20 世纪 60 年代兴起至今,已在工程领域得到了广泛的应用。从目前来看,单一模态参数的诊断方法比较常见。而基于模态参数识别的振动诊断属于动力学的反问题,从数学上看,反问题往往不唯一和不确定,从而使识别结果可能不唯一,出现误判。所以目前多参量的综合诊断技术已经兴起,模式识别,人工智能,专家系统等的研究成果为损伤诊断注入了新的活力。其中模式识别是人工智能的一个重要分支,将模式识别或人工智能的理论和方法应用于结构损伤诊断,发展智能化的损伤诊断技术已成为当今故障诊断技术的发展的主流。本文选取悬臂梁作为典型构件,将模式识别理论和方法应用到基于模态分析的振动诊断技术中,利用由模态参数构造的损伤信息特征量建立损伤模式,对结构损伤的位置和程度进行分步识别,从而提高了损伤诊断的精度。

【Abstract】 Engineering structure is closely linked with human Beings’ life; itsquality problem not only connects to the safety of the people’s life and wealth,but also causes great loss of social wealth. But engineering structure sometimesexists many types defects and damage in fact. The reliability and security ofstructure can be improved by structural damage detection, the occurrence ofmomentous accident can be avoided, the harm degree of accident can be minished,and the great potential economy benefit and social benefit can be obtained. There are many kinds of structural damage detection method, but dynamicdamage detection technology is prone to realize automatization andminiaturization of system, the foundation of its theory is abundant, theequipment to analyze and test is perfect, the detection result is nicety andcredibility, and dynamic damage detection technology have many abilities, forexample, real time, on line, remote test, controllability and convenience ofpicking up signal, so it is in dominant in engineering structural damagedetection technology, it has wide use foreground in structural damageidentification. There are many kinds of dynamic damage detection methods, there are three<WP=6>types of methods in common use: direct analytic method, time order analyticmethod and parameter recognition method. In these three methods, parameterrecognition method recognizes the changes of modal parameter or physicalparameter from metrical input and output signal directly, so it is convenient.From 1960s to this day, the dynamic damage detection technology based on modalanalysis has been used widely. Detection method based on single modal parameteris relative common at present. But the dynamic damage detection based onrecognition of modal parameter belongs to “inverse problem” in dynamics,“inverse problem” sometimes is not single and certain, so the detection resultis likely to be not single, and the wrong detection occurs. So the syntheticallydetection technology based on multi-parameter rises, the research productionof pattern recognition, artificial intelligence and expert system etc. infusesdamage detection with fresh energy. In these methods, pattern recognition isan important embranchment of artificial intelligence, it is a leading directionin today’s development of damage detection technology to developintelligentized technology by appliance of theory and technique of patternrecognition or artificial intelligence to structural damage detection. In this paper, cantilever is chosen as typical component, the theory andtechnique of pattern recognition is used to the dynamic detection technologybased on modal analysis, damage pattern is established by damage informationeigenvector from modal parameter, the structural damage position and extensionare recognized respectively, consequently precision of detection is improved.

  • 【网络出版投稿人】 华北工学院
  • 【网络出版年期】2004年 04期
  • 【分类号】TU317
  • 【被引频次】4
  • 【下载频次】317
节点文献中: 

本文链接的文献网络图示:

本文的引文网络