智能故障诊断

Intelligent Fault Diagnosis

一、引言长期以来,基于动力特性的结构故障诊断方法一直是国际学术界和工程界关注的热点,但至今缺乏有效的解决方案。结构故障诊断的核心是损伤识别与定位,其中,损伤定位问题尤为困难。损伤识别属于广义模式识别的范畴,因此,从逻辑的角度讲,要进行损伤识别与定位,首先需要解决损伤标识量的选择问题,即决定以哪些物理量为依据能够更好地识别和标定损伤的程度与方位。我们发现,用于损伤识别的物理量可以是全局量,如结构的固有频率等,但用于损伤定位的物理量最好是局域量。

A systematic introduction is made on the two type of methods for recognition and location of structural damage, one of which is based on strain type parameters (such as strain, strain modal shape, curvature modal shape, and so on), and another is based on displacement type parameters (such as displacement, displacement modal shape, flexibility matrices, and so on). The following phoneme is discovered that the methods based on displacement type parameters' exist possibility of giving a wrong damage location. Based on the localized characteristics of damage, some new formula for recognition and location of damage based on local parameter changes are suggested. Besides, some related problems such as face recognition or signature recognition are also discussed , and some of ideals are presented.