结构裂纹时序分析的几种识别法

Some Recognition Methods of Time Series Analysis for Structural Crack

本文利用随机时间序列模式识别技术对具有闭裂纹和开裂纹的工程梁结构进行了故障及其类型的诊断和判别。利用AR(n)模型建立并推导了Fisher 线性判别函数、广义残差方差判别函数、格林函数面积特征量和距离判别函数(即欧几里德距离、K—L 信息距离和广义残差方差距离)。对一悬臂梁在不同状态下的62组随机响应数据进行了故障诊断和判别分析。结果表明,本文所提出的识别方法具有良好的准确度和可靠性,可直接用于结构故障的实时在线诊断和状态监测。

In this paper,the diagnostic and decisive analyses of the failure and its type for an engineering beamwith close and open crack have been presented using pattern recognition techniques of random time series.The Fisher linear decision function,generalized residual variance decision function,Green function areaquantities and three distance decision functions are developed by use of model AR(n).Sixty-two sets ofrandom response dada for a cantilever beam in different conditions are computed.The results show thatthe methods presented in this paper have a good accuracy and reliability,and may be directly used for real-time,on-line diagnosis and condition monitoring of a structural failure.