基于统计模式识别的压力管道损伤检测

Damage detection of pressure pipe based on statistical pattern recognition theory

基于统计模式识别理论,采用AMRA时序分析方法,通过长自回归模型计算残差法和最小二乘法的结合来估计模型参数,从而建立了系统模型。运用均值控制图的方法对压力管道的振动信息特征进行提取、选择,从而有效地判别压力管道的无损与损伤问题。数值模拟结果表明,基于统计模式识别的均值控制图的结构异常检验方法,能够准确诊断检测结构的损伤状态,而且对于损伤程度和损伤位置都有很强的敏感性。

Based on the statistical patt ern recognition theory, t he AMRA tim ing analysis method combined w ith the long autore-gressive model residual method and the least squares method w ere used to estimate the model parameters and then to develop the system model. The mean cont rol chart method w as used to extract and select the vibration information and feature of the pressure pipe, therefore the damage of the pressure pipe can be ident ified effectively. T he simulation result s show ed that struc-tural abnormality test method of the mean control chart, w hich is based on the statist ical pattern recognition theory, can diag-nose the structural damage accurately, and is sensitive to the damage degree and location.