基于参数降阶模态的分层贝叶斯在线裂纹检测  

Hierarchical Bayesian Online Crack Detection Based on Parameter Reduced Order Mode

作　　者：魏保立[1] 冯雅珊 罗坤[1] 付伟 WEI Baoli;FENG Yashan;LUO Kun;FU Wei(College of Electromechanical Engineering,Zhengzhou Railway Vocational&Technical College,Zhengzhou Henan 451460,China;School of Energy Science and Engineering,Henan Polytechnic University,Jiaozuo Henan 454003,China)

机构地区：[1]郑州铁路职业技术学院机电工程学院,河南郑州451460 [2]河南理工大学能源科学与工程学院,河南焦作454003

出　　处：《机床与液压》2024年第4期206-217,共12页Machine Tool & Hydraulics

基　　金：河南省科技攻关项目(222102230025)。

摘　　要：为了保证检测的稳定性和准确性,并且有效处理参数的关联性与强非线性,提出一种基于参数降阶模态的分层贝叶斯在线裂纹检测方法。提出一种用于递归输入、状态和参数估计的分层贝叶斯滤波器,它通过使用空间不完整和噪声可以仅输出测量振动。将所寻求的参数视为具有有限个更新状态的随机变量,参数状态的动力学由改进估计策略控制,该策略能够探索参数空间并识别目标值。进一步利用参数降阶模型对所提出的方法进行裂纹识别。最后通过航空航天应用中的真实组件对提出方法进行仿真,结果显示提出方法检测准确性和稳定性较高,证明了提出方法的有效性。In order to ensure the stability and accuracy of detection,and effectively deal with the correlation and strong nonlinearity of parameters,a hierarchical Bayesian online crack detection method based on parameter reduced mode was proposed.A hierarchical Bayesian filter for recursive input,state and parameter estimation was proposed,which could only output the measured vibration using spatial incompleteness and noise.The parameters sought were regarded as random variables with finite update states.The dynamics of the parameter states were controlled by an improved estimation strategy,which could explore the parameter space and identify the target val⁃ue.Furthermore,the parameter reduced order model was used to identify the cracks of the proposed method.Finally,the proposed method was simulated by real components in aerospace applications.The results show that the proposed method has high detection accuracy and stability,which proves the effectiveness of the proposed method.

关 键 词：裂纹检测 分层贝叶斯 参数降阶模态 随机变量 

分 类 号：TH133.33[机械工程—机械制造及自动化] TP18[自动化与计算机技术—控制理论与控制工程]