工程动态信号的精密测试与智能评估方法  

High-Precise Measurement and Intelligence Evaluation of Engineering Dynamic Signal

作　　者：沈松[1] 刘进明[1] 鄢明 王治敏 SHEN Song;LIU Jinming;YAN Ming;WANG Zhimin(China Orient Institute of Noise and Vibration Beijing,100085,China)

机构地区：[1]北京东方振动和噪声技术研究所,北京100085

出　　处：《振动．测试与诊断》2023年第6期1049-1057,1239,共10页Journal of Vibration,Measurement & Diagnosis

摘　　要：为实现工程动态测试中的精密计算,首先,讨论了频率、幅值、相位和阻尼等基本参量的校正计算方法;其次,研究了动态范围、频响特性和微积分转换等动态系统特性的扩展;最后,举例介绍了基于综合系统建模的高精度动态信号测试方法。对测量后的信号使用人工智能方法进行识别和评估,讨论了将机理特征提取与深度学习相结合的方法,介绍了样本增强、迁移学习和自编码学习等方法,用于解决工程中少样本或无负样本的问题。For precise calculation in engineering dynamic testing,the correct method of basic parameters is firstly discussed,including frequency,amplitude,phase and damping.Then the dynamic feature improvement method of dynamic range,frequency response and calculus is studied.Finally the comprehensive system modeling method is introduced for high-precise dynamic signal measurement.For signal identification and evaluation base on artificial intelligence,the combination method of feature extraction and deep learning is discussed,and several methods are introduced including sample enhance,transfer learning and self-encode learning,in order to solve the problem of few or no negative samples in engineering.

关 键 词：动态测试 精密测量 信号处理 深度学习 

分 类 号：O329[理学—一般力学与力学基础] TB123[理学—力学] TH825[理学—工程力学]