机器学习算法下通信电子线路微损伤检测  

Microdamage Detection of Communication Electronic Lines Based on Machine Learning Algorithm

作　　者：王膺钦 赵春 WANG Yingqin;ZHAO Chun(School of Information Engineering,Nanchang University,Nanchang,Jiangxi 330000,China;Chengdu Jincheng University,Chengdu,Sichuan 611731,China)

机构地区：[1]南昌大学信息工程学院,江西南昌330000 [2]成都锦城学院,四川成都611731

出　　处：《计算技术与自动化》2024年第3期70-74,共5页Computing Technology and Automation

摘　　要：在通信电子线路损伤检测过程中,由于一些微损伤特征较难察觉,从而导致损伤检测准确率较低。为此,提出了机器学习算法下的通信电子线路微损伤检测方法,实现对电子线路微损伤的精准、高效检测。引入小波包特征分析,提取小波讯号动能特征,计算损伤敏感度,构建基于一维卷积神经网络的通信电子线路微损伤检测模型,将提取到的动能特征作为模型输入,经特征样本区别标明、训练数据集合有放回抽样,实现模型参量更新,最终在最佳训练模型中输入待测线路样本以实现电子线路微损伤检测。实验表明:所提算法有效检测出通信电子线路的微损伤,对微损伤检测应用性和可行性较高,且抗噪声干扰能力强。In the process of damage detection in communication electronic circuits,the accuracy of damage detection is low due to some microdamage features that are difficult to detect.Therefore,a machine learning algorithm based microdamage detection method for communication electronic circuits is proposed to achieve accurate and efficient detection of microdamage in electronic circuits.Introducing wavelet packet feature analysis,extracting the kinetic energy features of wavelet signals,calculating damage sensitivity,and constructing a communication electronic circuit microdamage detection model based on one-dimensional convolutional neural network.The extracted kinetic energy features are used as model inputs,and the feature samples are distinguished and labeled,and the training data set is put back for sampling to achieve model parameter updates.Finally,the electronic circuit microdamage detection is achieved by inputting the tested circuit samples into the optimal training model.The experiment shows that the proposed algorithm effectively detects microdamage in communication electronic circuits,has high applicability and feasibility for micro damage detection,and has strong anti noise interference ability.

关 键 词：机器学习算法 通信电子线路 微损伤检测 小波包 小波讯号 卷积神经网络 

分 类 号：V448[航空宇航科学与技术—飞行器设计]