摘要
液压缸的工况错综复杂,为了确保液压缸的正常运行,寿命预测系统采集了大量数据以获悉液压缸的寿命状况。针对液压缸监测信号噪声大、单一分类器分类性能不佳的问题,提出了一种基于深度学习的液压缸寿命预测方法。利用DAE算法对噪声数据进行重构,以完成数据的特征提取;利用BP神经网络对数据中各特征子集进行分别训练构成弱分类器,然后采用Adaboost算法对弱分类器进行加权合并成强分类器以实现数据的特征选择。通过实验验证,提出方法可有效提高液压缸的寿命预测精度。
The working condition of hydraulic cylinder is complex,in order to ensure the normal operation,the life prediction system collects data to get the life-span status.Aiming at the problem of high-noise of monitoring signal and poor classification of single classifier,a method of life-span prediction of hydraulic cylinder based on deep learning is proposed.DAE algorithm is used to reconstruct the noise data to extract features of data,BP neural network is weak classifier for training feature subsets of data,Ada⁃boost algorithm is strong classifier to merge weak classifiers for the feature selection of data.Experiments show that the proposed method can improve the life-span prediction accuracy of hydraulic cylinders effectively.
作者
高谦
肖维
GAO Qian;XIAO Wei(Hohai University,Nanjing 210000;Shenzhen University,Shenzhen 518000)
出处
《计算机与数字工程》
2021年第1期36-40,共5页
Computer & Digital Engineering
关键词
液压缸
数据降噪
分类器
寿命预测
hydraulic cylinder
data denoising
classifier
life-span prediction
作者简介
高谦,男,硕士研究生,研究方向:计算机视觉,故障诊断;肖维,男,硕士研究生,研究方向:智能制造,机器人轨迹规划。
