工业机器人的传动效率直接影响到工业节能效果.针对齿轮箱故障诊断能力,设计了一种基于深度主动学习(Deep Active Learning,DAL)的工业机器人RV齿轮箱故障识别方法,并开展测试分析.研究结果表明,边缘抽样、深度学习、深度主动学习、随机抽样分别达到77.23％、94.35％、97.18％和84.26％的准确率,表明该方法可以有效地过滤出较难鉴别的样品.分析混淆矩阵表明,在只利用小样本进行训练的情况下,DAL算法取得了比传统学习算法更优的处理结果.该研究有助于提高工业机器人控制效率,具有广泛的应用价值.
Fault Identification and Analysis of RV Gear Box of Industrial Robot Based on DAL
The transmission efficiency of industrial robots directly affects the effect of industrial energy saving.Aiming at the fault diagnosis capability of gearbox,a fault identification method of RV gearbox of industrial robot based on deep active learning(DAL)was designed and tested.The results show that the accuracy of edge sampling,deep learning,deep active learning and random sampling can reach 77.23％,94.35％,97.18％and 84.26％respectively,indicating that the method can effectively filter out difficult samples.The analysis of confusion matrix shows that the DAL algorithm can get better results than the traditional learning algorithm when only using small samples for training.This research is helpful to improve the control efficiency of industrial robots and has wide application value.
industrial robotRV gear boxfault identificationdeep active learning
万方数据
维普
