测控技术2025，Vol.44Issue(7)：11-18,8.DOI:10.19708/j.ckjs.2025.07.301
基于增量学习的转台故障诊断研究
Research on Turntable Fault Diagnosis Based on Incremental Learning
摘要
Abstract
The existing diagnostic methods for common types of faults such as overspeed,abnormal noises,reso-nance,and emergency stops during the operation of the turntable rely on engineering experience,are suscepti-ble to control signal interference,and have insufficient accuracy.To overcome these issues,a tumtable fault di-agnosis method based on incremental learning is proposed.Operational data from a three-axis vertical turntable are analyzed,focusing on the time and frequency characteristics of motor current signals selected for detection.A one-dimensional convolutional neural network(CNN)is used to extract key features from the data.An incre-mental learning strategy is used,and the herding algorithm is used to construct representative samples.A distil-lation loss is added to the original loss function to retrain the model,enabling the identification of new fault types.Fault detection experiments under various operating modes are conducted and compared with existing methods.Results show that the proposed method maintains high accuracy for historical faults and achieves over 98％recognition rate for new faults.Training time is reduced by 73.68％,effectively preventing catastrophic forgetting.The model demonstrates high accuracy and robustness,providing a reliable solution for turntable fault diagnosis.关键词
电机电流/增量学习/故障诊断/卷积神经网络Key words
motor current/incremental learning/fault diagnosis/CNN分类
机械制造引用本文复制引用
成喆敏,马雅琼,闫斌斌,高峰,刘梦..基于增量学习的转台故障诊断研究[J].测控技术,2025,44(7):11-18,8.基金项目
航空科学基金项目(20200001053005) （20200001053005）
