摘要
提出基于多状态时间序列预测学习的超精密机床主轴故障诊断方法,并对该方法进行了仿真测试。通过构建DAFDC-RNN模型(Dual-stage attention and full dimension convolution based recurrent neural network,时间序列预测模型),引入注意力、全维度卷积和时间注意力机制,生成运行状态过程中主轴间的相关性,模型输出即主轴运行故障预测值,同时去除预测数据内噪声,利用EMD-AR分析方法(Empirical mode decomposition-auto regressive,检验模态分解和谱分析结合法)分解时域信号中一阶分,并排除生成全新的信号,直至不再分解为止结束,在预测值范围内以及L-D算法帮助下对分解后信号计算求解,得出故障诊断结果。实验结果表明:本文方法的故障预测精度可达0.95~1.0,耗时可控制在10 ms以内,研究方法下主轴运行信号分解得到的正常信号分解幅值波动情况与实际结果基本吻合。
A fault diagnosis method for ultra precision machine tool spindle based on multi state time series predictive learning was proposed,and the simulation test of the method was completed.By constructing a DAFDC-RNN model(Dual-stage attention and full dimension convolution based recurrent neural network)introduces attention,full-dimensional convolution and time attention mechanisms to generate the correlation between spindles in the process of running state.The output of the model is the predicted value of spindle operating failure.The EMD-AR analysis method(Empirical mode decomposition-auto regressive)decomposes the first-order components in the time domain signal,and excludes the generation of a new signal until it is no longer decomposed.After decomposition,the signal is calculated and solved,and the fault diagnosis result is obtained.The experimental results show that the fault prediction accuracy of the proposed method can reach 0.95~1.0,and the time-consuming can be controlled within 10 ms.The amplitude fluctuation of the normal signal decomposed by the spindle running signal decomposition under the research method is basically consistent with the actual results.
作者
张朝刚
侍中楼
李敏
ZHANG Chao-gang;SHI Zhong-lou;LI Min(Engineering Training Centre,Jianghan University,Wuhan 430056,China;Engineering Training and Innovation Practice Center,Wuhan University,Wuhan 430056,China)
出处
《吉林大学学报（工学版）》
EI
CAS
CSCD
北大核心
2023年第11期3056-3061,共6页
Journal of Jilin University:Engineering and Technology Edition
基金
武汉市教育局项目(2021041)
国家级创新实验项目(202010265013).
关键词
多状态时间序列
序列预测
超精密机床
故障诊断
EMD
multi-state time series
sequence prediction
ultra-precision machine tools
fault diagnosis
EMD
