基于Word2vec和LSTM-SVM的数控机床故障报警预测.pdfVIP

第4期 组合机床与自动化加工技术 No. 4 2023年4月 Modular Machine Tool Automatic Manufacturing Technique Apr. 2023 文章编号： 1001 -2265(2023)04 -0071 -05 DOI ： 10.13462/j. cnkb mmwmh 2023.04. 017 基Word2vec和LSTMEVM的数控机床故障 报警预测 王梓琦，张铫,夏雨风，任杰文 (东北大学秦皇岛分校计算机与通信工程学院，秦皇岛066004) 摘要 :规模自动化工业生产中的集群数控机床因各种故障导致停机而造成生产线效率的下降 ，若能 及时准确地预测故障对数控机床进行预检预修有利于提高全线生产效率 。在工业智能制造背景 下，以数据驱动为支撑，数控机床积累的大量历史故障报警数据为依托，设计了一种基于WoW2vac 和LSTM-SV 的故障报警预测方法对机床未来可能发生的故障进行预测 。首先通过词嵌入技术将 报警文本向量化，然后将报警向量作为输入构建长短期记忆神经网络long show term memo— net­ work, LST )预测模型，并使用支持向量机(suppoW vectcr machina, SV )代替传统的softwax作为模 型的末端分类器 ，实验结果表明该方法具有更高的预测准确率。 关键词：故障预测；报警数据;词嵌入；长短期记忆神经网络；支持向量机 中图分类号:TH16 ；TG659 文献标识码 :A Fault Alarm Prediction of CNC Machine Tools Based on Word2vec and LSTM-SVM WANG Zi-qi ,ZHANG Yao ,XIA YuWeng ,REN Jic-wen (School of Computes and Communication Engineering, Northeastern University at Qinhuangdao, Qinhuangdao 066004 , China ) Abstract: The efficiency of the production line is reduced because of the shutdown due to various faults of the cluster CNC machine tools in scale automated industhal production. Timely and accurately fault predic­ tion and pre-inspection and pre-repai.c of CNC machine tools are beneficial to improve the production eai- ciency of the whole —e. UndeT the background of industrial inteUigent manufacturing, a fault alam pedic- hon method based on Word2vec and LSTM-SV is designed to predict the possible future faultt of CNC machine tools according to a large number of hitorical fault alarm data accumulated by CNC machine tools and supported by date de ven. First of all, the alam text is vectorized by the word embedding technology, and then