基于LSTM循环神经网络的织机了机预测  

Prediction of loomwarp-out time based on LSTM recurrent neural network

作　　者：徐开心 戴宁 汝欣[1] 胡旭东 XU Kaixin;DAI Ning;RU Xin;HU Xudong(Key Laboratory of Modern Textile Machinery Technology of Zhejiang Province,Zhejiang Sci-Tech University,Hangzhou 310018,China;College of Textile Science and Engineering(International Institute of Silk),Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]浙江理工大学浙江省现代纺织装备技术重点实验室,杭州310018 [2]浙江理工大学纺织科学与工程学院(国际丝绸学院),杭州310018

出　　处：《现代纺织技术》2023年第3期70-80,共11页Advanced Textile Technology

基　　金：浙江省博士后科研项目择优资助项目(ZJ2021038);浙江理工大学科研启动基金项目(11150131722114)。

摘　　要：不同织机由于生产情况和影响参数各异,实际的织布效率和了机时间也存在着很大的差别。针对利用预先设定好的计划生产静态参数对织机了机时间进行计算时,存在理论计算值与实际织机了机时间偏差过大的问题,提出了一种基于长短时记忆(Long short term memory,LSTM)循环神经网络的织机了机预测方法。从织机经纬向停车情况、人员工作效率、加工布匹品种3个方面出发,分析影响织机了机时间的各类因素,构建了具有时间序列特性的织机生产情况数据集。通过设置时间进度系数动态调整模型在织轴整个生命周期内的预测情况,并从损失程度和训练耗时两方面考虑对模型性能进行优化。最后,利用8组实验数据对模型的可靠性进行验证。结果表明:模型在了机预测截止时间的前30 h至前6 h,模型的预测结果值与实际值之间的平均误差范围为0.84 h至1.52 h,满足对实际生产时的所需指标要求。Loom production mainly refers to the process of weaving weft and warp on the weaving axis into cloth in vertical and horizontal directions.When the yarn of the loom axis,the axis,drop wires,heald,reed and warp yarn need to be cleaned from the loom.This process is called the loom changing the axis.If the axis is changed too early,the yarn material of the weaving axis will be wasted.If the warp axis is not set in time,too long stagnation of machine and insufficient length of remaining warp will cause problems such as inability to knot the new axis.Accurate prediction of warp-out time,timely arrangement of personnel for warp threading,warp knotting and axis change,so that the new weaving axis can be put into production in time,which has an important effect on improving the production efficiency of weaving.When calculating the looms warp-out time by using the pre-set static parameters of planned production,the deviation between the theoretical calculation value and the actual value is too large.Aiming to solve this problem,a looms warp-out time prediction method based on LSTM recurrent neural network proposed.Based on the analysis of the factors affecting the loom warp-out time from three aspects:the warp and weft stop of the loom,the working efficiency of personnel and the variety of cloth processed,a data set of loom production with time series characteristics s constructed.The prediction of the model in the whole life cycle of the weaving axis was dynamically adjusted by setting the time schedule coefficient,and the performance of the model was optimized from two aspects of loss degree and training time.On this basis,propose to set the dynamic loss change threshold to determine the optimal number of iterations of the model under different training data to improve the generalization ability of the model.Finally,the statistics of the consumption time of all kinds of stop states in the actual production of the loom re carried out,and the conclusion s drawn that the average percentage of the loom's stopping time in th

关 键 词：织机了机 LSTM循环神经网络 时间序列 经纬向停车 织轴 

分 类 号：TS111.8[轻工技术与工程—纺织材料与纺织品设计]