Study On Adaptive Data Collection And Failure Prediction For Equipment Lifecycle

As more and more complex technology are applied to welding equipment in automatic welding flexible production lines,the possibility of welding equipment failure is also increased.Unplanned shutdown maintenance caused by unexpected failures results in not only the halt of some welding production lines which affects the production plan of company,but also missing the best maintenance opportunity and increased the maintenance cost and maintenance difficulty.Therefore,the design of a reasonable remote monitoring system for welding equipment can still monitor the status of the equipment remotely through the network when no one is on duty.The remote monitoring system is very important to predict the running trend of the equipment and predict failures of welding equipment.However,the traditional equipment monitoring system adopts the data collection method at equal time intervals without purpose,which not only wastes limited bandwidth and storage resources,but also has defects in the fitting accuracy of the equipment state.At the same time,the traditional equipment monitoring system seldom considers the performance of the failure rate to make the collected data that is not precise and representative,which affects the accuracy of equipment failure prediction.This thesis relies on Sichuan Science and Technology Plan "Automatic Flexible Production Line Lifecycle Management Intelligent Maintenance and Remote Monitoring System" Project(No.2017GZ0060),and takes the welding equipment of flexible manufacturing line for automobile manufacturing as the research object,aiming at improving the traditional equipment remote monitoring system.This thesis focuses on the following aspects:Firstly,according to the equipment failure rate curve,the life cycle theory was introduced to divide the equipment life cycle phase,and Hidden Markov Model(HMM)was used to estimate the HMM model parameters based on historical monitoring data.Select the monitoring point of welding equipment by analyzing the welding equipment failures always happening.Determine the life cycle stage of the equipment by remote monitoring data and determine its impact factor.Secondly,by analyzing the smoothness of remote monitoring data and combining the impact factors of the lifecycle stages,the system adjusted the data collection interval to collect more targeted and representative equipment data adaptively,thus making rational use of bandwidth and storage.Based on resources,it was also conducive to more accurate equipment state fitting.Finally,according to the collected monitoring data adaptively and the impact factors of the life cycle,the ARIMA-SVR combination forecasting model was improved.The prediction difference was revised,and the future state trend of the equipment state was predicted.Comparing with the standard threshold value,the equipment failure was judged to provide reference for preventive maintenance.Based on the above research,the remote monitoring and forecasting system for welding equipment was developed,and the failure prediction results was verified for specific.This research content has theoretical guidance and practical application value for the improvement of equipment remote monitoring system and fault prediction.