Device Health Prediction Based On Real-time Information And Maintenance Optimization Method Design

Equipment is a basic resource to ensure the normal production and operation of enterprises.Therefore,mastering the real-time health status of production equipment is of great significance to ensure the safe and reliable operation of production systems.This dissertation takes the processing equipment in the complex mechanical product processing system as the research object,and proposes a device health prediction method based on real-time monitoring information.Based on this method,the equipment maintenance management system is established to realize the scientific management of the processing equipment.First,the article reviews the significance of equipment fault diagnosis and health prediction,and describes several prediction methods.On this basis,the hidden half Markov model(HSMM),which can achieve fast response,high prediction accuracy and mature development,is selected to realize the real-time health state prediction of processing equipment.The support vector machine that can effectively solve the pattern recognition problem is determined to deal with the state recognition problem.Secondly,according to the operation process of the processing equipment,a data acquisition method based on program fragments is proposed.Due to the diversity of monitoring data,the feature extraction and dimension reduction methods of multi-class data are studied.Based on this,an implicit semi-Markov health prediction model based on multi-class monitoring data is established,and the experimental verification is carried out.The multi-factor failure prediction failure problem is found.Thirdly,the problem of predicting failure is analyzed,and the concept of true impact factor of fault is proposed.Then the support vector machine method(SVM)which can mine the real influencing factors of faults is expounded,and a combined forecasting model which can adapt to new faults and distinguish similar faults is proposed.And the test verification was carried out again to solve the multi-factor failure problem.Finally,based on the combined model and a plant information management system,the equipment maintenance system with real impact factor library,equipment health prediction,maintenance plan development,spare parts demand forecasting and other functions is designed and developed.