Research On Remain Useful Life Prediction Of Intelligent Equipment Based On Multi-domain Degradation Features

As the key basic parts of complex equipment,the core components of complex equipment have a very important impact on the safety and reliability of the equipment operation.Predicting the remaining useful life(RUL)of the core components effectively and replacing or maintaining the core components in advance according to the prediction results can reduce the probability of failure of complex equipment.So as to reduce costs and increase efficiency.In recent years,tremendous attempts have been made to predict the RUL of components from the data-driven perspective.And it is important to extract features that can reflect the degradation states of equipment when predicting the RUL of complex equipment.Therefore,based on the original operating data,considering the core component mechanism,this paper extracts the multi-domain degradation features to reflect the performance degradation states of the core components.On this basis,the RUL prediction methods are studied.And the research achievements are applied to the intelligent operation and maintenance service system of hydroforming equipment for the purpose of ensuring the equipment to work reliably for a long time.The main research work and results of this paper are as follows:Firstly,the multi-domain degradation features that reflect the performance states of core components of complex equipment are extracted.On the basis of preprocessing the original operating data and considering the core component mechanism the multi-domain degradation features including multi-time domain feature indicators and mixing domain feature indicators are extracted.In order to eliminate the interference of redundant information between features,unique value features,collinear features,zero importance features and low importance features are selected.In addition,Kernel Principal Component Analysis(KPCA)is used for feature fusion to gain features that can reflect the performance degradation states of the components.Secondly,methods for predicting the RUL of core components of complex equipment are studied.First of all,support vector regression(SVR)and least square support vector regression(LSSVR)models are used to predict the RUL of the components.The study has found that the overall prediction performance of the LSSVR model is better than that of the SVR model.In order to improve the prediction accuracy of the core components' RUL,this paper proposes a RUL prediction method based on the Bagging model.Besides,the SVR model and LSSVR model are used as individual learners.It is found that the prediction error of the RUL prediction method based on the fusion of the Bagging model is smaller and can reflect the actual situation more effectively when compared to a single model.Finally,based on the algorithm studied in this paper,a set of operation and maintenance service system for hydroforming equipment is designed and developed on the basis of Spring Cloud architecture,which includes operation and maintenance management platform,equipment user mobile terminal and maintenance personnel mobile terminal.It supports equipment status monitoring,health analysis,life prediction and accident risk warning,etc.,realizing the intelligent and digitalization of the entire process of operation and maintenance services.