Research On Remaining Useful Life Prediction Based On DS Evidence Theory Information Fusion

The prediction of remaining useful life is the foundation and premise for the establishment of system equipment maintenance methods.Accurate residual life forecast can provide useful information for the maintenance program,reduce the cost of maintenance,and enhance the security and efficiency of the equipment.Therefore,it is very important to predict the remaining useful life for the medical management of complicated systems in industry.In the era of big data,the health status data of equipment during operation can be easily obtained through various sensors,which provides important data support for the prediction of the remaining useful life of equipment.In the face of different types of sensors and different sensitivities,how to conduct information fusion of multi-sensor data has become a research hotspot.Based on the Wiener process,this paper adopts the information fusion method of Dempster-Shafer’s evidence theory to build the forecast model of the system’s remaining useful life.In the case of a multi-sensor device,based on the uncertainty of sensor monitoring data and the contradiction between the data of sensor data and the information fusion,Dempster-Shafer’s evidence theory is improved.Then the remaining lifetime of the equipment is researched and validated with the aeroengine data set.The main contents of the study are as follows:(1)Based on the fusion of multi-sensor data,a method of residual life forecasting based on Dempster-Shafer Evidence Theory is presented.First,the health status of the system was established by Wiener Process Degradation Model,and then the degradation data of every sensor was used as evidence.Then,the basic probability distribution function and the evidence recognition frame are constructed,and the method of discount is used to improve the evidence source.At last,Dempster-Shafer’s Evidence Theory Synthesis Rule was used to fuse the unknown parameters,and the remaining life forecast results were obtained.Finally,an example is given to validate the proposed method by the air engine degradation monitoring data,which proves the validity of the proposed method.(2)Based on the traditional Dempster-Shafer evidence theory synthesis rule,the remaining life prediction method of evidence theory considering evidence credibility is proposed.Firstly,a basic probability distribution model and evidence recognition framework are constructed according to the evidence data sources provided by multiple sensors.Then,evidence credibility is introduced to improve the traditional Dempster-Shafer evidence theory synthesis rule.Finally,the remaining useful life prediction model of data fusion is constructed to deduce the probability density function of the remaining useful life of equipment.Through the sensor degradation data set of aircraft engine,the corresponding experimental demonstration is carried out to verify the practicability of the proposed method.