Research On Staged Remaining Useful Life Prediction Method Of Components Based On Wiener Process

The method based on degradation modeling has been recognized as the basic and effective method for estimating the remaining life of various health management activities.This method can not only ensure the reliable and safe operation of the degraded system,but also avoid the occurrence of major accidents and economic losses of some large equipment due to the expiration of the life of some components.As one of the most effective stochastic modeling methods,the research of single-stage remaining useful life prediction method based on Wiener process has achieved significant development and extensive application in the past decades,but the staged remaining useful life has not been studied in depth.In fact,due to the influence of internal factors(such as mutation of degradation mechanism)or external factors(such as dynamic environment,state switching,etc.),the degradation characteristics of many components show two-stage or even multi-stage degradation characteristics.This staged degradation process is more consistent with the degradation trend of components under actual conditions.Therefore,in order to make the degradation model more consistent with the degradation trend of components,this paper studies the construction of the staged degradation model and the degradation indicator and does the following work:(1)For a single type of drift function,the degradation trajectory of components cannot be tracked accurately for a long time,which reduces the accuracy of model prediction.In this paper,a two-stage remaining useful life prediction model for degradation angle identification is proposed.Two drift functions are used to estimate the degradation state of components and predict the remaining useful life of components.First,the concept of degradation angle is proposed to identify the abrupt moment of slow degradation and accelerated degradation of components.Then,based on the definition of first arrival time and degradation angle,this paper obtains the probability density function of slow degradation stage based on power function type drift function and the probability density function of accelerated degradation stage based on exponential function type drift function,and then estimates the degradation state of components.(2)The degradation information of components collected from a single sensor is insufficient and the degradation indicator construction process cannot be updated,resulting in that the constructed degradation indicator cannot accurately represent the degradation track of components.Based on this,this paper proposes a degradation indicator construction method for multi-sensor and multi-feature fusion.The degradation information collected by multiple sensors is initially fused,and the sensor fusion technology and feature fusion technology are updated using the prognosis information,and then the degradation indicator of the component is constructed.(3)The degradation model cannot adaptively track the staged degradation process of components,resulting in the reduction of the accuracy of the remaining useful life prediction method.Therefore,this paper proposes an adaptive staged prediction method.According to the degradation mechanism of the component,the degradation process of the component is divided into four degradation modes.The degradation process of the component is divided adaptively,and then different degradation models match the different degradation stages of the component.Then,a four-step method is proposed to estimate and update unknown parameters in the model to solve the problem of solving the complexity of parameters.Then,based on the definition of first arrival time,the probability density function of adaptive staged remaining useful life is obtained to predict the remaining life of components.