Non-linear equipment residual life prediction method based on Box-Cox transformation

Abstract

The invention belongs to the technical field of life prediction, and particularly discloses a non-linear equipment residual life prediction method based on Box-Cox transformation, which comprises unknown parameter estimation of historical random degradation equipment and residual service life prediction of online service equipment. According to the invention, the non-linear degradation data is subjected to transformation processing by utilizing Box-Cox transformation, so that the linear degree of the degradation data of the random degradation equipment is improved; an accurate analytic solution of the residual life can be obtained by modeling the transformed degradation data by using the existing linear Wiener process, so that the constraint that the selection of the existing time scale transformation function is limited by the failure data of the random degradation equipment is avoided.

Description

technical field [0001] The invention belongs to the technical field of remaining life prediction and relates to a method for predicting the remaining life of key equipment, in particular to a method for predicting the remaining life of non-linear equipment based on Box-Cox transformation. Background technique [0002] Remaining life prediction and health management technology is a key technology for modern complex engineering systems, major products, and major facilities to improve operational reliability, safety, and maintainability. It can provide an important guarantee for the long-term safe and reliable operation of major equipment. With the development of advanced sensing and condition monitoring technology, it has become possible to obtain monitoring data of performance degradation process that can reflect the health status of equipment. Under this background demand, data-driven residual life prediction technology based on stochastic process modeling for stochastic deg...

AI Technical Summary

Problems solved by technology

For the modeling of nonlinear degraded data, two ideas are generally adopted: one is to use logarithmic transformation or time scale transformation to approximate the nonlinear degraded data into linear degraded data, and then use the linear Wiener process to model the degraded data, and then obtain the remaining The exact analytical solution of the life probability density function, the common time scale transformation functions include power function, polynomial function, etc. The choice of the specific function form depends on the degradation data of the specific object, which is feasible only when the equipment degradation and life data are sufficient. However, for For high-reliability equipment, it is unrealistic to obtain a large amount of degradation and life data. At the same time, the linearization technology of nonlinear degradation data based on logarithmic transformation requires the degradation data to have exponential characteristics, so its applicable objects are limited; the second is based on a The general nonlinear time-varying drift coefficient function, through the time-space transformation of the nonlinear stochastic degradation model, transforms the problem of the nonlinear stochastic process first reaching the fixed boundary into the problem of the standard Brownian motion first reaching the time-varying boundary, which is different from Based on nonlinear data reconstruction techniques such as time scale transformation and logarithmic transformation, however, based on this general nonlinear Wiener process model, only an approximate solution to the remaining life can be obtained

Images