Aero-engine remaining life prediction method based on multi-stage information fusion

Abstract

The invention discloses an aero-engine remaining life prediction method based on multi-stage information fusion. The method comprises the steps that multi-source monitoring parameter denoising processing and feature extraction are conducted; stable analysis is conducted on multi-source monitoring time sequences, sudden change points of all parameter monitoring time sequences are calculated and parameter degeneration proportions at the positions of the sudden change points are calculated; multi-stage division is conducted on multi-source parameters, a regression fusion model is established, sample training is conducted by using historical monitoring data and parameters, in multiple stages, of the fusion model are obtained; according to monitoring data in a training set, the multi-source monitoring parameters are fused and a health indicator HI is obtained; by using the Kalman filtering algorithm, best fit is conducted on an engine in the whole process that the performance fails from being complete, and the error of a prediction model is minimized; according to real-time monitoring data in a test set, the multi-source monitoring parameters are fused and a health indicator HI is obtained; time-varying parameters of the prediction model are estimated in real time by using the Kalman filtering algorithm; the prediction model is determined, the time mechanism is introduced and the failure time of the engine is estimated in real time.

Description

Technical field: [0001] The invention relates to a method for predicting the remaining life of an engine, in particular to a method for predicting the remaining life of an aeroengine based on multi-stage information fusion. Background technique: [0002] As the core component of the aircraft, the structure of the aero engine is extremely complex. Therefore, a single monitoring parameter cannot accurately characterize its performance. In order to comprehensively use observation information to accurately describe the real-time health of the engine, information fusion technology has been widely used in engine health management, such as fault diagnosis based on information fusion, performance evaluation, and performance trend prediction. Among them, the prediction of the recession trend is particularly important for the predictive maintenance decision of the engine, and it is a research hotspot in the process of the current planned maintenance to the predictive maintenance. [0003] I...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the problems that cannot be solved by common engine remaining life prediction methods, the present invention improves the existing engine health state prediction technology, realizes multi-source data fusion in stages, establishes a prediction model based on multi-stage information fusion results and utilizes multi-stage training The parameters of the model are estimated by samples, real-time monitoring data is introduced, and intelligent algorithms are used to update and predict the parameters of the multi-stage model, and finally predict the performance decline trend of the engine, and obtain an accurate estimate of the remaining service life

Images