Limited data driving long-life part residual life prediction method

Abstract

The invention discloses a limited data driving long-life part residual life prediction method comprising the following steps: using a wavelet-envelope to analyze and preprocess long life part history monitoring data, thus reducing data noises; building a multiparameter residual error correction autoregression gray long-term prediction model for the preprocessed data; using combined cooperate gaming to map autonomous selection optimal parameters; mapping a failure mode and a failure mode in an influence analysis system as reliability evaluation indexes of the prediction model, and thus real time evaluating prediction result reliability. The advantages are that the cooperate gaming and the residual error correction autoregression gray long-term prediction model are organically combined, thus fully utilizing long life part limited monitoring data, and solving a series of problems in the prior art that a conventional prediction method cannot respond in long time, is large in data amount demand, multiparameter selection depends on expert knowledge, and the prediction reliability cannot be evaluated in real time; the novel method can improve long-life part residual life prediction result reliability and accuracy under the limited data background; the method is suitable for residual life prediction of the highly reliable long life part with limited data and high test cost; the novel method has universality.

Description

technical field [0001] The invention relates to a method for predicting the remaining life of long-life components driven by limited data, which is a data-driven life prediction method for high-reliability and high-cost long-life components, and belongs to the cross field of engineering application and information science. Background technique [0002] With the discovery of new materials and the development of manufacturing technology, the reliability of equipment components in the field of engineering applications (such as spacecraft inertial navigation components, electronic equipment components, telemetry sensors, etc.) is getting higher and higher, and the working life is getting longer and longer. , ranging from a few years to decades. The high manufacturing cost and high reliability of these long-lived components have brought great challenges to the study of their final service life and the determination of maintenance time nodes. How to obtain the remaining life of l...

AI Technical Summary

Problems solved by technology

This method fully considers the characteristics of high reliability and high cost of long-life components, and solves the problems that traditional prediction models cannot respond for a long time, requires a large amount of data, and has poor nonlinear data processing capabilities; at the same time, based on this method, specific long-term Independent selection of multi-parameters based on the data characteristics of life parts to solve the problem that the multi-parameter prediction model cannot independently select the optimal parameter combination

Images