Aero-engine life prediction method based on singular value decomposition and GRU

Abstract

The invention discloses an aero-engine life prediction method based on singular value decomposition and GRU. The method comprises the following steps: preprocessing original data, performing dimensionreduction and reconstruction on residual characteristic data through SVD, removing redundant information, and screening effective characteristics to serve as characteristic data for measuring the service life of an engine; and then deeply mining historical information through a GRU neural network to achieve prediction of the service life of the aero-engine. The method can extract the main characteristic data of engine information, eliminates the interference of irrelevant factors, and improves the prediction precision of the service life of an aero-engine.

Description

Technical field[0001]The invention belongs to the field of aviation safety, and specifically relates to a life prediction method.Background technique[0002]As a kind of power plant system, aero engine plays a key role in the development of aircraft. It is not only the "heart" of the aircraft, but also the source of power to promote the rapid development of aircraft. However, due to long-term service under conditions of high temperature, high pressure, high speed, alternating load, etc., aero engines have extremely high requirements for safety and reliability. Once a key component fails, it will often cause high maintenance costs and even huge flights. disaster. Therefore, in order to ensure the safe and stable operation of the aircraft, it is important to predict the aircraft status in advance, especially the safety status of aero engines. Predicting the remaining service life of an aero engine can grasp the engine state in advance, which is essential for safe flight.[0003]With the r...

AI Technical Summary

Problems solved by technology

[0004] However, in the past research methods based on physics and knowledge, the engine information cannot be fully utilized, and it is even more difficult to extract the main feature information for analysis.

However, most of the existing data-driven research is based on the analysis of a kind of sensor information, or part of the sensor data is discarded based on empirical values. This will not only lead to a large loss of sensor information, but also cause the risk of misjudgment, which will seriously affect the safe operation of the aircraft. potential risks

In addition, some data-driven algorithms cannot deeply mine the historical information between data, resulting in insufficient utilization of historical data, resulting in a series of problems such as low accuracy of prediction results

Images