Variable cycle engine multi-fault diagnosis device based on self-association neural network

Abstract

The invention provides a variable cycle engine multi-fault diagnosis device based on self-association neural networks. The device comprises eight self-association neural networks, An n-dimensional vector m composed of sensor measurement parameters of the engine, the corresponding oil supply amount Wf and the opening degree msv of the mode selection valve MSV is input to the eight self-association neural networks, and the eight self-association neural networks output m1', m2',..., m8', comparing with the input m to obtain residual errors r1, r2,..., r8, and judging the fault condition of the engine according to the relationship between the residual errors and a set threshold value. According to the method, effective fault diagnosis and isolation can be carried out on the variable cycle engine, faults of a single component and faults of multiple sensors can be diagnosed and isolated at the same time, economic losses caused by flight stopping and other reasons can be effectively avoided, and replacement of some unnecessary components can be avoided; the high stability and reliability of the engine can be effectively guaranteed, safe work of the engine is guaranteed, the performance of the engine is fully played, and the safety and performance of an aircraft are improved.

Description

technical field [0001] The invention relates to the technical field of variable-cycle engine control, in particular to a multi-fault diagnostic device for variable-cycle engines based on an autoassociative neural network. Background technique [0002] Modern warfare requires advanced fighter jets to have the ability to cruise at subsonic speeds with a long range, and at the same time to have the ability to respond quickly during combat. In the future, variable cycle engines will continue to develop in three directions: long cruising range, high thrust-to-weight ratio, and wide operating range. By studying the speed characteristics of conventional engines, the researchers found that the turbojet engine has higher unit thrust and lower unit fuel consumption rate in the supersonic state, while the large bypass ratio turbofan engine has lower unit fuel consumption rate in the subsonic state . Considering the performance requirements of fighter propulsion systems in modern warfa...

AI Technical Summary

Problems solved by technology

At present, many fault diagnosis technologies can complete the diagnosis and isolation of faults, but they all need to be carried out in two or more steps.

Images