Aero-engine sensor fault diagnosing and isolating method based on fuzzy membership degree

Abstract

The invention discloses an aero-engine sensor fault diagnosing and isolating method based on a fuzzy membership degree. For the defects in the prior art, a solving idea of the aero-engine sensor faultdiagnosing and isolating method is that analytic redundancy and hardware redundancy are fused based on fuzzy logic and a fault threshold value; sensor fault analysis is performed based on a fused signal; once a signal membership degree is lower than an average value, a proportion of a signal of the signal membership degree in the fused signal decreases; after the signal membership degree exceedsthe fault threshold value, the signal is completely expelled out of the fused signal; thus, self-adaption fault tolerant and fault isolation are achieved. Compared with the prior art, the aero-enginesensor fault diagnosing and isolating method disclosed by the invention can achieve self-adaption fault tolerant and fault isolation under the single channel situation of a dual-channel sensor.

Description

technical field [0001] The invention belongs to the field of system control and simulation in aerospace propulsion theory and engineering, and specifically relates to a fault diagnosis and isolation method for aeroengine sensors based on fuzzy membership. Background technique [0002] In aero-engine control systems, sensor failures account for more than 80% of total failures. In order to improve the safety and reliability of the engine control system, sensor fault diagnosis technology has been widely concerned, and has developed into an interdisciplinary applied discipline in the past 40 years. [0003] For the key sensors of the engine control system, multi-channel and multi-sensor hardware redundancy technology is often used. For sensors with three hardware redundancy and above, the election method is often used to vote on whether the sensor fails. If the deviation exceeds the threshold, it is impossible to determine which sensor has failed. However, multiple hardware re...

AI Technical Summary

Problems solved by technology

However, the engine has a large flight envelope and variable work. The offline training neural network is difficult to adapt to the uncertainty of the engine work, while the online training neural network has a complex network structure, many inpu

Images