Model-based aircraft system fault positioning method

Abstract

The invention belongs to the field of aircraft complex system fault diagnosis design, and particularly relates to a method for achieving LRU-level rapid fault positioning of an aircraft system. The aircraft system fault diagnosis method based on the model is provided and used for rapidly achieving LRU-level fault positioning after an outfield aircraft system breaks down, so the replacement of a fault unit is rapidly completed, and finally the integrity rate and the attendance rate of an aircraft are increased.

Description

technical field [0001] The invention belongs to the field of aircraft complex system fault diagnosis design, and in particular relates to a method for realizing LRU-level fast fault location for aircraft systems. Background technique [0002] At present, the existing aircraft system fault diagnosis methods mainly include: rule-based reasoning, artificial neural network-based and case-based methods, etc. There is a knowledge acquisition bottleneck in the rule-based diagnosis method. The connection chain between the symptoms observed by the system and the corresponding faults is often strengthened with the increase of the complexity of the system. The rule of expert experience is often not unique, and There is considerable difficulty. Neural network-based diagnostic methods usually require more training examples for neural network learning in order to make the network converge and obtain stable diagnostic results, and cannot make reasonable explanations for the diagnostic res...

AI Technical Summary

Problems solved by technology

There is a knowledge acquisition bottleneck in the rule-based diagnosis method. The connection chain between the symptoms observed by the system and the corresponding faults is often strengthened with the increase of the complexity of the system. The rule of expert experience is often not unique, and quite difficult

Neural network-based diagnostic methods usually require more training examples for neural network learning in order to make the network converge and obtain stable diagnostic results, and cannot make a reasonable explanation for the diagnostic results

The case-based diagnosis method also has limitations. Usually, it is very time-consuming to retrieve cases from a large case base, and it is difficult to decide which symptoms and their weights should be selected. It is difficult to deal with the consistency test when the case is revised. The acquisition of case knowledge is still a difficult problem in the process of

Images