Intelligent diagnostic method for airplane functional failure and system thereof

Abstract

This invention relates to one plane fault intelligent diagnose method and its system, which comprises the following steps: collecting data expression data and initial fault diagnose based on neural network; collecting initial fault results, message and rules expression fly parameters for secondary diagnose to output the result. the system comprises the following parts: dialogue module based on neural network to collect the data expression data and for initial dialogue based on the net; second dialogue based on blur special system for collecting fault message and initial dialogue result to output second dialogue output.

Description

technical field [0001] The invention relates to an aircraft fault intelligent diagnosis method and system, in particular to a fault diagnosis method and system for flight parameters and fault messages uploaded and downloaded in real time from the aircraft. Background technique [0002] At present, in the field of aircraft fault intelligent diagnosis, a single diagnostic technology such as expert system or neural network is generally used. This method has some disadvantages, such as the disadvantages of the expert system mainly include poor diagnostic accuracy and difficulty in overcoming the "bottleneck" problem of knowledge acquisition, while the neural network is only a process of data calculation in a sense, and we cannot accurately understand it. What exactly the neural network has learned, due to the lack of expert experience, it is impossible to correctly interpret the calculation results, and the neural network needs to be trained repeatedly for a long time and with a...

AI Technical Summary

Problems solved by technology

This method has some disadvantages, such as the disadvantages of the expert system mainly include poor diagnostic accuracy and difficulty in overcoming the "bottleneck" problem of knowledge acquisition, while the neural network is only a process of data calculation in a sense, and we cannot accurately understand it. What exactly the neural network has learned, due to the lack of expert experience, it is impossible to correctly interpret the calculation results, and it takes a long time and repeated training of the neural network with a large amount of data before it can have the diagnostic function

[0003] In particular, if the fault involves aircraft subsystems such as the engine system, due to its complex structure, using a single diagnostic technology will make it difficult to build a diagnostic system

For example, if only neural network is used for diagnosis, a large amount of data is required to train the neural network from fault symptoms to the lowest fault component directly; only expert system is used, which will increase the number of layers of system knowledge base construction and reasoning, which will It will bring problems such as infinite recursion and combinatorial explosion

[0004] In addition, currently in the field of aircraft fault intelligent diagnosis, only a single flight parameter data or fault message is used as the input data source of the diagnosis system

The disadvantage of a single data source is: the flight parameter data is limited and incomplete; fault messages will generate false alarms

Therefore, the accuracy of the diagnostic system using a single data source is low

Images