The Research Of Infrared Temperature Model Of Airborne Electronic Board Components

With the rapid development of the global and civil aviation industry, the structure of airborne electronic equipment system is becoming more and more complex. However, the domestic aviation enterprise is only able to perform maintenance on board level, which is inefficient and costly. Therefore, improving the traditional fault detect scheme becomes a must.This research is based on practical applications. First of all, it discusses the heat transfer modes and fault types of different kinds of components, analyses and compares two kinds of image processing methods: the method based on differential infrared images and the one based on sequence infrared images, and then arrives at a plan to collect components’ infrared temperature image. Next, the research analyzes several border detection methods of infrared image segmentation and simulates them respectively, compares their advantages and disadvantages, then expounds the whole process of target recognition. It then puts forward a new feature vector combining invariant moments and infrared characteristics to identify infrared temperature image of components of airborne electronic boards. By recognizing the infrared image of components in the work phase, the research establishes a multi-signal temperature model of components, and studies modeling steps.Finally, the research discusses an analyzing method of the correlation matrix based on the multi-signal temperature model. Aiming at the deficiency of the correlation matrix analysis, it proposes a method of fault diagnosis based on bayesian network analysis. Using a certain type of airborne electronic board as an example, it establishes a multi-signal temperature model, studies the model elements and modeling process in detail, and conducts the correlation matrix analysis and the analysis based on the bayesian network. The analysis results of the two methods show that the bayesian network analysis method is more practical than the correlation matrix analysis, because it presents detailed quantitative indicators of the fault diagnosis of components whereas the correlation matrix analysis does not. This has a great practical value to guiding the fault diagnosis of airborne electronic boards on a component level.