Analysis & fault diagnosis of cockpit voice signals based on information fusion

Abstract

Aircraft cockpit voice signals recorded by aircraft black boxes are one of the key evidences to analyze and diagnose the flight faults. Firstly, the paper describes information fusion principle, and constructs three-level information fusion model. Then, many typical cockpit voice signals as examples (such as wind shear, ground proximity warning, take off form warning, fire alarm, background sound of stick shaker, over speed warning, background ground of high speed with translation, and so on), their characteristic warehouse have been set up. Thirdly, the new analysis and diagnosis methods for cockpit voice signals has been put forward according to production rule and information fusion principle. Finally, binary system diagnostic trees on cockpit voice signals are formed and some analysis and diagnosis results are obtained.