Abstract: Cockpit sealing system of aircraft is an important air-tight system to ensure safety of crew in high-altitude flight, which should be carried out regularly or emergency quantitative detection of gas leakage. Because of high error and leakage detection rate and complex noise environment, it is difficult to realize quantitative detection and accurately identify gas leakage signal. Based on principle of ultrasonic detection and characteristics of small hole leakage of cockpit sealing system, according to corresponding relationship between sound pressure, sound intensity of ultrasonic signals and gas leakage at gas leakage hole, quantitative detection model of cockpit sealing system was derived. An improved chaotic Duffing oscillator equation was proposed for detection of gas leakage signals in presence of strong background noise. By comparing chaotic threshold with standard deviation of noise and maximum Lyapunov index, and analyzing state change of system's phase trajectory, as well as developing aircraft cockpit gas leakage detection example analysis, this method was proved to be more suitable for identification of gas leakage signals of aircraft cockpit sealing system.