Research On The Vaccum Leakage On Line Detection Method Based On Acoustic Emisson

With the development of space science and technology, space activities are increasing. At the same time, the amount of space debris has increased rapidly,which makes the probability of impact are rising between in-orbit spacecraft and the kinetic energy of particles. Spacecraft leak will not only cause the failure of space activities, but also endanger the life safety of the astronauts. It is great significance for in-orbit spacecraft vacuum on-line detection research. Vacuum leak is common failure in gas dynamics system, however, a weak leakage is a potential safety hazard easily to ignore. So how to effectively and quantitative detect the weak leakage signal is the key point of this study.In this paper, we use acoustic emission testing technology combined with BP neural network method to detect several weak leakage signals. The main research contents are as follows:(1) Build experiment platform of simulation in-orbit spacecraft vacuum leak,and use sensors to collect different kind of leakage signals by acoustic emission detection method.(2) Use acoustic emission parameter analysis method and acoustic emission waveform analysis to extract the leakage signal feature. In the acoustic emission waveform analysis, the characteristic processing of the leakage signal by the methods of average power spectrum and wavelet packet energy spectrum in the frequency domain and time domain are discussed.(3) Build the BP neural network, and use the acoustic emission parameter analysis method, the average power spectrum method, the wavelet packet energy spectrum method and the method of multi-parameter fusion to extract the feature parameters. Then sent them into neural network to pattern recognition. So we can compare the four methods of recognition effect.(4) Make chaos optimization of BP neural network and apply in weak signal detection in the vacuum leak. Compare with the standardized BP algorithm and adaptive BP algorithm on convergence curve. Chaotic BP algorithm can effectively speed up the convergence. Compare with sampling signals of different distance between the sensor and leak, and analyze the result of recognition of different distance between sensor and leak.