| Adhesive structure in machine manufacture process is abroad applied in spaceflight , avigation and else industry field, because its weight is lighter and its shape is velvet. It can ensure the quality of production and prevent accident when the adhesive structures were tested effectively. The normal testing methods are not fit for testing on-line due to its low speed and pollution; while the traditional discriminating methods are complicated with mathematics pattern and have low rate of identification , so that it can not meets the needs of real-time production.Test technique of acoustic excitation presented in this paper can discriminate if the adhesive structure is intensional enough to endure certainty draw strength ,through a series of process ,for example bringing draw strength to bear on adhesive structure .testing signal through microphone array, choosing signal's character, recognizing automatically through manual nerve network, and so on. Thereinto ,Where the shoe pinches is how to choose character related to adhesive capacity from acoustic signal, and it is primary content in this paper ,too.Wavelet analysis is a newly developed theory, which overcomes the disadvantages of traditional Fourier analysis. It has good localized features in both the time domain and the frequency domain and has important application value in the fields of signal processing ,image processing ,speech analysis and so on. acoustic excitation signal is processed with wavelet analysis in this paper ,and chooses characters related to adhesive capacity from acoustic signal in the time domainand frequency domain. These characters is the input of nerve network which is used to non-mangle test about mechanics capacity of adhesive structure, and establish the base for classify distinguishing effectively and forecast.Experimentation result show that the method provided in this paper can choose effective characters from acoustic signal and establish the base for classify distinguishing effectively and forecast.
|
PDF Full Text Request