Розпізнавання сигналів випробувань вибухових речовин на стійкість до механічного впливу методом вейвлетного аналізу

Abstract

This work is devoted to one of the most important and complex problem of the digital signal processing – process recognition. Applied to the testing of explosives for resistance to mechanical impact, the researched method leads to the automatically objective deciding of the experiment result. This research is extremely urgent because of deep influence of accuracy and reliability of the tests to efficacy and safety of mine and underground construction building. The algorithm of explosion acoustic signal was developed in this work. It is based on using packet wavelet transform technique for spectral analysis and artificial neural networks for pattern recognition. Classification of signals is based on such artificial neural networks architectures as multilayer perceptron, radial-basis and probabilistic networks.