The Research On Ground-moving Target Recognition Based On Vibration Signal

With the constant progress of science and technology, energy and important chemical raw materials play an important role for the development of the society. In recent years, however, the important energy and materials theft all over the country occurs frequently, which caused serious economic loss, environmental damage, and even a great threat to people’s life safety. Therefore, it is important to discriminate the type of ground-moving target in the field of security and pre-warning.Ground-moving target recognition is an important signal processing task and technology. The ground-moving target signal is typical non-stationary signal. The time-frequency analysis method is an effective analysis method to the target signal. In recent years, the time-frequency analysis method of empirical mode decomposition(EMD) with many good qualities is widely applied to the feature extraction of vibration signals.The paper studies the principle which elaborates a variety of seismic waves in the stratum and the dissemination mechanism. It confirmed the method which makes use of the Rayleigh wave to detecting the ground-moving targets. The experiments were designed to gather several different ground-moving target signals. When the moving targets are different, the vibration signals will change at the same time. After these signals from the experiments are pre-processed and the targets’ features are extracted, the pattern recognition is done to identify the ground-moving targets.The algorithm development of EMD is still in its growth stage, and it needs further development and research to become more perfect. For modal aliasing effect of empirical mode decomposition and small sample characteristics of the practical signal, the paper put forward a feature extraction method based on Ensemble Empirical Mode Decomposition(EEMD) energy value. Then, it provides effective feature vectors to the Support Vector Machine(SVM) classifier established for ground-moving target recognition. The experiment result shows that, the average discrimination accuracy of proposed method is higher. In addition, compared with the method based on EMD and SVM, the proposed approach could acquire better classification result.For endpoint effect and negative frequency phenomenon of empirical mode decomposition method, local mean decomposition(LMD) is put forward to combine with approximate entropy, energy value for feature extraction of ground-moving target signal respectively. Then, the feature vectors are input into the SVM classifier for ground-moving target recognition. Verified by the analysis of experimental data, the proposed method is effective.