The Application Of Hilbert-Huang Tansform In Extraction Of Mine Characteristics

In the detection and the recognition of mine, which method can extract the characteristics information of echo effectively is the key. Because of the lack of time-frequency analysis methods exist, in this paper, as a new method of time-frequency analysis, the Hilbert-Huang Transform (HHT) has been applied to the target characteristics extraction. Feasibility study on this method is the focus of this paper.This paper dissertates the idea of HHT and how to realize it in detail, discusses related issues, and improves the algorithm through the simulation of its algorithm. This paper proves the local characteristics of HHT through contrasting this method with Wavelet Transform; then it researches anti-noise and anti-reverberation ability of HHT by simulation, and provides the value of SNR and SRR which can detect signal effectively. Then, this paper applies HHT to extract the mine characteristics, basing on the simulation of highlight model of mine. It focuses the discussion on the resolution of time delay and magnitude of two highlights, and what condition is satisfied by the time delay and the relative magnitude, HHT can extract the structure of the highlights effectively.At the end of this paper, having HHT application in a lot of experimental data, the validity of this method in dealing with the actual mine target is discussed. First of all, this paper extracts the time-frequency characteristics of different targets echo by HHT, then, classifies the true or false target and the different geometry of real targets by BP neural network classifier, finally, discusses whether the experimental results fit with the theoretical discussion. The results show that this method can extract the highlight structure effectively, and achieve better results of classification and recognition.