Research On CEEMD And Ant Colony Optimization Applications In Ship Target Recognition Technology

The target identification technique is not only the prerequisite of true strike, stealthy attack under the advanced technology circumstance,but also one of the key technique to put intelligent at mines and torpedos, it is valuedby every advanced countries. This paper investigates the applications of Complementary Ensemble Empirical Model Decomposition,Multi-ayer autocorrelation, Ant Colony Optimization, and Neural Network to the feature extraction and classification of ship radiated noise. The simulation data and processing of real ship radiated noise validate the effectiveness of the algorithms. The investigation includes as folow,1. Explored the mechanism of ship radiated noise,on the basis of analysis the typical feature of ship radiated noise, the paper studies the mathematics model and implement method of line spectrum, broadband constant spectrum and noise modulation. Comparing and analysis the real ship noise, the simulation model proposed in this paper can simulate typical feature of specific ship radiated noise well.2. This paper makes systematical description of the Hilbert-Huang Transformation and decomposition the signal utilizing Complementary Ensemble Empirical Model Decomposition, it can overcame aliasing efficiently in the process of Empirical Mode Decomposition, detailed steps and parameter selection method are given.3. Investigated the theory of Multi-layer Autocorrelation, analysis the mechanism of Multi-layer Autocorrelation process in cancelling the uncorrelated additive noise, and applied it to the weak signal detection in comqplex background noise. Simulation and real noise processing results suggest Multi-layer Autocorrelation algorithm has a fairly good property in noise-restrain. The multi-layer power spectrum is much smoother and the line spectrum is more outstanding4. A new method to extract the modulation feature of ship noise is proposed. Combine the CEEMD decomposition and Multi-layer autocorrelation, reduce noise of ship noise,then extract modulation feature utilizing Hilbert demodulation method. Simulation and real noise processing validate this algorithm can extract the modulation feature efficient and highlight the line spectrum in low SNR and complex background noise.5. On the basis of deep study on the ant colony algorithm, a new algorithm for spectrum line detection and extraction based on ant colony optimization algorithm is proposesed in this paper. The algorithm ideas and processes are elaborated. In addition, the operator design and parameter selection method are put forward. The algorithm greatly improves the anti-noise ability to detect spectrum line.6. This paper studies ship radiated noise features extraction based on CEEMD and MFCC and elaborates the concept of Mel frequency cepstrum coefficients. The Mel frequency cepstrum coefficients are extracted as the target classification features of ship radiated noise signals. The definition of high order statistics and important properties are discussed in the paper. It obtained the high order statistical feature extraction of IMF component by decomposing the CEEMD,and regards it as the important features of the ship noise.7. The paper elaborates underwater acoustic target classifier design method. The ant colony optimization algorithm is applied to select the initial weights of neural networks and the threshold value. The parameters of neural networks are optimized in it. The paper verifies the effectiveness of feature extraction and classification by the processing and analysis of experimental data and classification results.