The Application Of Chaos To Underwater Signal Detection

The technology of underwater signal detection has significant theoretical and military values in our national defensive construction. In traditional detection methods underwater signal has always been processed as stochastic signal. With the deeper research and the high speed of the chaotic theory development, the applications of chaotic theory to signal detection has attracted more and more investments and some results have been obtained. On the basis of these results, the dissertation takes up the research of practical use of chaos on signal detection.With the signal detection as the main topic of this dissertation, chaotic signal processing methods are applied to the characteristic extraction of the ship-radiated noise. Utilizing the theory of state space reconstruction from chaotic series, research on how to reduce the noise and give prominence to chaotic feature is done. Further more the mechanism of the ship-radiated noise is simulated using the neural-network, which provides the essential premise for the practical use of underwater signal detection. The main endeavor and contributions of the dissertation are as follows:1.The chaotic theory is discussed systematically, which includes the chaotic signal processing methods and the characters of the chaotic system. After comparing the various calculating methods of the chaotic characteristic parameters, it is concluded that the strange attractor dimension and the Lyapunov exponent are main parameters that well describe the chaotic feature.2.On the basis of the theory of the state apace reconstruction, varieties of estimating methods of the characteristic parameters are discussed and the related parameters are classified and discussed too. In order to effectively reduce the add-in noise, PCA is introduced to realize the state space reconstruction.3.Three preprocessing algorithms are discussed to restrain the undesired noise, which consist of PCA, derivative coordinates and the algorithm based on the high-order statistics. The simulation results show that the PCA produces the best noise-cut effect.4.The chaotic feature of the ship-radiated noise is analyzed. Applying PCA to the sate space reconstruction, the strange attrator dimension and the Lyapunov exponent are estimated and compared with the correspondent result of the sea background noise. The conclusion shows that the correlation dimension provides the effective detection statistics.5.With the knowledge of the local predictability of the chaotic signal, the mechanism of the ship-radiated noise is simulated using the neural network trained by the BackPropagation algorithm. The result is the essential basis for the positive signal detection.6.The practical experiment of the target signal detection from the ship-radiated noise is done. Applying the trained neural network and the error detection algorithm the positive underwater signal detection is realized. The experiment verifies the effectiveness of the chaos applying to the signal detection.