Optimization Techniques With Application To Underwater Acoustic Signal Processing

The dissertation investigated Simulated Annealing (SA) and Genetic Algorithm (GA) for underwater signal processing. Due to their recently shown remarkable performance,SA and GA are receiving considerable attention. The application of SA and GA to underwater acoustic signal processing is considered in details here. The applications include underwater acoustic channel estimation,de-convolution,target tracking and inverse problems of underwater sound field. It is shown that SA and GA have the capability of solving the specific problems. SA and GA perform fairly well for the problems in conjunction with combinatorial optimization.The characteristics of SA and GA are analyzed intensively,which involves search mode,the mathematical analysis and experimental aspects of the underlying processes in the algorithms,convergence discussion and detailed operation procedures. Then the modified versions are presented. These new algorithms are MOSSA (Multiple Operation Structure SA),NGA (Normalized GA) and HGA (Hybrid GA). The algorithms are evaluated by test function methods. Compared with the CSA (Classical SA),FSA (Fast SA) and LGA,the new versions of the algorithms seem to gain better performance.There are several application examples investigated and presented in the dissertation. De-convolution techniques based on MOSSA is shown in Chapter 7,underwater acoustic channel estimation and data recovery in Chapter 8,and the inversion problems in underwater sound field in Chapter 9. Chapter 10 summarizes the investigation and gives the major conclusions.