| Recently , with the development of computer software and hardwaretechnology, Blind Source Separation——BSS has became a novel method in thesignal processing. Because it can reconstruct the original signal from the observed signal without any prior knowledge of the mixing system and source signal, BSS is a powerful tool to solve those problems ranging from wireless communication to medical signals processing, to image enhancement and to audio mixtures separation.First, this paper introduces the important mathematic concepts of BSS and expounds the corresponding physical meanings.Second, this paper introduces the basic principle of BSS. Through introducing four separation standards in the BSS: minimization of mutual information, maximization of entropy, maximum likelihood, non-Gaussian, we introduce two classical BSS algorithms: relative gradient algorithm and fixed-point algorithm. Through experiment contrasting, we conclude that the fixed-point algorithm is better than relative gradient algorithm in speed and reliability. At the same time, we introduce method of estimate in the unknown amount of source in usually signals separated. Through experiment, we prove the method that based on second-order statistics is feasible.Last, we research the special structure of longitudinal sensor array and the array speciality of detected signal and analyse the cause of disturbing between each sensor. In order to reduce the disturbing between each sensor, we tentatively use BSS method to process multi-channel signal and we design a multi-channel detected signal process system that based on BSS. Through MatLab software, we simulate the system on the PC. The result shows that this system can reduce the disturbing between each sensor and meets the detected equipment requirements.The research in this paper promotes the progress of BSS and expands the application scope of BSS in theory and reduces the disturb between each channeh improves the detecting equipment reliability in practice. |
PDF Full Text Request