中文摘要：

根据战场环境复杂多变的特点,提出采用小波变换对目标声信号进行特征转换,用基于Daubechies小波和证据理论（即D-S证据理论）对基于多传感器的声目标进行融合识别。其步骤为：首先,针对Fourier分析在处理目标非平稳随机信号方面的不足,运用Daubechies小波变换对信号进行处理,即将256个数据为一组的采样信号在Daubechies小波第三尺度上进行变换处理,在保留信号的峰峰值位置、个数等原始特征的前提下,数据处理量由256个减少到32个,从而减少了后续数据的处理量和处理难度;其次,对经过Daubechies小波变换的数据采用FOBW编码进行特征提取,并建立常见声目标的特征信息库;最后,分析并研究数据融合在声目标探测识别中的应用。研究结果表明：与单一传感器识别和多传感器融合识别效果相比,采用D-S证据理论的声目标识别,系统的识别率提高,系统的误判率降低,达到甚至超过了预定的技术指标。

英文摘要：

According to the characteristics of complex battlefield environment,the wavelet transform was used to extract the acoustic target features,and D-S evidential theory was used to recognize the acoustic targets.The procedures were as follows： Firstly,aimed at the shortcoming of Fourier analyses,discrete wavelet transform（DWT） was used for signals processing.Using 256 data as a set of sample signals in the Daubechies wavelet transform processing on the third scale,while retaining the signal peak to peak position,the number of original features such as the premise,data processing capacity from 256 down to 32 to reduce the amount of follow up data processing and handling difficulties.Secondly,Daubechies wavelet transform was used to extract the acoustic target features of FOBW codes.The database of acoustic targets features was built.The target detecting and recognizing techniques were studied based on data fusion.Related theory and key techniques of D-S evidential theory were studied.Lastly,D-S evidential theory was used to recognize the targets.By comparing the simulation results of targets identification based on separate data and fusion data respectively,the advantages of this algorithm were testified.The results show that compared with the single sensor recognition and multi sensor recognition,the discrimination rate of system is improved,and the discrimination error reate is reduced using the D-S evidential theory.