连续隐马尔科夫模型在多基地目标识别中的应用

Multi-static underwater target recognition method based on continuous hidden Markov model

多基地声纳组网探测系统是目前大范围水下安保领域的研究热点。综合利用多基地系统中各个声纳节点的信息进行水下目标识别是亟待解决的问题。利用传统的多传感器融合的方法进行多基地水下目标识别,往往忽略了各声纳节点之间的相关性,效果并不理想。针对这一问题,本文提出了利用连续隐马尔科夫模型(CHMM)进行多基地水下目标识别的方法。首先利用RELAX算法提取了目标在不同分置角上回波的强散射点特征,组成观测向量,利用Baum-Welch方法对CHMM参数进行训练,然后计算待识别目标的特征值观测序列在不同模型下的似然概率。对所有目标重复此过程,取概率最大值对应的目标类别为最后的识别结果。在消声水池开展多基地模拟实验,对四类目标进行了识别,利用CHMM方法得到的多基地水下目标融合识别率比多基地声纳下单声纳节点的最高识别率提高了30%。

Netted multi-static sonar detection system is a research hotspot in the wide range of underwater security. How to utilize each sonar node in multi-static sonar system comprehensively is a problem demanding prompt solution. The method using traditional multi-sensor fusion neglects correlation between adjacent sonars,and the result is unsatisfactory. To solve this problem, multi-static underwater target recognition method based on continuous hidden Markov model(CHMM) is proposed in this paper. The strong scattering points features obtained by RELAX as recognition features from different sonars are combined as observation vectors. BaumWelch method is used to train CHMM parameters. The likelihood probability of the observation sequence of test data in different model is calculated, and the target type corresponding to the maximum value is the recognition result. Multi-static simulation experiment is conducted in anechoic tank, and the fusion recognition rate using CHMM method is 30% above the maximum recognition rate of single sonar node of multi-static system.