中文摘要：

船舶自动识别系统（Automatic Identification System,AIS）数据链缺乏保障机制,常报告错误信息,干扰船舶导航。在实际使用中,需对这些信息予以剔除。作者借鉴人工识别过程,建立基于证据理论的AIS错误目标自动识别算法。首先,采用概率推理、专家评估和模糊数学方法,分别对船舶航速、航向和轨迹特征进行建模,获得船舶航速、航向和轨迹位置的信度分布函数。然后,利用PCR6证据合成规则对这些特征的信度分布函数予以融合,综合评判AIS目标的真实性。通过与传统Dempster规则的合成结果对比,发现对于路过识别水域的客货船的AIS报文,PCR6的识别准确率高于Dempster规则,对于巡逻艇类船舶的AIS信息,PCR6和Dempster规则的识别准确率下滑,但总体上依旧可靠。

英文摘要：

Due to lack of data-link guarantee mechanism,sometimes AIS will broadcast error messages and distract attention of navigators.Therefore,it is necessary to remove these error messages in practical application.In this research,an error AIS targets identification algorithm based on the evidence theory was proposed by using artificial identification process.Firstly,probabilistic inference,expert assessment and fuzzy mathematics were utilized to do modeling the velocity,course and track characteristics of ships,respectively,and belief distribution functions of velocity,course and track of ships were acquired.Then,the belief distributions were combined with PCR6 evidence synthesis rule,to evaluate the authenticity of AIS targets.By comparing the results of PCR6 and Dempster＇s rule,it is found that PCR6 performs better than Dempster＇s rule when identifying passing cargo/passenger ships.Their accuracy both decreased when identifying patrol ships and ferries,but it still reliable generally.