参数自适应动态贝叶斯舰船态势估计算法

为减少动态海域下贝叶斯网络舰船态势估计算法较大的误差,提出了一种改进的动态贝叶斯网络舰船态势估计算法。该算法根据多路传感器的数据和新获取的态势信息进行推理,通过计算新态势要素与原有态势要素间的互信息构建并更新动态贝叶斯网络参数。与传统贝叶斯网络态势估计算法对比,在仿真情况下对10000艘舰船进行合作态势估计,改进动态贝叶斯网络的舰船态势估计算法合作舰船错误率降低了7.1%;用实测数据,目标的合作态势提升了4.2%。改进的算法不仅能够实时地反映舰船环境变化,同时还提高了目标态势估计的准确概率,为海监提供了一种舰船态势分析与决策的技术支持。

Algorithm for the assessment of ship situation based on the parameter adaptive dynamic Bayesian network

In order to reduce the error of the Bayesian network algorithm for the assessment of ship situation in the dynamic area of the ocean, an improved algorithm for the assessment of the ship situation is proposed based on the dynamic Bayesian network. The algorithm makes an inference based on the data from multiple sensors and newly acquired situation information. By calculating the mutual information between new situational elements and original situational elements, the dynamic Bayesian network parameters are constructed and updated. Compared with the model of the traditional Bayesian Network, the error rate of the cooperative target of the ship reduces by 7.1% through simulation of about 10,000 ships. By using the improved dynamic Bayesian network algorithm for the assessment of ship situation, under the measured data, the cooperation of situation for the target has increased by 4.2%. The algorithm proposed in this paper not only reflects the environment of ship changes in real time, but also improves the accuracy of the target situation, thus providing a technical support for analysis and decision-making of the situation of ships for Marine Surveillance.