内河AIS数据可靠性与修复研究

Real-time shipping traffic information is an essential part of the intelligent shipping system, which would be acquired by AIS (automatic identification system) normally. Due to the indeterminacy of AIS data transmission, it is necessary to study the reliability of AIS system, find a way to restore the data link, which would be great improvement on the inland maritime information administration..The application would take the DSmT method to filter the origin AIS data, which probably contain the GNSS and Static information errors. With the data pretreated, we do the statistics about the PER of AIS data link, further more, do the filed strength calculation over different distance, geography, interference, set up the prediction model for the AIS data transmitting. Meanwhile, for repairing the missing data, the application would adopt the high order Markov algorithm to create a model to restore the messages which are lost in the AIS link by using the motion characteristics of the vessels, and keep the model as simple as possible by the limit of the accuracy we need. Eventually, the application would set up the point to area field strength prediction model, describing large area which is plain, upland, urban, and different interference and weather. With the help of the model, it would build a simulation environment based on GIS, which would be used for the AIS base station location planning.

实时船舶交通信息是构建智能航运系统的关键，船舶自动识别系统（AIS）是获取内河船舶交通信息重要手段。针对AIS数据传输存在不确定性，研究AIS数据可靠性模型以及数据修复方法，对于获取完备的船舶交通信息具有重要意义。.本项目首先采用DSmT方法甄别AIS错误数据，并利用统计分析的方法大范围评估区域性的AIS数据掉包特征，进而逆推场强分布，同时采用偏微分方法描述各参数的特征，针对不同地理特征、天气、干扰，建立AIS信号衰减计算模型。然后根据船舶运动规律对数据进行修复，并建立高阶Markov修复模型，寻求近似简化算法，在保证修复精度的条件下有效降低算法复杂度。最后利用AIS信号点-面场强预测模型，描述大范围的平原、山区、城区、干扰源、天气因素下的AIS信号场强分布状况，建立基于地理信息系统的仿真环境，进而评估不同布设原则下的最佳AIS基站布设点。

本研究通过实地实验采集AIS信号场强，与简易地理模型匹配，改进传统基于经验的中值场强预测模型，建立包含地形特征的AIS基站信号场强点对面计算模型。根据船舶操纵性能，制定AIS错误数据清洗规则，然后基于船舶在航道中的位置、速度和方向，采用似然度建模方法，将这3个特征映射为可用于评估AIS数据可靠性的证据，最终采用证据推理算法建立AIS数据可靠性评估模型。以大量数据样本为基础，验证常用插值法在船舶运动轨迹数据修复方面的精度，得到满足精度要求的短距离丢失AIS数据快速修复方法。然后，针对长距离丢失的AIS数据，根据历史相似轨迹数据库建立基于最小二乘支持向量机（Least Squares Support Vector Machine，LSSVM）的数据修复方法。以均方误差和结点数据错位误差之和最小为优化目标，采用粒子群优化算法（Particle Swarm Optimization，PSO）搜寻合理模型参数。最后，搭建内河AIS通讯链路可视化评估平台，融合区域自动划分方法统计各区域的AIS数据丢包率，实现基于AIS数据分析面向海事监管的AIS基站覆盖范围可视化呈现，并在武汉水域展开实例应用。.本项目研究实现了AIS数据应用中的数据完整性和可用性问题以及AIS平台建设中的基站布设评估问题，为AIS数据应用研究提供坚实的基础，同时为海事AIS基站优化布局提供指导。