基于大数据平台的海上杂散目标识别模型

Recognition model of marine stray targets based on big data platform

针对杂散目标主要包括雷达杂波(非正常回波、雷达回波噪音)以及海浪等非正常目标,设计一种基于大数据平台的杂散目标识别模型。该模型利用目标点经纬度的变化、航速航向等信息,对目标点是否为杂散目标进行判断。杂散目标的判断不包括固定目标、漂浮物等,如何对多个雷达收集到的数据进行处理和判断,对识别的准确率、模型的有效性有着直接的影响。对AIS正常船航行的轨迹记录按时间进行截取,截取后的单轨迹时间不超过300 s。分别对处理后的来自AIS的数据和雷达数据进行特征处理,从而筛选拟合测试数据以及速度过小的数据。通过规则将数据打上标签并作为输入,利用决策树进行训练,通过对实验结果的验证判断实验对离散目标识别的效率及准确率。

Aiming at spurious targets including radar clutter(abnormal echo,radar echo noise) and abnormal targets such as ocean waves,a spurious target recognition model based on a big data platform is designed. The model uses information such as the latitude and longitude changes of the target point,speed and heading,etc.,to judge whether the target point is a stray target. The judgment of stray targets does not include fixed targets,floating objects,etc. How to process and judge the data collected by multiple radars has a direct relationship with the accuracy of recognition and the effectiveness of the model. The trajectory record of normal AIS sailing is intercepted by time,and the single trajectory time after interception does not exceed 300 seconds. The processed data from the AIS and the radar data are feature-processed separately,so as to screen the fitting test data and the data with too low speed. The labeled data is used as input through the rules,and the decision tree is used for training,and the efficiency and accuracy of the experiment on the discrete target recognition are judged through the verification of the experimental results.