Abnormal Behavior Detection Of Fishing Boats Based On AIS Data

Fishing vessel plays a crucial role as the primary tool in fisheries production and plays an essential role in the development of the fishing industry.In recent years,with the growth of fishing operations and the number of fishing vessels,along with the increasing freedom in vessel navigation and the uncertainty of their routes,illegal transportation and unauthorized fishing activities by fishing vessels have become more frequent.Currently,the analysis of fishing vessel AIS trajectory data allows for the identification of vessel navigation characteristics and behavioral patterns,which is significant in reducing the occurrence of maritime accidents and detecting abnormal fishing vessel behavior.This paper aimed to address the issue of detecting abnormal fishing vessel behavior and specifically focused on the detection of "Ship Identity Spoofing," where a vessel uses false identity information to impersonate another vessel’s identification code.Based on AIS data,a method for detecting abnormal fishing vessel behavior was proposed.After performing relevant preprocessing steps on the fishing vessel trajectory data,the trajectory data was compressed.Subsequently,the abnormal trajectories were detected,and the identified abnormal behaviors were analyzed.The main research work included.(1)Data preprocessing.Its purpose was to improve the quality and accuracy of the raw AIS trajectory data,addressing issues such as data missing,outliers,and inconsistent formats.It provided an accurate and reliable data foundation for subsequent abnormal behavior detection.(2)The improved Auto-Douglas-Peucker method(ADP),based on the Douglas-Peucker algorithm,was proposed for compressing fishery vessel trajectory data.It addressed the issue of distance threshold setting in existing fishery vessel trajectory compression algorithms,which relied on empirically set thresholds.First,the turning point set was determined based on the difference in heading between trajectory points,preserving the temporal sequence of the trajectory and compressing redundant points.Second,a segmented compression approach was applied to the turning point set,extracting feature points from each sub-trajectory segment and constructing a feature point set.Finally,the feature point set was fitted using a curve-fitting method,and the threshold was adaptively adjusted based on the angle change rate between adjacent feature points to accommodate different trajectory data characteristics.Experimental results demonstrated that the ADP algorithm achieved an average compression rate of 96%.Although slightly lower than the average compression rate of the DP algorithm,it had greater advantages in preserving trajectory features.The compressed trajectories were closer to the original trajectories,facilitating accurate analysis of fishery vessel abnormal behavior.(3)A fishery vessel abnormal behavior detection method based on improved LCSS(Longest Common Subsequence)was proposed.It aimed to address the issue of "ship identity spoofing" in detecting fishery vessel abnormal behavior.Firstly,trajectories with the same time span were selected to eliminate the influence of differences in trajectory starting points on similarity assessment.Secondly,the trajectories were segmented,and the distance calculation was improved from point-to-point distance to segment-to-segment distance,enhancing computational efficiency.Vertical,angular,and centroid projection distances were introduced in the distance calculation.Thirdly,the similarity between trajectories was calculated by solving the longest common subsequence problem.Finally,the trajectories within clustered clusters were discriminated to determine whether they were abnormal trajectories.Compressed fishery vessel AIS trajectory data was used in the experiments.The results showed that the method achieved satisfactory performance in detecting fishery vessel abnormal behavior by identifying abnormal trajectories within clusters,effectively addressing the issue of "Ship Identity Spoofing".