An Algorithm for Mining Track Law Based on Adaptive Clustering

In many sailing trajectories, the curve can be approximated, and the target often has a certain law of activity. So how to use unsupervised learning to mine the law of the target trajectory in the route is of great significance to situational awareness. More importantly, the law of the mining activity should be suitable for different scenarios, so the adaptiveness of the target trajectory mining algorithm should be considered. In this regard, firstly, it is based on the distance formula between the vectors to consider the similarity between the tracks, and the multi-dimensional distance calculation method is used to calculate the distance relationship between the tracks. The multi-dimensional calculation method uses longitude, latitude, altitude, and speed. The distance calculation method of attributes such as, heading, etc. under different influence factors, and then use the improved density-based clustering algorithm to mine the law of the route, and finally display the mined fitting track in the interface to represent the characteristic track. The experimental results show that the mining effect of the adaptive clustering mining algorithm is better than the fixed value of the parameters. The track accuracy and recall rate that can be mined by the adaptive method are significantly higher than the previous algorithms, and it can be extended to different scenarios.