The information obtained through forced detection is often inaccurate, and targets frequently change course unpredictably. This degrades the performance of target maneuver detection and hampers the analysis of the target motion pattern. Therefore, this paper proposes a detection method for maneuvering maritime targets based on prior knowledge.

The method incorporates two types of prior knowledge derived from expert experience. The first is that significant differences in target heading occur before and after maneuvering, whereas the target heading remains relatively stable during non-maneuvering periods. The second is that the heading difference before and after maneuvering reaches a local extremum. The maneuvering point in the trajectory tends to maximize the heading difference between adjacent sub-trajectories. Based on the definition of trajectory smoothness metric, a calculation method is proposed to calculate the course maneuver evaluation factor based on principal component analysis (PCA). This factor enables preliminary screening of potential maneuvering points. In order to find trajectory points that satisfy the second prior knowledge, a maximum filtering-based maneuvering point screening method is proposed.

Simulation results show that, compared with the mainstream interactive multiple model (IMM) algorithm and information entropy-based algorithm, the target maneuver inflection points detected by the proposed method are closer to the actual inflection points, with the lowest false detection rate and missed detection rate. Moreover, when track compression is performed using the maneuver positions extracted by this method, the distance error relative to the original track is minimized.

The findings confirm the superiority of the proposed algorithm, which can effectively improve the accuracy and robustness of target maneuver detection and provide strong support for target behavior analysis and operational decision-making at sea.