Effective identification of ship behavior is crucial for the rational planning of port waters. However, the intricate ship trajectories and diverse behavior patterns within port waters significantly increase the difficulty and challenge of behavior identification. To address the issue of incomplete navigation status data, an unsupervised ship behavior identification method is proposed. The core of this method lies in the application and improvement of the DBSCAN principle: it replaces the traditional density threshold with average speed to define core points and subsequently refines threshold intervals based on ship behavior characteristics. This innovation achieves multi-mode synchronous clustering, enabling the simultaneous and accurate identification of various ship behaviors. This method demonstrates significant advantages in ship behavior identification, not only improving the accuracy of identification but also enhancing its practicality. More importantly, the ship behavior data obtained through identification allows us to delve deeper into the actual distribution of ship anchorages and berths within port waters. To verify the accuracy of the algorithm, we conducted a detailed comparison with actual conditions, fully demonstrating that the algorithm has high accuracy and provides strong support for the rational planning of port waters.