Research Progress on Ship Anomaly Detection Based on Big Data

The purpose of ship behavior anomaly detection is to identify and monitor some non-expected behaviors of ships, so as to improve the navigation safety of ships. Its research is of great significance to the safety guarantee of maritime navigation, intelligent monitoring of sea areas and the development of port management. This paper summarizes and evaluates the research progress of ship anomaly detection based on big data and points out the future development trend. First of all, the concept of ship abnormal behavior is introduced, and the process of data-driven ship abnormal detection and its data basis are described in detail. Secondly, the data-driven ship anomaly detection methods are divided into statistical method, machine learning method and neural network method, and their research status and existing problems are reviewed respectively. Finally, focusing on maritime big data, temporal and spatial correlation of scenarios, online real-time anomaly detection and other aspects, the current problems and challenges in the study of ship anomaly detection are discussed, and the future research direction is introduced.