Abstract: The unmanned surface vessel is a small surface warship with autonomous planning and autonomous navigation. It is often used to carry out different kinds of dangerous battle missions by many countries in the military field. In order to ensure the safety and reliability of surface vessels at work, this paper discusses the advantages of applying data-driven methods to the detection and diagnosis on the navigation states of surface vessels, analyzes the application performance of PCA-based, SVM-based and BP-neural-networkbased methods, and proposes practical methods for detection and diagnosis, which provide references to researchers devoted to the related work.