An Effective Model-based Thruster Failure Detection Method for Dynamically Positioned Ships

This paper presents an effective model-based thruster failure detection and isolation method for dynamically positioned (DP) offshore surface vessels. A DP vessel is supposed to maintain its position and heading at a reference point exclusively by means of thrusters. The occurrence of thruster failure may cause significant performance losses. Therefore, it is of great practical importance to timely detect and isolate thruster failures. In our proposed method, according to the prior knowledge of mathematical model of a DP ship, estimated model states can be obtained as reference. Wind disturbances, due to its great influence on the thruster diagnosis of the DP vessel, is taken into account. A new attitude based residual generator is designed. A failure can be identified once it exceeds a threshold. To further isolate the failure, a slide window concept together with a probability analysis is applied to the residual, until a concrete thrust failure is found. Simulation experiments of DP operation under different thruster failure cases are conducted in a professional simulator. The results show the proposed method is able to detect and isolate these thruster failures.