On intelligent risk analysis and critical decision of underwater robotic vehicle

Abstract

The marine community has witnessed a remarkable growth of underwater robotic vehicles (URVs) for undersea exploration and exploitation in recent decades. Yet, it is critical to intelligently diagnose the fault and evaluate the risk of the onboard system, and render critical decision to ensure the safety of the URV with high-value assets. In this paper, a dedicated two-layer fault treatment system including risk analysis subsystem and intelligent decision subsystem is proposed to enhance the onboard safety of the URV. First, a hierarchical fault tree model of the URV is built by integrating the state information of sensors, actuators and running status. Second, in the risk analysis subsystem, the onboard system risk is analyzed based on the adaptive learning and fuzzy inference capabilities of the Mamdani fuzzy neural network (MFNN). Third, in the safety decision subsystem, the risk level of the URV is evaluated by adopting the maximum membership and threshold principles, which enables the intelligent decision to take critical operation and ensure the safety of the URV. Finally, the proposed fault treatment system is validated by numerical simulation and hardware in loop test. Experimental results demonstrate the feasibility and efficiency of the intelligent fault treatment system for the URV.