To evaluate and improve the reliability of an electric propulsion ship power grid and optimize its operation strategy, this paper proposes an online power quality assessment method based on grey cloud clustering and evidence reasoning. First, an improved group analytic hierarchy process and variable weight theory are used to obtain the index weights. The grey cloud clustering model is used to transform the evaluation indices into the basic reliability distribution of the corresponding evaluation level. Then, based on historical information, an improved conflict measurement method is used to measure the reliability of the indices. Finally, through evidence reasoning rules, information of the time domain and the airspace corrected by reliability and weight are combined to obtain real-time evaluation results of the system. The simulation analysis shows that the method can accurately reflect the real-time operating state of the ship's power grid and has strong anti-interference ability. This work is supported by National Natural Science Foundation of China (No. 61503240) and Innovation Foundation for Graduates of Shanghai Maritime University (No. 2016ycx078).