Optimizing configuration strategy of sensors in lubrication system monitoring of marine diesel engines offers significant benefits. It facilitates efficient acquisition of operational status data and enables accurate analysis of system characteristics. These improvements deliver reliable data support for maintenance research and engineering applications. A comprehensive evaluation index for optimizing configuration strategy of sensor was constructed based on fault mode recognition，sensor failure tolerance，and sensor quantity. A fault mode correlation matrix was also constructed based on the improved grey relational analysis(IGRA) method，and then the evaluation index was calculated by using rough set theory. A Pareto-optimal sensor configuration was finally achieved by using the non-dominated sorting genetic algorithm Ⅱ(NSGA-Ⅱ). Experimental results demonstrate that the initial configuration with nine sensors monitoring temperature and pressure parameters，including lubricating oil pump，filters，main engine oil inlet，and oil cooler can be optimized to five sensors while maintaining equivalent fault identification capabilities. The optimized configuration exhibits a robust fault tolerance，sustaining a fault isolation rate from 81.14% to 96.00% during single-sensor failure scenarios.