Abstract: This paper designed a water fullness detection algorithm for train water tanks based on feature matching to solve the prominent problems in train water filling operations, including low intelligence of tank fullness detection, heavy reliance on manual operation and serious waste of water resources. It illustrated the full-water detection principle and control system, and adopted improved gradient detection and dynamic time warping technologies to accurately extract feature information from pressure curves under full-water conditions. The paper built an experimental platform by simulating actual on-site water filling operation environments and carried out experiments to explore the variation laws of water pressure before and after tank overflow in multiple complex scenarios. Experimental results show that the proposed algorithm is applicable to water tanks with various structures and capacities, and delivers favorable real-time performance and robustness. It effectively improves the automation degree and operation safety of train water filling operations.