<p>This paper presents the design and innovation of fuzzy control algorithms in an underwater weed localization system, which aims to achieve accurate underwater weed identification and localization by integrating sensor arrays, data processing centers, fuzzy logic controllers, and actuation units. The system model combines optical imaging, acoustic sensing, and underwater robot dynamics control and employs a deep fuzzification layer with reinforcement learning algorithms to optimize the fuzzy inference process, as well as particle swarm optimization techniques to dynamically adjust the fuzzy parameters. The experimental evaluation utilizes a comprehensive dataset to verify the localization accuracy and robustness of the system in complex underwater environments. Through model training and environmental impact testing, the system demonstrates excellent performance and adaptability, providing an advanced technical solution for underwater ecological management and aquatic environment monitoring.</p>