Multimodal Data Fusion and Adaptive Optimization in Tennis Training Based on Deep Deterministic Policy Gradient and IoT Sensors

Abstract

This paper proposes a novel framework integrating IoT technologies, multimodal sensor networks, and the Deep Deterministic Policy Gradient (DDPG) algorithm for intelligent tennis training. We employ the DDPG algorithm for adaptive training adjustments, which dynamically optimizes the training policy based on real-time feedback. Experimental evaluation on 10 athletes shows that the DDPG algorithm improves performance metrics in multiple training scenarios, increasing the average game score from 50 to 80 points and reducing the error rate in high-pressure scenarios from 13% to 6%. The system’s success rate reached 85%, with swing stability enhanced by 27% (0.1 rad deviation). These quantifiable outcomes highlight the framework’s effectiveness in optimizing training strategies, with potential applications in industrial automation and healthcare monitoring.