Big Data and AI-Driven Optimization of Action Analysis and Training System for Table Tennis Players

Aiming to meet the demands for fine and personalized training in table tennis, this study discusses how to use big data and AI technologies to optimize the motion analysis and training system of table tennis players. The sensor fusion model, combined with Kalman filter technology, enables effective integration of multi-source data, while key motion features are extracted through principal component analysis. On this basis, a motion recognition model is developed using Long Short-Term Memory (LSTM) networks and the attention mechanism, enabling accurate recognition and classification of table tennis players' movements. Meanwhile, a personalized training model is established using machine learning algorithms, with training intensity dynamically adjusted through multi-objective optimization and reinforcement learning strategies to ensure training efficiency and safety. The system adopts modular and hierarchical architecture and combines Hadoop/HDFS, Spark, TensorFlow, and Kubernetes to realize data storage, real-time analysis, and model deployment. Experiments show that the experimental group using this system is significantly better than the control group regarding movement stability, physiological load balance and competition results, with the movement stability improved by 34.4% and the simulated competition winning rate increased by 18.2%. In addition, the system has good real-time performance, the average delay is 150ms, and 95% of the response delay is less than 185ms, which meets the real-time requirements. Results conclude that the integration of big data and AI technologies enables the creation of a more scientific, accurate, and personalized training system for table tennis players, significantly enhancing training efficiency and competitive performance.