Abstract
Background
Table tennis serve is the only technical action fully controlled by athletes during competition. Its biomechanical characteristics directly determine the coherence of the first three strokes, the initiation of active offense, and overall match performance. Conventional serve training relies predominantly on coaches
’
empirical judgment and athletes
’
subjective perception, which cannot satisfy the growing demand for refined optimization of multi-dimensional biomechanical parameters including spin, speed and placement.
Objective
Existing studies remain limited in elucidating the coupling relationships among serve biomechanical parameters, establishing a comprehensive evaluation framework that balances competitive efficacy and motor executability, and deploying intelligent methods for individualized serve parameter optimization. This study aims to construct a data-driven multi-objective optimization framework for table tennis serve kinematic and performance parameters.
Methods
Five core kinematic and serve-control variables were selected as decision parameters: racket face angle, ball contact height, hitting timing, swing velocity, and landing coordinates. A multi-objective evaluation model was developed, integrating spin intensity, ball speed, placement stability, return suppression effect and motor adjustment cost. An improved particle swarm optimization (PSO) algorithm embedded with dynamic inertia weight, constraint correction, non-dominated sorting and crowding distance mechanisms was applied to generate Pareto optimal solution sets and screen athlete-specific serve schemes. The calibration dataset included 42 athletes (1,512 trials), and the validation dataset included 18 athletes (648 trials). A separate serve–return block involved 18 independent receivers, and individualized recommendations were evaluated in a 4-week pre–post intervention.
Results
The improved PSO-Pareto method outperformed conventional algorithms in convergence efficiency, solution set diversity and distribution uniformity. Six categories of Pareto-optimal serve schemes with distinct biomechanical and tactical orientations were identified. Individualized parameter recommendations showed descriptive improvements in overall serve quality, landing-point accuracy, return-suppression effect and third-stroke connection performance across different athlete types.
Conclusions
The proposed framework enables quantitative and individualized optimization of table tennis serve kinematic parameters based on measured movement and performance characteristics. It provides a methodological basis for data-driven and precision-oriented serve training, and offers a reference for biomechanical parameter optimization of other sport techniques.