Design of intelligent optimization of sports strategy and training decision support system based on deep reinforcement learning
Abstract Existing training decision support systems mostly rely on preset rules or frameworks based on prior knowledge. When faced with significant individual differences or rapid changes in the environment, it is difficult for the system to dynamically adapt to the physiological conditions, skill l...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-08-01
|
| Series: | Discover Artificial Intelligence |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44163-025-00473-9 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Existing training decision support systems mostly rely on preset rules or frameworks based on prior knowledge. When faced with significant individual differences or rapid changes in the environment, it is difficult for the system to dynamically adapt to the physiological conditions, skill levels, and training progress of different athletes. To achieve efficient and personalized training support for athletes, this article constructs a new training decision support system that combines DRL (Deep Reinforcement Learning) technology. Through multi-sensor fusion technology, the system comprehensively collects athletes’ physiological and external environmental data. The data is preprocessed by a sliding window average filter algorithm to eliminate noise and outliers. The system adopts the DQN (Deep Q-Network) architecture and applies dual DQN technology to improve model stability. The priority experience replay strategy is used to accelerate model convergence and improve training efficiency. The system combines real-time strategy adjustment and optimization modules to dynamically adjust training strategies according to athlete performance to meet individual needs. The system feedback mechanism continuously optimizes the training plan according to athlete performance to ensure long-term effectiveness and personalized services. The research results show that the average response time of the DRL-based system in the initialization and data loading test scenarios is 13.21 ms, and the average accuracy, consistency, and failure rate of the system are 96.31%, 95.0%, and 2.37%, respectively. This shows that the system response time period and long-term operation stability are better, and it can provide personalized suggestions, thereby significantly improving training efficiency and providing scientific support for athletes. |
|---|---|
| ISSN: | 2731-0809 |