Application of recurrent neural network in predicting athletes' sports achievement

Abstract

The application of recurrent neural network (RNN) is studied in athletes' sports achievement prediction under the internet of things (IoT) environment. Concretely, the 3000 m steeplechase and athletes’ corresponding achievements are analyzed using RNN. Then, the athletes' achievement prediction model for 3000 m steeplechase is established through different algorithms, in which the IoT technology is used for prediction analysis of the relationship between athletes’ physical parameters and achievement. The experimental analysis finds that when the athletes' morning pulse is relatively stable, the achievement is better, while when the athletes' morning pulse fluctuates greatly, the achievement is not ideal. Meanwhile, when the athletes' oxyhemoglobin saturation is high, they will achieve better results, while when oxyhemoglobin saturation decreases, athletes' performance will decline significantly. Moreover, when the athletes' oxyhemoglobin saturation increases gradually, the athletes' performance will also be improved. At the same time, when athletes' blood pressure is abnormal, the training methods and training status should be adjusted reasonably, and the rest and training time should be rearranged to avoid long-term high-intensity training. There is no significant correlation between athletes' achievement and weight, but life experience tells that people with normal weight may have a better physical condition. Finally, under comparative analysis, the long short-term memory neural network prediction model can more accurately predict the achievement of athletes of 3000 m steeplechase. The results provide important theoretical support for the training plan of athletes, and an effective scheme for the prediction of 3000 m steeplechase achievements of athletes.