Analysis Method of Special Physical Training Mode of Basketball Teams in Colleges Based on WeChat Applet and FTTA Optimised LSTM

OBJECTIVE: this paper proposes a basketball special physical training mode analysis method based on WeChat applet and optimization algorithm to improve the deep learning network. METHODS: Using the applet data set and the coaches' record data as model input data, the proposed method is used to analyse and thus improve the performance of the basketball team's special physical training pattern. RESULTS: Comparing the analysis effect between FTTA-Attention-LSTM analysis model and LSTM, FTTA-LSTM, FTTA-GRU, FTTA-BiLSTM models, the WeChat mini-program oriented basketball team's special physical fitness training mode analysis index system contains 14 factors affecting the analysis model; in analysing the relationship between the size of FTTA population and Attention-LSTM network hidden layer node number experiments, it was found that the selection of the population size of 80, the number of hidden layer nodes for 90; by analysing the FTTA-Attention-LSTM analysis model and other comparative models, it was found that the analysis accuracy of the FTTA-Attention-LSTM analysis model is the smallest, and the analysis time meets the real-time requirements, controlled within 0.001s. CONCLUSION: In the future, principal component analysis technology can be introduced for feature selection to further achieve intelligence and improve the analysis efficiency of the model.