Research And Implementation Of Badminton Technical Characteristics Statistics And Pace Training Based On Machine Learning

From the Internet and mobile Internet to cloud computing and big data,to machine learning and artificial intelligence,information technology has changed people's production and lifestyles over the past decade.The wave of new technology iterations and digital innovation has driven traditional industry changes.With the rapid development of machine learning and artificial intelligence,the society has gradually entered the era of intelligence.A series of research results and intelligent products based on machine learning and artificial intelligence have emerged,such as Baidu intelligent robots,Google AlphaGo intelligent go masters,etc.The resulting "artificial intelligence+" wave of new industrial revolution is rapidly infiltrating into various fields.This paper applies the machine learning method to the classification and identification of badminton shots,and builds a technical characteristics statistics and pace training system for badminton based on this algorithm.The main innovations and achievements of the thesis include:(1)An algorithm for recognizing the ball-attacking action of badminton is proposed,and on this basis,a real-time recognition system for badminton action is implemented.In this paper,a single acceleration sensor fixed on a badminton racket grip is used to collect the data of badminton movements,and a sliding window data segmentation technique is used to extract the shot signals.An unsupervised learning algorithm such as k-means is used for clustering evaluation and vector quantization,by exploring the characteristics of Hidden Markov Model(HMM)and training methods,proposes an improved HMM training model algorithm for identifying common ten badminton shots.Including serving,forehand rubbing,backhand rubbing,forehand lunging,backhand lunging,forehand pushing,backhand pushing,forehand picking,backhand picking,and forehand lofty.Experiments show that the algorithm model system designed in this paper can recognize ten common hitting motions in real time.The average recognition rate of the improved HMM is 6.4%higher than that of the traditional HMM,and the overall recognition rate of the final hitting movement can reach 94%.(2)Design and implementation of badminton technical characteristics statistics and pace training system.Based on the above-mentioned badminton action recognition algorithm and other related algorithms(such as score and error assessment algorithm,step recovery algorithm,etc.),this paper has constructed a set of badminton technical characteristics statistics and pace training system.This system realizes the technical statistics of badminton games and the players' pace training.The technical statistics mainly include the scores and errors of each of the player's batting movements,the technical characteristics of various batting movements,and other information;the pace training is mainly used to quantitatively analyze and compare the pace of several types of shuttlecock used in actual combat.It has achieved targeted training for athletes at specific paces.The system test experiment shows that the comprehensive accuracy rate of the technical statistical function data is 96.7%.The player's pace training system can accurately measure the player's pace information and help to quickly improve the player's pace.The above research has solved the hardware and software engineering technical problems from data acquisition to algorithm programming,and provided a functional platform for intelligent analysis and assisted training of badminton sports.It can be applied to more professional athletes and coaches to accumulate large amounts of data and refine more technical parameters that match the rules of badminton sports,eventually formed a set of professional products for intelligent analysis and auxiliary training of badminton.