The Research On Diagnosis And Evaluation Of Skills And Tactics Of Table Tennis Matches Based On Artificial Intelligence

The research is from Research and Implementation of Winning Factors in the Racket Events (NO.2006BAK12B03)and Research on the Skills and Tactic Analysis and Diagnosis of the Racket Events in Competitive Sports (NO.072705128), those are Specific Plan for High-Tech Olympics of National Science and Technology Project and the Key Technology Research and Development Programs of Shanghai.The purpose of the article is to explore the new diagnosis method of skills and tactics of table tennis matches and to try the application of artificial intelligence for diagnosis and analysis of table tennis matches. It also discusses the method of evaluation and predicting based on the outcome of diagnosis of table tennis matches.The artificial intelligence methods study how to simulate human's intelligence and behaviors by means of computer technology. The decision-tree method can build the tree according to goal data set,obtain decision rules and provide support of table tennis matches. The rough set method can get the minimum set of goal set and optimization attribute combinations. Depending on its efficient nonlinear fitting abilities, the artificial neural network method can find the implied functions. For their advantages, it is very important to study the application of diagnosis evaluation and predicting of table tennis matches systematically.The study of skills and tactics indicator system of table tennis matches for diagnosis and evaluation. Based on the analysis and summary of related literature, we found that the study on diagnosis of table tennis' skills and tactics developed in two different directions:abstract and concrete. The article manages to build new skills and tactics indicator system to bound five elements of table tennis together. Firstly, it builds skills system by combining skills and placement of table tennis, which includes serve, forehand loop drive (fast attack), backhand loop drive (fast attack), flick, Chop, Chopping short, Push(block). Secondly, it builds tactics system by combing tactics and time series of competition, which includes three phases such as serving phase, receiving phase, countering phase and four tactics such as attack, control, defense and balance.The contrast of skills-tactics diagnosis of table tennis matches based on artificial intelligence, which includes the application of decision tree and rough set based on the winnowing results of decision tree and reduction results of rough set. The results indicated that ANN and DT method are stable, the reduction results of rough set are different among several tests. ANN method is able to fit goal set precisely if the goal error is selected justly, DT method is good and rough set method is poor. The results of DT and rough set method are tend to attributes with large value, but ANN method is tend to find the function relationship between skills attributes and winning percentage according to its better transfer function.The improvement of method of skills and tactics diagnosis of table tennis matches based on artificial neural network, firstly, the way of improvement of delta is changed, secondly, the matches are divided into three kinds of circumstances such as the best, good and poor performance.The study of skills-tactics evaluation and predicting of table tennis matches based on artificial intelligence, which discusses the importance, approach, principle and processing procedures of evaluation, that is divided into two circumstances such as evaluation between players and some player.The results showed that the evaluation method has a strong approach and better results and can reflect the difference reasonably among matches. The article try to predict the results of table tennis matches based on the results of diagnosis and discuses the meaning, classification and procedures of predicting. It studies the combination of the decision tree and the artificial neural networks to predict the outcome of table tennis matches.Practices have shown that the algorithm is applicable. The new methods used data from 70 matches to develop predictive models of excellent ping-pong players and 32 matches for test. Combination prediction, which takes on average 3.1094s, takes only 18.61% of ANN would take.The accuracy is 0.9285;is close to ANN.It studies the characteristics of skills of excellent table tennis players based on the improved ANN diagnosis method. The results showed that:The forehand loop (fast attack) is a major pointing method of all excellent table tennis players. The skills of serving are also important pointing method, especially, at middle half placement. The characteristics of skills between Chinese male table tennis players and foreign are simlar; there are some differences in the importance of forehand loop drive.For foreign male players, the importance of forehand loop drive is different from Chinese layers such as the hitting placement; the series of their goal average of forehand loop drive is backhand middle forehand; Chinese players are backhand forehand middle. For Chinese female players, the most important skills are forehand loop drive and backhand loop drive. The placement of forehand and backhand are primary pointing method; however, those are backhand and middle for foreign female players.It also studies the characteristics of tactics of excellent table tennis players. The results showed that:The most important tactics of male table tennis players are serving-attack and reserving-control.The most important tactics of female table tennis players are serving-attack and countering-balance.Compared to the foreign male players, the tactics of serving-control and reserving-control of Chinese male players are better. The tactics of defense are important to foreign players. The tactics around attack are most widely used for Chinese female players. The uses of receiving countering tactics play a decisive role in foreign female players'matches.The improved diagnosis method is used to excellent table tennis players at the first time such as CHNMA, CHNMB, CHNMC, CHNMD, CHNME, CHNFA, CHNFB, CHNFC, CHNFD, CHNFE, FORMA (GER), FORMF (GER), FORMB (BLR), FORMC (KOR), FORMD (KOR), FORME (TPE), FORFA (SIN), FORFB (SIN), FORFC (SIN), FORFF (SIN), FORFD (JPN), FORFE (KOR). The diagnosis conclusion is used for competitor analysis of the 29th Olympic Games and the 50th World Table Tennis Championship, which has been proposed and broadly accepted by Chinese coaches and players.