VIEWS: 145 PAGES: 33 CATEGORY: Education POSTED ON: 5/16/2010 Public Domain
High School Soccer Head Coach Scouting Report Aid EE5390 Fuzzy Logic and Engineering (Project Proposal) Ismael O. Leyva ID#800216811 April 20, 2006 Abstract: The objective of this project is to help the soccer head coach with the scouting report. This system was accomplished through fuzzy logic. Simply by creating a basic set of base rules that define the best output for the coach’s use in his team that corresponds to the statistics that the coach gave as an input of the opponent team to the system. 1 Introduction: Soccer is 90% strategic and 10% physical. Before a game, the player does not only prepare physically, but they also prepare their minds to have a game plan for their opponent team. In order to have this preparation, they need the help of their head coach to teach them the strategies they will use during the game. The Scouting Report is essential to a soccer head coach’s game plan. When a coach does not have a well prepared game plan things can go very wrong when his team steps on the field. For example, if team A steps on the field with a formation and a strategy that is inappropriate or weaker against team B, then team A will be at a disadvantage, making team B’s job easier. The Soccer Head Coach Scouting Report Aid system will help the coach to prepare the game plan, setting a strategy that is appropriate to defeat the opposing team. The system will consist of several inputs that the soccer coaches use when doing the scouting report such as: What is the formation that the other team is more likely to play? Which game strategy will the opposing team use? Which offensive and defensive strategy will they use? Where are they more likely to start their offense (Left, Right, or Center)? After indicating the appropriate inputs, the system’s output will let the head coach know what the best formation and strategy is to use against the opposing team. Soccer has five basic strategic field formations which are; 5-4-1, 5-3-2, 4-4-2, 4- 3-3, and 3-5-2. The 5-4-1 strategic field formation consists of five defensive players, four mid-field players, and one offensive player. This strategy is highly defensive because there are four flat back defensive players and one sweeper. This strategy is used when the opposite team is faster from midfield up. The 5-3-2 strategic field formation 2 consists of five defensive players, three mid-field players, and two offensive players. This strategy is used against teams whom their weakness is in the defensive center. This strategy requires a good scouting report because due to the lack of outside midfielders, offensive plays are likely to occur through the midfield. The 4-4-2 strategic field formation consists of four defensive players, four midfield players, and two offensive players. This strategy is the most commonly used by beginning teams, whether it is a new team or a new coach, because it gives stability to the three field areas. The forward is moved into midfield to put pressure on the two players to score goals. The midfield player bottles up the opposition in the midfield before attacking field. The 4-3-3 strategic field formation consists of four defensive players, three midfield players, and three offensive players. This strategy is not used very often because there is too much imbalance when they are defending. The defense cannot count on much support from the midfield because the midfields have only three players and they are supposed to be constantly giving the ball to the three forwards. The 3-5-2 strategic field formation consists of three defensive players, five midfield players and two offensive players. This strategy is used by teams that have players that have been together for a while and have good communication and team work. This system is for all high school soccer coaches that do not count on videos of the opposing team, or time to watch the opposite team play, but the only thing they could rely on are the opposite team statistics reported in the newspaper or the web. The system will aid the coach when he does the scouting report or is preparing the game plan. The only thing that the coach would have to do is insert the statistics of the opposite team into 3 the system and system will provide the better strategic alignment against the opposite team. Experimental Procedures: First of all, one decides which of the three types of membership function and which of the two basic function types best suits the project. In this case, the interview type was chosen along with the triangular form and the trapezoidal form. The trapezoidal form was only used for the extremities of the membership function. After that was decided, the information was collected of which five basic formations were known by various high school and beginner coaches. The information along containing the five basic formations also has the strategy of the strength and weakness of their offensive and defensive stand. After the information was collected it was decided which were going to be our fuzzy sets and the fuzzy elements of each set. In this system, we used five fuzzy sets and each set contained five elements (see Figure B2 to Figure B6 in APENDIX B). For example, fuzzy set “formation” with elements 5-4-1, 5-3-2, 4-4-2, 4-3-3, and 3-5-2. formation = F = {( x, μ F ( x) : x ∈ X , μ F ( x),∈ [0,1])} Eq (1) strategy = S = {( x, μ S ( x) : x ∈ X , μ S ( x),∈ [0,1])} Eq(2) defense = D = {( x, μ D ( x) : x ∈ X , μ D ( x),∈ [0,1])} Eq(3) offense = O = {( x, μ O ( x) : x ∈ X , μ O ( x),∈ [0,1])} Eq(4) attack = A = {( x, μ A ( x) : x ∈ X , μ A ( x),∈ [0,1])} Eq(5) The output would only happen when all these were active. In this case, it was going to be the interception of the five sets: F, S, D, O, and A (F∩S∩D∩O∩A). 4 ∨ x ∈ X , μ F I S I D IOI A ( x) = min( μ F ( x), μ S ( x), μ D ( x), μ O ( x), μ A ( x)) Eq (6) Matlab fuzzy logic toolbox was used to simulate (design the fuzzy logic rules) of the system. The toolbox was fairly easy to use because once the FIS Editor (see Figure 1B in APENDIX B) is opened one has to decide how many inputs and outputs are needed for the system. In this case they were five inputs and five outputs, from formation to how to start an offensive attack. Once this was decided the one had to decide how many elements each set has (see Figure 2B to Figure 6B in APENDIX B). After this, one has to write the rules in the Rule Editor (see Figure 7B in APENDIX B), and then go to the Rule Viewer (see Figure 8B in APENDIX B) to simulate the system. Fussy Controller: There are four main steps in the fussy controller. One is fuzzification of the crisp data. Then apply the if-then rules to the fuzzified data. Then defuzzify the fuzzified output data (see Figure 9B in APENDIX B). Finally, read the crisp data results. Our crisp data is the input that the coach will give to the system. For example, in the formation set the element 5-4-1 goes from 0 to 0.3 and 4-4-2 goes from 0.3 to 0.7. The coach chooses a value of 0.6 as an input to represent a 4-4-2. The coach is not aware of this, because when he inputs this value the system will fuzzify that input converting it into a linguistic variable, in this case 4-4-2. Or if the coach decides to choose 0.1 as an input then the fuzzified linguistic variable will be 5-4-1. Once the coach chose all his inputs for the five sets and the system has fuzzified the inputs, the fuzzified linguistic variables will go through the rule based system. This rule based system will decide the output. The rule based system consists of if-then rules (see APENDIX A). For example, 5 If (FORMATION is 5-4-1) and (STRATEGY is DEFENSIVE) and (DEFENSE- STRATEGY is TWO-ON-ONE) and (OFFENSE-STRATEGY is COUNTER-ATTACK) and (START-OFFENSE is CENTER), Then (IDEAL-FORMATION is 3-5-2), (IDEAL- STRATEGY is OFFENSIVE), (IDEAL-DEFENSE-STRATEGY is OFFSIDE-TRAP), (IDEAL-OFFENSE-STRATEGY is FULL-FIELD-PRESS), (IDEAL-START – OFFENSE is CENTER). The “and” between the inputs means the interception of the inputs. Therefore, the output of each rule will be the minimum membership value out of the five inputs given the respective rule output the minimum membership value of the inputs (see Eq (6)). Once the rules are applied to the inputs, the output is going to be fuzzy or in a linguistic variable. Then one needs to defuzzify the output in order to convert it into a crisp output. One of the most basic forms to defuzzify the fuzzy output is with a centroid. Once the centroid is found, then it obtains the mean value in a crisp data form out of the fuzzy output (see Figure 1). Results: The results were that when the proper inputs are applied the fuzzy system will provide the best strategic formation a coach might use with its team. The strategic formation corresponds to the input data that was applied to the system taken from the opponent team’s statistics. For example, if the coach chooses 0.4(5-3-2) for the formation and 0.6(balance) for the strategy and 0.6(neutral) for the defense strategy and 0.34(ball control) for the offense strategy and 0.6(center) for the start offense then the output will have a value of 0.5 for 4-3-3 formation, will have an offensive strategy, an offside-trap defensive strategy, a wing-play for the offensive strategy and a center attack 6 as a way to start their offense. The results are shown in figure 1. Walk through a complete example in APENDIX C. Figure 1. Example for formation 5-3-2 Conclusion: My motivation to create this project is that someday in the near future I would love to be a high school soccer coach. Since I lack from the experience and the education of being a soccer coach, this system could give me a head start for the soccer program, making my job easier while I gain experience. This system was tested with a team of children less than 16 years of age. The only thing the coach did to prepare for the game was to put the permitted parameters in the system to know how to stand his team in the 7 soccer field. The results were favorable, because the team was superior to the opponent team. Recommendations for future enhancement: One of my ideas would be that the system would have more intelligence, that instead of one having to give the system inputs, the system would have the ability of pattern recognition. By this, it would be capable to read a recorded video game of the opponent team and obtain the different parameters. For example it will be able to recognize the formation of the team, their strategy, how to defend, how to attack, where they attack frequently, and others. In the future, this system will not only be used by soccer coaches, but by football coaches now that their game is based on strategy in each down. This could benefit the sport. For example, the quarterbacks, sometimes have time to call an audible before the play starts, and if they have this type of system, the coach can tell the quarterback what audible to call now that the defense is in its position. 8 APENDIX A IF-THEN Based Rules: 1. If (FORMATION is 5-4-1) and (STRATEGY is VERY-DEFENSIVE) and (DEFENSE-STRATEGY is WAIT) and (OFFENSE-STRATEGY is COUNTER- ATTACK) and (START-OFFENSE is AIR-LEFT) Then (IDEAL-FORMATION is 3-5-2), (IDEAL-STRATEGY is VERY- OFFENSIVE), (IDEAL-DEFENSE-STRATEGY is PRESSING), (IDEAL- OFFENSE-STRATEGY is FULL-FIELD-PRESS), (IDEAL-START –OFFENSE is AIR-RIGHT). 2. If (FORMATION is 5-4-1) and (STRATEGY is DEFENSIVE) and (DEFENSE- STRATEGY is TWO-ON-ONE) and (OFFENSE-STRATEGY is COUNTER- ATTACK) and (START-OFFENSE is CENTER) Then (IDEAL-FORMATION is 3-5-2), (IDEAL-STRATEGY is OFFENSIVE), (IDEAL-DEFENSE-STRATEGY is OFFSIDE-TRAP), (IDEAL-OFFENSE- STRATEGY is FULL-FIELD-PRESS), (IDEAL-START –OFFENSE is CENTER). 3. If (FORMATION is 5-4-1) and (STRATEGY is BALANCE) and (DEFENSE- STRATEGY is NEUTRAL) and (OFFENSE-STRATEGY is COUNTER- ATTACK) and (START-OFFENSE is AIR-RIGHT) Then (IDEAL-FORMATION is 3-5-2), (IDEAL-STRATEGY is BALANCE), (IDEAL-DEFENSE-STRATEGY is NEUTRAL), (IDEAL-OFFENSE- STRATEGY is FULL-FIELD-PRESS), (IDEAL-START –OFFENSE is AIR- LEFT). 9 4. If (FORMATION is 5-3-2) and (STRATEGY is DEFENSIVE) and (DEFENSE- STRATEGY is TWO-ON-ONE) and (OFFENSE-STRATEGY is BALL- CONTROL) and (START-OFFENSE is GROUND-LEFT) Then (IDEAL-FORMATION is 4-3-3), (IDEAL-STRATEGY is VERY- OFFENSIVE), (IDEAL-DEFENSE-STRATEGY is PRESSING), (IDEAL- OFFENSE-STRATEGY is WING-PALY), (IDEAL-START –OFFENSE is GROUND-RIGHT). 5. If (FORMATION is 5-3-2) and (STRATEGY is BALANCE) and (DEFENSE- STRATEGY is NEUTRAL) and (OFFENSE-STRATEGY is BALL-CONTROL) and (START-OFFENSE is CENTER) Then (IDEAL-FORMATION is 4-3-3), (IDEAL-STRATEGY is OFFENSIVE), (IDEAL-DEFENSE-STRATEGY is OFFSIDE-TRAP), (IDEAL-OFFENSE- STRATEGY is WING-PALY), (IDEAL-START –OFFENSE is CENTER). 6. If (FORMATION is 5-3-2) and (STRATEGY is OFFENSIVE) and (DEFENSE- STRATEGY is OFFSIDE-TRAP) and (OFFENSE-STRATEGY is BALL- CONTROL) and (START-OFFENSE is GROUND-RIGHT) Then (IDEAL-FORMATION is 4-3-3), (IDEAL-STRATEGY is BALANCE), (IDEAL-DEFENSE-STRATEGY is NEUTRAL), (IDEAL-OFFENSE- STRATEGY is WING-PALY), (IDEAL-START –OFFENSE is GROUND- LEFT). 7. If (FORMATION is 4-4-2) and (STRATEGY is DEFENSIVE) and (DEFENSE- STRATEGY is WAIT) and (OFFENSE-STRATEGY is NEUTRAL) and (START-OFFENSE is GROUND-LEFT) Then (IDEAL-FORMATION is 4-4-2), (IDEAL-STRATEGY is OFFENSIVE), (IDEAL-DEFENSE-STRATEGY is OFFSIDE-TRAP), (IDEAL-OFFENSE- STRATEGY is NEUTRAL), (IDEAL-START –OFFENSE is GROUND- RIGHT). 10 8. If (FORMATION is 4-4-2) and (STRATEGY is BALANCE) and (DEFENSE- STRATEGY is NEUTRAL) and (OFFENSE-STRATEGY is NEUTRAL) and (START-OFFENSE is CENTER) Then (IDEAL-FORMATION is 4-4-2), (IDEAL-STRATEGY is BALANCE), (IDEAL-DEFENSE-STRATEGY is NEUTRAL), (IDEAL-OFFENSE- STRATEGY is NEUTRAL), (IDEAL-START –OFFENSE is CENTER). 9. If (FORMATION is 4-4-2) and (STRATEGY is OFFENSIVE) and (DEFENSE- STRATEGY is OFFSIDE-TRAP) and (OFFENSE-STRATEGY is NEUTRAL) and (START-OFFENSE is GROUND-RIGHT) Then (IDEAL-FORMATION is 4-4-2), (IDEAL-STRATEGY is DEFENSIVE), (IDEAL-DEFENSE-STRATEGY is WAIT), (IDEAL-OFFENSE-STRATEGY is NEUTRAL), (IDEAL-START –OFFENSE is GROUND-LEFT). 10. If (FORMATION is 4-3-3) and (STRATEGY is BALANCE) and (DEFENSE- STRATEGY is NEUTRAL) and (OFFENSE-STRATEGY is WING-PLAY) and (START-OFFENSE is GROUND-LEFT) Then (IDEAL-FORMATION is 5-3-2), (IDEAL-STRATEGY is OFFENSIVE), (IDEAL-DEFENSE-STRATEGY is OFFSIDE-TRAP), (IDEAL-OFFENSE- STRATEGY is BALL-CONTROL), (IDEAL-START –OFFENSE is GROUND- RIGHT). 11. If (FORMATION is 4-3-3) and (STRATEGY is OFFENSIVE) and (DEFENSE- STRATEGY is OFFSIDE-TRAP) and (OFFENSE-STRATEGY is WING-PLAY) and (START-OFFENSE is CENTER) Then (IDEAL-FORMATION is 5-3-2), (IDEAL-STRATEGY is BALANCE), (IDEAL-DEFENSE-STRATEGY is NEUTRAL), (IDEAL-OFFENSE- STRATEGY is BALL-CONTROL), (IDEAL-START –OFFENSE is CENTER). 11 12. If (FORMATION is 4-3-3) and (STRATEGY is VERY-OFFENSIVE) and (DEFENSE-STRATEGY is PRESSING) and (OFFENSE-STRATEGY is WING- PLAY) and (START-OFFENSE is GROUND-RIGHT) Then (IDEAL-FORMATION is 5-3-2), (IDEAL-STRATEGY is DEFENSIVE), (IDEAL-DEFENSE-STRATEGY is TWO-ON-ONE), (IDEAL-OFFENSE- STRATEGY is BALL-CONTROL), (IDEAL-START –OFFENSE is GROUND- LEFT). 13. If (FORMATION is 3-5-2) and (STRATEGY is BALANCE) and (DEFENSE- STRATEGY is NEUTRAL) and (OFFENSE-STRATEGY is FULL-FIELD- PRESS) and (START-OFFENSE is AIR-LEFT) Then (IDEAL-FORMATION is 5-4-1), (IDEAL-STRATEGY is BALANCE), (IDEAL-DEFENSE-STRATEGY is NEUTRAL), (IDEAL-OFFENSE- STRATEGY is COUNTER-ATTACK), (IDEAL-START –OFFENSE is AIR- RIGHT). 14. If (FORMATION is 3-5-2) and (STRATEGY is OFFENSIVE) and (DEFENSE- STRATEGY is OFFSIDE-TRAP) and (OFFENSE-STRATEGY is FULL- FIELD-PRESS) and (START-OFFENSE is CENTER) Then (IDEAL-FORMATION is 5-4-1), (IDEAL-STRATEGY is DEFENSIVE), (IDEAL-DEFENSE-STRATEGY is TWO-ON-ONE), (IDEAL-OFFENSE- STRATEGY is COUNTER-ATTACK), (IDEAL-START –OFFENSE is CENTER). 15. If (FORMATION is 3-5-2) and (STRATEGY is VERY-OFFENSIVE) and (DEFENSE-STRATEGY is PRESSING) and (OFFENSE-STRATEGY is FULL- FIELD-PRESS) and (START-OFFENSE is AIR-RIGHT) Then (IDEAL-FORMATION is 5-4-1), (IDEAL-STRATEGY is VERY- DEFENSIVE), (IDEAL-DEFENSE-STRATEGY is WAIT), (IDEAL-OFFENSE- STRATEGY is COUNTER-ATTACK), (IDEAL-START –OFFENSE is AIR- LEFT). 12 APENDIX B Figure 1B: FIS Editor Inputs: (5 Sets) Outputs: (5 Sets) 1. Formation 1. Formation 2. Strategy 2. Strategy 3. Defense-Strategy 3. Defense-Strategy 4. Offense-Strategy 4. Offense-Strategy 5. Start-Offense 5. Start-Offense 13 Figure 2B: Membership Function Editor (SET: Formation) Formation = {5-4-1, 5-3-2, 4-4-2, 4-3-3, 3-5-2} 14 Figure 3B: Membership Function Editor (SET: Strategy) Strategy = {Very-Defensive, Defensive, Balance, Offensive, Very-Offensive} 15 Figure 4B: Membership Function Editor (SET: Defense-Strategy) Defense-Strategy = {Wait, Two-On-One, Neutral, Offside-Trap, Pressing} 16 Figure 5B: Membership Function Editor (SET: Offense-Strategy) Offense-Strategy = {Counter-Attack, Ball-Control, Neutral, Wing-play, Full-Field-Press} 17 Figure 6B: Membership Function Editor (SET: Start-Offense) Start-Offense = {Ground-Left, Air-Left, Center, Air-Right, Ground-Right} 18 Figure 7B: Rule Editor 19 Figure 8B: Rule Viewer and Fuzzy Output (For Simulation) 20 FORMATION (5) IDEAL-FORMATION (5) coach STRATEGY (5) IDEAL-STRATEGY (5) (mamdani) DEFENSE-STRATEGY (5) IDEAL-DEFENSE-STRATEGY (5) 15 rules OFFENSE-STRATEGY (5) IDEAL-OFFENSE-STRATEGY (5) STARTS-OFFENSE (5) IDEAL-START-OFFENSE (5) System coach: 5 inputs, 5 outputs, 15 rules Figure 9B: Fuzzy Controller Setup (Fuzzy Inputs, Rule Based System, and Fuzzy Output) 21 APENDIX C (Example) Coach Inputs: Formation = 0.2; which selects Element “5-4-1” and Element “5-3-2” with a degree of membership = 0.5. 22 Strategy = 0.4; which selects Element “Defensive” and Element “Balance” with a degree of membership = 0.5. 23 Defensive-Strategy = 0.4; which selects Element “Two-On-One” and Element “Neutral” with a degree of membership = 0.5. 24 Offensive-Strategy = 0.2; which selects Element “Counter-Attack” and Element “Ball-Control” with a degree of membership = 0.5. 25 Start-Offense = 0.4; which selects Element “Air-Left” and Element “Center” with a degree of membership = 0.5. 26 With this inputs, three base rules are activated: Rule #2, Rule #3, and Rule #5. 2. If (FORMATION is 5-4-1) and (STRATEGY is DEFENSIVE) and (DEFENSE-STRATEGY is TWO- ON-ONE) and (OFFENSE-STRATEGY is COUNTER-ATTACK) and (START-OFFENSE is CENTER) Then (IDEAL-FORMATION is 3-5-2), (IDEAL-STRATEGY is OFFENSIVE), (IDEAL-DEFENSE- STRATEGY is OFFSIDE-TRAP), (IDEAL-OFFENSE-STRATEGY is FULL-FIELD-PRESS), (IDEAL-START –OFFENSE is CENTER). 3. If (FORMATION is 5-4-1) and (STRATEGY is BALANCE) and (DEFENSE-STRATEGY is NEUTRAL) and (OFFENSE-STRATEGY is COUNTER-ATTACK) and (START-OFFENSE is AIR- RIGHT) Then (IDEAL-FORMATION is 3-5-2), (IDEAL-STRATEGY is BALANCE), (IDEAL-DEFENSE- STRATEGY is NEUTRAL), (IDEAL-OFFENSE-STRATEGY is FULL-FIELD-PRESS), (IDEAL- START –OFFENSE is AIR-LEFT). 5. If (FORMATION is 5-3-2) and (STRATEGY is BALANCE) and (DEFENSE-STRATEGY is NEUTRAL) and (OFFENSE-STRATEGY is BALL-CONTROL) and (START-OFFENSE is CENTER) Then (IDEAL-FORMATION is 4-3-3), (IDEAL-STRATEGY is OFFENSIVE), (IDEAL-DEFENSE- STRATEGY is OFFSIDE-TRAP), (IDEAL-OFFENSE-STRATEGY is WING-PALY), (IDEAL- START –OFFENSE is CENTER). First, the rule will select the minimum degree of membership out of the five inputs for each rule. ∨ x ∈ X , μ F I S I D IO I A ( x) = min( μ F ( x), μ S ( x), μ D ( x), μ O ( x), μ A ( x)) Eq (6) Then, it would plot it to its corresponding output. Once all outputs for each rule is plot it, the system will combine all outputs into a single one. Then it will get the centroid for each output. 27 Outputs: Ideal-Formation = {Centroid = 0.8} 28 Ideal-Strategy = {Centroid = 0.6} 29 Ideal-Defense-Strategy = {Centroid = 0.6} 30 Ideal-Offense-Strategy = {Centroid = 0.8} 31 Ideal-Start-Offense = {Centroid = 0.4} 32 Actual Simulation Result at the Rule Viewer: 33