Design And Implementation Of Agent Based On Adaptive Genetic Algorithm In Short Track Speed Skating Simulation System

Computer simulation, as an integrated discipline, makes remarkablecontributions to the development of various fields, including sports. The intelligencein simulation system is a significant research branch in the field of artificialintelligence, especially to the intelligence-oriented system, artificial intelligence hasplayed a more pivotal role. With the development of artificial intelligence, a lot ofacademic theory and application results of which can be introduced to simulationsystem to varying degrees.Firstly, different types of agent decision-making model are described, thenaccording to the features that the short track simulation system is more commendingreal-time and dynamic, this paper selects the reactive-agent system as the frameworkof the agent’s decision-making process. And it also carries out the details of eachmodule in the framework, which greatly makes the chosen of the method in agentdecision-making process more convenient.Secondly, considering the system is about short track speed skating, this paperpresent the definition, classification, storage, and calculation of tactic by full chapter,which is the basis of the paper to start work, since most of work the simulationsystem commit to training the best tactical in accordance with a particular athlete.Meantime tactical driving mechanism meets the different performance of agent.Finally, this paper defines interface between agent module and server with thedetailed analysis of the factors that influence the instructions. In the design ofdecision-making algorithm, taking into account of the requirement that agent canimpersonate the high-level athletes with different personal styles, this paper adoptsadaptive genetic algorithm, gives encoding scheme and implementation of fitnessfunction aiming at practical problem in decision-making of the simulation system.By applying the adaptive genetic algorithm to the agent decision-makingprocess of the system, the agents achieve well effect, no matter in tactical training orsmart decisions. Through comparison of existing decision tree, the paper gives thepros and cons of different algorithms supported agents, and after several timestraining, obtain the best tactics based on a particular model of athlete.