Research On The Method Of Formation And Automatic Evaluation Of Aggregation Path In Group Simulation

Crowd simulation is a kind of technology about simulation research on the group behaviorthrough the computer, which has made gradually progresses with the development of computergraphics and virtual reality, and has been widely used in various fields. In particular, played animportant role in urban planning, pre-judge of mass emergencies, large anime game production,fire safety, building design, defense and security. Swarm intelligence algorithm is an importantbranch of artificial intelligence. According to the simulation of biological group behavior, itproduced some new ideas and new methods to deal with complex optimization problems andother practical problems. For example, the successful design heuristic algorithm by birdsforaging behavior has effective applied in combinatorial optimization, group path planning,resource scheduling, virtual networking and communications, and other practical issues. Pathplanning is one of the most important researches of the crowd simulation. There are manyanimals and insects, although individual insignificant in nature, but the clear division of laborgroups, orderly. Swarm intelligence algorithm is a kind of collective intelligence algorithms tomimic nature. There is a natural link between crowd simulation and swarm intelligence. In crowdsimulation, how to plan the group path through swarm intelligence algorithm is the currentresearch focus, it not only can provide material and technical support for the group animationcreation, but also for large-scale evacuation simulation to provide underlying support. Decisiontree algorithm is a kind of classification and evaluation method for data through a set of rules,and which is also a method of machine learning. It can be used in path evaluation, and toimprove classification efficiency.The main work in this paper are as follows:(1) In this paper, it gives a method which is based on improved Particle SwarmOptimization division multiple group gathered path planning to improve the shortcomings ofconventional method according to the features of aggregation in crowd simulation. First, changethe parameters of algorithm according to the aggregation; Secondly, give a group divisionmultiple strategy; Finally, provide a dynamic adjustment strategy for the congestion queuingwhich is possible in multi-objective point accumulation. Overcome the problems lack scientificbasis and unclear in the traditional multiple group division. The change effectively improve theefficiency of the path planning of crowd simulation, and enhance the group simulation realism.(2) In this paper, it gives a path-based automatic evaluation model to improve the problemscaused by the slow evaluation and low accuracy and lack of effective features for path evaluationof a large group of particle in crowd simulation. In the aggregation process of group particles,extracted by analyzing the properties which affect their movement behavior, and combined withdecision tree algorithm to generate the judgment and evaluation of good or bad about path. Thismodel is effective in improving the shortcomings of the traditional evaluation methods, with highpracticality and effectiveness.(3) Combined with the specific research project, based on ACIS HOOPS modeling softwarein VS2003development platform, carried out simulation experiments by using the aboveresearch, which verify the feasibility and effectiveness of path planning method based on multi-population particle swarm optimization combined division strategy with dynamicadjustment strategy. Evaluate and compare the traditional path of its improved method in Matlab.The result of experiment shows that the path-based automatic evaluation model based on thedecision tree algorithm improves the speed and accuracy of evaluation.