Research on multi-agent-based simulation of supervision issues

Supervision issues exist among almost every industry of this world. Based on a multi-agent modeling and simulation approach, this paper, which is different from the existing researches, designs the interaction rules of supervision and violation, builds the evolving mechanism using genetic algorithm, and simulates the evolving process on the swarm platform. On the basis of the results of the simulation, it reach the conclusion that under the condition of ldquobounded rationalityrdquo and ldquoincomplete informationrdquo, to eliminate the violation, the supervisor must obtain almost the ldquocomplete informationrdquo of the behavior of actors, apart from making their expected revenue of obedience greater than the expected revenue of violation. A heavier punishment can accelerate the progress of the evolution.