The Research And Simulation Of Ant Colony Labor Division Model For Virtual Enterprise

With global competition getting fiercer and fiercer, market opportunities vanish quickly. To survive and develop better in the competitive market, enterprises need to cooperate and share resources with each other and to exert the respective core capabilities. Therefore, the organization framework of Virtual Enterprise (VE) is applied more and more in the activities such as research, manufacture and sales of the enterprise.This thesis researches the disadvantages of the Labor Division Model, improves the task property, individual property, learning capability and each probability computation formula in the model according to the similarities between the features of VE and the labor division model, and brings forward an extended ant colony labor division model oriented to the emulation of building VE and verifies the effectiveness of this model by programming.Aiming at the deficiency of existing heuristic methods and stochastic methods for task assignment problem, then this thesis designs a hybrid ant colony labor division algorithm based on the heuristic search of ant colony optimization algorithm and stochastic search of ant colony labor division. This algorithm adds the learning and forgetting coefficients on the basis of the strategy of updating threshold values in the labor division model and introduces the elite ant colony benefiting from the elite strategy in the genetic algorithm and ant colony forage algorithm. Besides, this algorithm creates the tactic of keeping the current optimal solution and global optimal solution based on the elite ant colony and updates threshold matrixes by rewarding and punishing the ant colony, then achieves optimal task assignment on the basis of the stochastic search of labor division algorithm. Experiments on actual problems show that this algorithm converges to the satisfactory solution in a short time, and features good stability.For the complexity of VE, this thesis proposes studying the establishment process of the VE with the Swarm Intelligence (SI), a novel method for complex system modeling. Based on the comparison between the characteristics of the operation process of VE and the labor division model of the ant colony, this algorithm defines each individual in VE as an artificial ant and redesigns the modules such as the attributes, task execution, learning rules, activity environment of the ants according to the requirements of the operation of VE. In the end this thesis designs a prototype system using the labor division model as the task assignment method of Multi-agent system, and then emulates the actual example, and verifies the effectiveness of the system with analysis into the results.