Research On The Virtual Enterprise's Partner Formation Process Based On Multi-Agent Technologhy And Its Corparter Selection

With the rapid development of economy globalization, the competitions among the manufacturing enterprises have been more and more fierce. The traditional enterprise working modes are unavailable to fit the great market change. Virtual enterprise, a new enterprise collaboration mode, has come up recently. It can realize mutual superiority share and greatly utilize resource among enterprises. Thus, market risk can be reduced and benefits can be shared. To enterprises, virtual enterprise is also a shortcut for fitting the fast mutative market. It helps to accelerate enterprise's response to the market, improve the competitive ability of the enterprise, and avail enterprise's self-development. Based on the research of the virtual enterprise, we propose a set of new virtual enterprise formation methods, which include the formation process and partner selection process of virtual enterprises. This paper has a detailed dissertation from the above two aspects. The main content of this paper is as follows:Firstly, this paper penetrates into the present research situation and main methods of virtual enterprise. Moreover, it points out that agility manufacture will be a trend of the future manufacturing development. The reason of virtual enterprise's creation is also analyzed in detail. The concept and characters, the lifecycle and the forward projects of virtual enterprise are all introduced in this paper.Secondly, through the deep study of the virtual enterprise's formation process and methods, we propose a whole uniform virtual enterprise's framework grounding on the multi-Agent technology. Besides, we design the function of the specific Agent. The network relation and the collaboration relation among the different Agents are also analyzed. This model not only has strong practicability, but also can effectively harmonize the relation among technology, organization and people. Furthermore, it possesses preferable opening and exploiting nature.Thirdly, we put forward the math model and object function by analyzing the process and concept of partner selection. On this base, we devise the genetic algorithm code, fitness function, selection arithmetic operators, the aberrance arithmetic operators, intercrossing arithmetic operators, genetic parameter and the terminal condition of the algorithm. Then we design process of genetic algorithm, including the initialization process, the intercrossing process and the aberrance process.Fourthly, given to the disadvantage of genetic algorithm, GAAA algorithm combining with ant arithmetic is proposed. In the aspect of time cost, this arithmetic is better than the ant arithmetic. From the angle of solution efficiency, this arithmetic is also better than the genetic algorithm. An improved heuristic algorithm is thus formed. If this arithmetic is applied to the process of virtual selection, the time and solution efficiency can be greatly improved compare to the primary algorithm.Fifthly, we use the Matlab as the emlluator, and do some contrast experiments on genetic algorithm and GAAA algorithm. Experiments prove that GAAA algorithm is more effective and practicable.