An Enterprise Multi-agent Model with Game Q-Learning Based on a Single Decision Factor

In recent years, the study of enterprise survival development and cooperation in the whole economic market has been rapidly developed. However, in most literature studies, the traditional enterprise multi-agent cannot effectively simulate the process of enterprise survival and development since the fundamental characteristics used to describe enterprises in social networks, such as the values of enterprise multi-agent attributes, cannot be changed in process of the simulation. To address this problem, an enterprise multi-agent model based on game Q- learning to simulate enterprise decision making which aims to maximize the benefits of enterprises and optimize the effect of inter-firm cooperation is proposed in this article. The Firm Q Learning algorithm is used to dynamically change the attribute values of the enterprise multi-agent to optimize the game results in the evolutionary game model and thus effectively simulate the dynamic cooperation among the enterprise agents. The simulation result shows that the evolution of the enterprise multi-agent model based on game Q-learning can more realistically reflect the process of real enterprise survival and development than the multi-agent simulation with fixed attribute values.