Modeling and simulation of intelligent behavior in complex systems: the extension of the X language

Currently, the dynamic and evolutionary nature of complex systems has gained increasing prominence. Intelligent behavior plays an important role in driving the dynamic evolution of complex systems. However, current agent-specific modeling languages (e.g. AML) struggle to integrate with complex system modeling tools (e.g. SysML), while SysML itself faces challenges in directly modeling and simulating intelligent behavior. Compared to traditional methods that combine agent-specific tools with complex system modeling languages, X language provides an integrated approach to modeling both complex systems and intelligent behaviors. However, its focus remains on single-agent system design, with limited support for multi-agent interaction and learning behavior modeling in agents. To solve this problem, this paper expands the agent class of the X language, and provides multi-agent interaction syntax and reinforcement learning (RL)-based learning ability for the X language. At the same time, based on the MADES agent framework, we implemented the compiler algorithm to compile the extended syntax into the MADES simulation file on the basis of the X language compiler. Finally, a multi-agent autonomous driving model was developed using the Q-learning algorithm to verify the modeling syntax. The results show that the extended X language syntax effectively supports building multi-agent interaction systems with learning capabilities.