Agent-based simulation (ABS) modelling had been a prevalent approach in simulating dynamic processes for a variety of domains. However, the constructed models are usually domain-specific in nature. The proprietary issue of existing modelling approaches limit the possibilities of reusing those constructed models in other application domains. Such limitation also poses issues in model extensibility and customization. Thus, this paper presents a generic ABS framework with self-adaptive modelling approach to address the aforesaid limitations. In the ABS framework, self-adaptive modelling architecture supports construction and re-construction of models autonomously. Simulations were conducted for a few dynamic processes derived from two case studies of New Student Registration and Crime Investigation. Simulation results show that different ABS models were successfully developed, extended and customized by the self-adaptive modelling component in the framework.
展开▼
