Artificial intelligence methodology for simulation modeling

The construction industry has been slow to accept the use of computer simulation as an analytical tool. One of the reasons for lack of acceptance may be the procedural difficulty of using existing computer simulation systems although they are more user-friendly than in the past. That is, those who know most about a process they want to simulate often have little understanding of computer simulation system. Thus, these users must rely on an intermediary to design a computer model of the process and translate it into language recognizable by a particular computer simulation system. The objective of this research, therefore, is to develop a transparent computer interface between a domain expert and a simulation system. That is, the user should not be aware of the requirements of the simulation package when entering information about the process. Therefore, the interface must be designed to avoid any ambiguities for the user. This can be done by separating the domain knowledge from the simulation knowledge and by using only domain language in the interface so that the user is not hampered by unfamiliar simulation terminology. In this way, simulation knowledge is processed internally to create data files in which only domain knowledge is kept so that it can also be changed manually by the user or domain expert. A Generic Simulation Modeler (GSM) is developed to test the concept of the research and includes: (1) Domain Knowledge Processor; (2) State Space Formulator; (3) Logic Builder; (4) Network Generator; (5) Source Code Translator. The purpose of the Domain Knowledge Processor is to translate information of a given system into unambiguous and structured pieces of information. To do this, a small knowledge-based system with about thirty rules identifies certain types of resources and activities. These resources and activities are then elements of the system and need to be logically connected each other to form a complete conceptual model. To establish system logic, the State Space Formulator identifies possible moves between activities as well as start and goal activity. With the state space formulated, the Logic Builder determines the system logic by searching for the Hamiltonian paths for all the resources involved in the system. The Network Generator then generates a conceptual network internally by connecting activities using the system logic. Finally the Source Code Translator translates the network into a source code of a chosen simulation language, CYCLONE.