Experimental frame structuring for automated model construction: Application to simulated weather generation

The source system is the real or virtual environment that we are interested in modeling. It is viewed as a source of observable data, in the form of time-indexed trajectories of variables. The data that has been gathered from observing or experimenting with a system is called the system behavior data base.; In this paper, automatic experimental system structuring, primarily generator model is developed by the serial and parallel composition for the given source data. The time indexed trajectories of variables provide an important clue to compose the DEVS (discrete event specification) model. Once an event set is derived from the time indexed trajectories of a variable, the DEVS model formalism can be extracted from the given event set. The process must not be a simple model generation but a meaningful model structuring of a request. The search engine searches the requested data in the data repository, and then model creation engine is launched to create the DEVS model. The source data is also designed with SES so that only user defined information is deposited. The source data and query designed with SES is converted to XML Meta data by XML converting process. The entity structure serves as a compact representation for organizing all possible hierarchical composition of system so that it performs an important role to design the structural representation of query and source data to be saved. For the real data application, the model structuring with the US Climate Normals is introduced. This data includes daily 1971-2000 normal maximum, minimum, and mean temperature, heating and cooling degree days, and precipitation for selected cooperative. The basic query for US Climate Normals consists of time, spaces, and variables. The complex systems are being developed at different levels of resolution. When the huge size of source data in US Climate Normals are implemented for the DEVS model, the model complexity is unavoidable. This issue is dealt with the creation of the equivalent lumped model based on the concept of morphism. As the one of the approaches for the model abstraction, aggregation is implemented in this paper. The US region is divided by United States Census Bureau to define the resolution level. Three primary components are coupled to provide more flexible system for user query so that the system can output not only DEVS model but also the resolution level itself or spread value. The system was designed to take a query from the user and give an output to the user. Therefore, the application to Web Service Oriented Architecture is a valuable approach. The recently researched DEVSML (DEVS Modeling Language) gives one solution of the way to return the DEVS model with XML language.