Research On Representation And Integration Of Model Based On Knowledge-Element

With the growing damages caused by emergency, scientific and efficient emergency has become a major issue. Emergency model can use this information, domain knowledge to scientific forecast and warning the emergencies, properly allocate resources and help in make scientific emergency decision. However, due to the unpredictable, dynamism and critical character of emergency seriously limited the use of model, these limits the better use of emergency model and domain knowledge to improve the model selection and integration, which cannot integrate several models to solve the complex problems.As part of the national natural science foundation research project "research on the evolution of unconventional emergency response methods" concerning the shortcomings in related studies, choose the emergency management model as the research object, this paper firstly present the model into frame model layer, the form model layer and the real model layer, which provide a foundation to model choose and model integration. Then, it building method to improve the efficiency and accuracy in model choose, by put the domain knowledge into the attribute selection algorithm during building the decision tree, it ensured that the selected model is most suitable for this event. After that, a dynamic model integration method based on knowledge Element net and the model hierarchy representation method was developed. Describes how to build the relationship between the three-level models and achieve model integration. At the end of this paper, coal and gas bombings are incorporated to demonstrate these methods.The three-layer model representation can map the model parameters to the attributes of the knowledge-element, and easy to model meta-represent, meta-reasoning and model reuse. The model choosing methods based on knowledge-network, not only reduce the requirements of the domain knowledge of decision-makers, but also improve the accuracy and timeliness of model selection. Use the model combination network to achieve model integration, we can catch the change of decision-making environments, ensure the scientific of the integrate relation between two models and solve the data mapping problem.