Dynamic Model Integration Method For Evolution Analysis Of Unconventional Incident

In recent years, with the rapid development of social economy and acceleration of urbanization progress, the unconventional incidents worldwide have experienced rising numbers of occurrences and expanding scales, which caused vital casualty and loss of social properties. The frequent incidents have drawn attentions of both government administrators and domain academic researchers, and how to establish scientific mechanism of decision making responding to unconventional incidents that then leads to effective operation of emergency events, has become a hot research topic recently. One fundamental component of unconventional incident decision making research is the paradigm of dynamic model integration for evolutionary analysis of unconventional incidents, which indicates practical significance. Due to the diversified nature of unconventional incidents, a systematic methodology to integrate heteroid models from various domains is of urgent need to support the evolutionary analysis of incidents and the decision making process.There is not sufficient research of methodology of model integration within the domain of unconventional incident management as far as we perceive, thus this paper attempts to fill the blank by proposing a dynamic model integration method based on knowledge unit to support unconventional incident decision making. Firstly, a model of unconventional incident scenario was proposed in accord with the widely agreed "scenario-response" paradigm within the emergency management domain. Then originated from the idea of primitive constituent of fundamental objective system that embrace unconventional incidents, model of general knowledge unit and knowledge network were developed, from which mathematical models were extracted to form a complete hierarchical network of unconventional incidents. Secondly, to meet the different needs of management of heteroid models from different disciplines, a model representation that separates both model from operator and model from data was proposed following the "analysis and synthesis" thought, and meanwhile the mapping mechanisms of model to operator and model to data were discussed. Thirdly, following the discussion of single model solving and dynamic integration of model network under temporal and spatial constraints, a systematic method of model dynamic integration was developed. Finally, the proposed method was verified and validated by results from a prototype system that simulated the incident of debris flow near an oil tank field. This paper is an attempt to solve the problem of dynamic model integration for unconventional incidents. Firstly, a hierarchical network model of unconventional incidents was proposed based on the theory of knowledge unit, which depicted the fundamental constituents and the mutual relationships of objective systems representing unconventional incidents. Secondly, considering the temporal and spatial characteristics of evolutionary process of unconventional incidents, concepts and formal representations of temporal and spatial scheduling engines were put forward, and the method of dynamic model integration under temporal and spatial constrains were discussed. The content of this paper supplements research of building a decision support system for systematic emergency management, and renders practical significance in supporting scientific decision making and effective response of unconventional incidents.