Концептуальные основы интеллектуального моделирования

Abstract

Рассмотрены вопросы интеллектуализации средств моделирования сложных процессов и систем. В результате аналитического обзора подходов предложена усовершенствованная концепция интеллектуального моделирования на основе анализа и структуризации знаний в предметной области построения моделей.

Розглянуто питання інтелектуалізації засобів моделювання складних процесів та систем. В результаті аналітичного огляду підходів запропоновано удосконалену концепцію інтелектуального моделювання на основі аналізу і структуризації знань у предметній галузі побудови моделей.

An analytical overview has been made on existing approaches to developing intelligent methods and tools for modeling complex processes and systems, including the support for the tasks of administrative decisions in various socio-economic sectors. It is concluded that the vast majority of the existing publications justify the implementation of the “intelligent modeling” simply by using the neural networks, evolutionary methods and other means of computational intelligence. In contrast, in this study a new concept of intelligent modeling as the complex processes and systems is developed, according to which it is proposed to distinguish the three main aspects: the intelligent offline modeling with the characteristics of a complex system from statistical data; the intelligent online modeling as a part of a control or decision-making process in the real time; a systemic intelligent modeling. Three of these types or levels of the modeling process can be specified as: • Intelligent modeling offline is a static task of the intellectual support of the process for building models out of the system control (from fixed base or data sample). It is shown that a proper system should be based on the inductive modeling tools, have a database and knowledge base as well as including tools of the intelligent interface. A methodology of their development is formulated based on a formalized structuring of knowledge about the subject area of the mathematical modeling from statistical data. The intelligence is focused here exactly in the interface. • Intelligent modeling online is a dynamic task of construction, adjustment and restructuring models in the system operation process (from changeable database). The appropriate system should include all the elements of the previous system and the tools supporting the knowledge-driven process of automatic or automated building models that plausibly describe the behavior of the objects in the conditions of uncertainty and incomplete prior information about the properties of the simulated objects and environment in which they operate, with accuracy being sufficient to making effective decisions by DMP under conditions of permanent changing the situation. • Systemic intelligent modeling should provide an intellectual support of processes of DSS modeling in a complex system to automatically detect optimal operating modes of a real system as well as the possible adverse or dangerous modes. The corresponding integrated system should contain the following key elements: information subsystem, which function is observing and the data storage; monitoring subsystem which is actually an online modeling; subsystem DSS which has formed the appropriate options for possible solutions and evaluated its effectiveness according to the certain criteria. This complex is practically a situational modeling system and has all the characteristics of intelligence. It includes two previous levels of the intelligent modeling. Such system has the necessarily accumulation function of knowledge about the object being modelled and the environment, as well as options for reasonable decisions in the changing situations.