Проектування нечітких моделей інтелектуальних промислових регуляторів і систем управління

Abstract

The problems of the design of industrial controllers and control systems with varying degrees of intelligence. Noted shortcomings of existing software and hardware tools to support management processes using the knowledge and methods of intelligent control systems. Emphasizes the need to develop a special methodology of designing control systems in its class. Provides a framework of multi-level design methodology of intelligent automatic control systems (ACS). Analyzes the characteristics of intelligent control systems.The use of fuzzy models and regulators, in particular control systems "intellectual in the large" can significantly improve the dynamic characteristics of the Shelter functioning in the conditions of uncertainty of input information or chaotically organized environment. Designing control systems of this class is a complex issue, since obtaining the required dynamic performance achieved in different ways: either by the complexity of the model executive level or through the involvement of intelligent tools with more features. The proposed methodology of the article makes it possible to set the desired correspondence between the executive and the intellectual level of the results of the simulation, depending on the orientation of the problem the problem, management objectives and modalities of the op amp.The basis of the design methodology of control systems with varying degrees (levels) of the intellectual component of simulation methods, learning and adaptation using neural networks, allowing to form a special knowledge base received intellectual level. This makes it possible to formulate the basic requirements for the respective CAD system, the base of which consists of such blocks as block simulation, fuzzy neural network learning and adaptation mechanisms of the block approximate reasoning (including fuzzy inference), driver knowledge base, driver and interpreter fuzzy operations.