A Model for Integration of Knowledge-Based Components in Existing Process Control Systems

Abstract

Integration of knowledge-based and conventional systems is a particularly hard problem because of the complete incompatibility of data structures of this two kinds of systems. The weak points of knowledge-based systems such as poor performance, enormous resource requirements and incomprehensibility hinder a wider acceptance of knowledge-based approaches in process control. The new approach for the integration of knowledge-based and conventional systems provides a unified data model for both worlds, offers a structuring mechanisms for the rule and the fact base and improves the performance. The main contribution of this work is a new statically and strongly typed knowledge-based language fully compatible with C++, which constructs provide for type-safe and efficient integration with conventional components. A particularly important point is that no modification of the conventional code and no conversions of the data are needed, providing for exchangeability of paradigms in process control applications.