AN APPROACH TO MONITORING AND DIAGNOSING ENGINEERING PROCESSES BY A KNOWLEDGE-BASED SYSTEM

Abstract

This article presents the approach used for building and using the knowledge-based system SUPER to fulfill several types of requirements corresponding to different engineering-domain contexts. SUPER was developed for maintaining and updating information about an evolving system as its slate changes due to abnormal events such as faults or as a consequence of external actions taken on the system. After analyzing different requirements corresponding to engineering-application domains, the successive steps of the approach are given, mainly based on structuring the knowledge base in the shape of a network at the time of knowledge acquisition and maintaining the database consistent at run time.

The first step was to develop a knowledge-acquisition module that uses a logical knowledge model to construct and compile the knowledge base as an AND/OR network and that maintains a consistent network of rules as it incrementally increases. The second step was to develop a looped inference engine to select an appropriate sequence of actions to be applied to the system to change its states and to update the database according to a change of states in response to an external action. In the third step, the capacity of the system was raised to the level of managing variables and propagating their domain constraints in the AND/OR network of rules. Finally, SUPER has been implemented on several host systems, particularly in a real-time multitasking environment. With the addition of temporal logic Junctions, SUPER has been used for real-time applications and was successfully applied in several engineering domains, such as process control, medical monitoring, and on-board diagnosis.