Knowledge-based simulation and identification of various metallurgical reactors
- PDF / 1,615,804 Bytes
- 15 Pages / 603.28 x 783.28 pts Page_size
- 69 Downloads / 151 Views
I.
INTRODUCTION
RESEARCH on the application of expert systems and artificial intelligence techniques in chemical engineering may be grouped into the following areas: (1) the synthesis of chemical engineering flow sheets in hybrid-type systems (hybrid system: knowledge-based system (KBS) integrated with equation-orientated simulation programs), t1'2"31 (2) selection programs, e.g., DECADE* for the selec*DECADE is a trademark of Combustion Engineering, Inc., Stamford, CT.
tion of catalysts, t4'51 (3) synthesis of operating procedures for chemical plants,tl,6-8] (4) malfunction diagnosis for complete chemical plants, t9-12j and (5) control synthesis, r1'~3-151 Although different architectures are used to define the knowledge base, all cited examples apply both qualitative (experimental and heuristic) and quantitative (equation-orientated) knowledge, also termed shallow and deep knowledge, respectively. In these systems, data and knowledge are mostly represented as (1) objects (a chunk of structured knowledge), t~l which are also called frames, (2) production rules, which are characterized by if-then type statements, and (3) functions, which define the equations for quantitative simulation of unit operations. These data structures form the basis of the various knowledge-based systems and are embedded in various architectures which include: (I) object-orientatedf~'121or frame-orientatedpl systems, M.A. REUTER, formerly Senior Lecturer, University of Stellenbosch, is Research Metallurgist with the Institute for Non-Ferrous Process Metallurgy, Aachen University of Technology, 5100 Aachen, Federal Republic of Germany, J.S.J. VAN DEVENTER, Professor, is with the Department of Metallurgical Engineering, University of Stellenbosch, Stellenbosch 7600, Republic of South Africa. Manuscript submitted May 29, 1990. METALLURGICAL TRANSACTIONS B
(2) hybrid systems in a Blackboard (Blackboard is the global data base permitting the interaction of all knowledge sources) architecture, ta'Sj (3) hierarchical diagnostic systems in which every node in the structure represents a malfunction hypothesis, the top node containing general malfunction information and vice versa (hence, the information is grouped into various levels of abstraction), t6'91 or (4) goal tree-success tree architecture used for the design of control systems for distillation columns, u51 Of the possibilities listed, an object-orientated approach was applied in the program discussed here to define the structure of the knowledge base for the proposed diagnostic and simulation system for metallurgical reactors. Whereas most of these types of applications implement object-orientated languages (e.g., KEE or CommonLISP)t~J for the programming of the qualitative model, Pascal, which permits the definition of complex data structures and performance of numerical analysis in the same program, was applied. This makes the timeconsuming communication between two different programs, i.e., one to perform the knowledge engineering and the other the numerical simulation, unnecessary
Data Loading...
