A knowledge-base approach for the design of chemical processes
A prototypical knowledge representation system has been developed. Its basic data structure is frame like and includes a forward chaining production system and, alternatively, a backward chaining production system supporting the certainty factor model or, a limited class of belief networks. The objective of the system is the representation of knowledge related with the conceptual design of chemical process structures in order to automate their generation. Frames have been used to represent the taxonomic relationship between the different classes of objects of the domain, forward chaining rules found their application in encoding decision making knowledge, and the inexact reasoning mechanisms were used to represent knowledge related to tasks requiring hypothesis assessment. The knowledge representation system has been integrated with a modular chemical process simulator. The system was used to develop a knowledge base capable of solving allocation problems dealing with multiple source and product streams. Remarkable aspects of the solutions found by the system are the possibility of understanding the solution path and the ability of augmenting the knowledge base incrementally.
- Research Organization:
- Washington Univ., St. Louis, MO (USA)
- OSTI ID:
- 7264140
- Country of Publication:
- United States
- Language:
- English
Similar Records
The representation of knowledge within model-based control systems
Decaying confidence functions for aging knowledge in expert systems