Knowledge-association networks for system design

Abstract

Knowledge-association networks, a new framework for a knowledge base for automatic system design, is proposed. With this framework, knowledge base developers can easily avoid verbose descriptions, errors and inconsistencies, and can reduce the computational complexity in automatic system design. The networks focus on the associations among pieces of knowledge about system components. The associations show orthogonal relationships for the traditional relationships, such as classifications and structures. The networks include four kinds of associative relationships: selective, direct, concatenated and meta-associations, which are derived from the dependencies among the system components and engineers' knowledge associations. To investigate the effectiveness of the knowledge-association networks, they are applied to an automatic power system design tool, the Advanced Plant Engineering Cycle tool, version 3 (APEC3). APEC3 supports electrical engineering processes by producing single line diagrams, specification balance statements and item lists from the customer's electrical load requirements. In practical use, the design results showed that APEC3 can reduce the design time to 1/8 th of the normal time. The networks and the design method can also be effective for information systems, because the type of design problem is not limited to power system design.