Knowledge-based system: Advisor on the selection of earth retaining structures

The objective of the work described was to produce a computer knowledge-based system (KBS) to advise the user on decisions concerning the selection of earth retaining walls. During the design of a retaining wall the designer has often to reason with incomplete and imprecise data and does so by using empirical rules that have been developed from experience. Two stages in the design procedure (wall selection and wall design) have been selected as being suitable for KBS development where the empirical rules are captured in separate knowledge bases. The way in which Grounded Theory is used to elicit and refine the rules of the knowledge bases is briefly discussed. The wall selection knowledge base forms part of a KBS which provides advice on retaining wall selection. The knowledge is represented in a modified Horn clause form of Prolog. Uncertainty is modelled by the theory of support logic in a computer program called FRISP. FRISP allows complete interrogation of the knowledge base and can explain the line of reasoning taken to arrive at a conclusion. It is also able to handle uncertain and incomplete data by either the minimum or multiplication inference models. © 1991.