Hybrid knowledge representation in a blackboard KBS for liquid retaining structure design

This paper highlights the importance of design expertise, for designing liquid retaining structures, including subjective judgments and professional experience. Design of liquid retaining structures has special features different from the others. Being more vulnerable to corrosion problem, they have stringent requirements against serviceability limit state of crack. It is the premise of the study to transferring expert knowledge in a computerized blackboard system. Hybrid knowledge representation schemes, including production rules, objectoriented programming, and procedural methods, are employed to express engineering heuristics and standard design knowledge during the development of the knowledge-based system (KBS) for design of liquid retaining structures. This approach renders it possible to take advantages of the characteristics of each method. The system can provide the user with advice on preliminary design, loading specification, optimized configuration selection and detailed design analysis of liquid retaining structure. It would be beneficial to the field of retaining structure design by focusing on the acquisition and organization of expert knowledge through the development of recent artificial intelligence technology. Introduction Liquid retaining structures can be classified as rectangular or circular according to shape. They can also be classified as underground or above the ground according to location. Liquid retaining structures are to be designed for hydrostatic pressure force, active earth pressure, wind force, seismic force, in addition to self-weight and imposed load. Being more vulnerable to corrosion problem, they have stringent requirements against serviceability limit state of crack. In order to select the appropriate design parameters, structural designers depend principally on their subjective judgments, intuition, professional experience, expertise, and rules of thumb. Empirical knowledge of expert designers constitutes a valuable source of information in structural design process. Yet, only a small proportion of such rules is well documented in the existing literature. It is desirable to facilitate transfer of expertise knowledge in this domain to less experienced design engineers. Structural design is a complex task integrating both design knowledge and analytical skills. Although structural analysis in the detailed design can be automated in an algorithmic fashion, the process of structural design cannot be reduced to a predetermined mathematical sequence. Conceptual design is largely performed in the mind of the designer solely by experience and is featured by a high degree of uncertainty with often more than a unique feasible solution. However, over-emphasis has been placed on algorithmic procedures in many computer-aided design packages, thus producing a large gap between model developers and users. This may produce inferior design and cause the under-utilization, or even total failure of these models. This is the Pre-Published Version.