TR058: Integrated Case-Based Reasoning for Structural Design

Recent knowledge-based expert systems for structural engineering design have focused on case-independent knowledge (abstract reasoning rules for designing), and while great strides have been made in that area, there is still a significant need to develop systems to take advantage of the wealth of knowledge contained in every substantial structural design. On the other hand, previous database-oriented design efforts have focused primarily on knowledge-poor databases of solutions, in which the traditional engineering handbook of solutions has simply been replaced by digital data. The challenge is to find a way to capture and apply the kind of case-dependent knowledge that structural engineers have traditionally used.

This thesis examines the new approach of implementing knowledge-based structural design systems using both case-dependent and case-independent knowledge. The resulting system, DDIS combines case-based reasoning with case-independent reasoning in a blackboard framework. In the blackboard model, the knowledge needed to solve a problem is partitioned into independent knowledge sources that are grouped into several knowledge modules in the knowledge base. The knowledge sources communicate design results via a global knowledge structure (the blackboard) and respond opportunistically to the changes on the blackboard. DDIS has two major knowledge modules: case-dependent and case-independent. The case-independent module represents abstract knowledge about the problem domain and problem solving strategies. The case-dependent module uses case-based reasoning techniques to transfer knowledge from previous designs to current design tasks. Using the blackboard control mechanism and the two knowledge modules, DDIS can apply both case-dependent and case-independent knowledge to perform collaborative and opportunistic design.

The most important elements of case-dependent knowledge identified in this study are design solutions, justifications, constraints, failures, plans and goals. Design solutions include final solutions, intermediate solutions and partial solutions. Design justifications are the calculations of previous design variables and the dependences of their values. Design constraints are used to evaluate designs and are very important information for understanding the history of a design case. Previous design failures can be used to avoid unsuccessful design alternatives in the future. Design plans are the strategies used to solve a design problem. Specific knowledge about how to achieve a particular design step is contained in a design goal. Design goals are included in plans to form complete case-dependent control knowledge of a case.

DDIS has a very flexible architecture and representation that can use any subset of the above mentioned case-dependent knowledge. Past design solutions can be applied to very similar new designs while previous design plans can be applied to guide designs with less surface similarity. However, this study did not address the similarity problem. DDIS relies on users to retrieve relevant designs from the case memory and to decide how similar they are to the new design.

Two structural design applications have been built using DDIS to demonstrate its integrated design approach. The demonstration applications design structural steel beam-columns and anchor base plates for electrical transmission poles.