A knowledge cell approach to processing design information

The aim of the knowledge cell approach reported here has been to create a prototype wide-purpose system to tackle relatively straightforward design analyses in mechanical engineering. In the present work, design analyses are carried out by performing a combination of equation solving and table selection activities, and at the same time, satisfying certain user-specified design constraints. These activities are automatically organised based on the captured design information and, depending on the situation, the system can automate certain design processes. Calculated results are then used to produce final geometric models. In this paper the knowledge cell approach is reviewed. Two examples are presented. One demonstrates tests in the design domain of machine tool spindle-bearing assemblies, whilst another relates to gearbox representations. Here one type of knowledge cell (the G-cell or graphics cell) is discussed and a series of tests to assess this cell is presented. Whilst the G-cells show some limitations when the complexity of the generic model is increased, satisfactory results can be achieved with the present CAD system when handling models of complexity appropriate to the aims of the research. All processes running in the associated intelligent environment are seen to be very effective and efficient.