Where traditional methods of Robust Design have their limitations in manufacturing systems, it appears that system modelling can provide advantages. When manufacturing systems can be split into subsystems, subsystem models and continuity and compatibility constraints are required to develop an overall system model. Where physical information is lacking, empirical models can be created. Herein, the Radial Basis Function Neural Network method for empirical modelling is discussed and procedures for connecting these models with physical system information is demonstrated using a cooling fin problem.

There has been a great deal of interest in quality and reliability in recent years. Researchers have defined these 'system' measures specifically for use within their discipline. However, some of these definitions are incompatible when applied to other areas. The purpose of this paper is to define quality and reliability for use in all disciplines. By using a discipline-independent method to define these measures, a working definition of quality and reliability can be found. Linear graph models will help establish the definition and relationship.