An Improved Affinity Propagation-Based Typical Process Route Discovery Approach Considering Clustering Effectiveness

With the increasing development of CAPP (Computer Aided Process Planning), an urgent requirement has arisen on how to extract reusable process knowledge effectively. Under this situation, an improved AP (affinity propagation) based clustering approach has been proposed to discover typical process routes from massive historical data under the consideration of clustering effectiveness. Because of the practical requirement of typical process routes, two soft constraints (i.e., quantity constraint and size constraint) have been introduced simultaneously in the clustering process to guarantee the quality and effectiveness of the typical process routes. In doing so, the traditional AP has been improved with two extra function nodes, which represent the two soft constraints mentioned before respectively. Finally, the results of numerical illustrations indicate that the improved AP can find a tradeoff between these two conflicting soft constraints and have better performance compared with commonly used clustering methods for typical process route discovery.