Application of Matlab to Injection Molding Quality Control

Injection molding is the most common manufacturing process for production of high volume commercial parts. However, injection molding is characterized by extremely complex process dynamics and material properties, which make it difficult to understand and predict the product quality and can lead to numerous unacceptable quality defects. There are many solutions to these problems – Moldflow is recognized as the leading global provider of valueadded solutions to the plastics industry. This paper describes the application of Matlab numerical software in Moldflow’s development process, focusing on prototype development and validation of an on-line quality control system for injection molding. As a specific example, one module of this quality control system is discussed in detail. This problem is unique in that quality is multi-dimensional and discrete (not continuous), the process is stochastic (not deterministic), and that quality data is non-uniform and sparse. The described module, Process Window, examines the local multi-dimensional quality surface using statistical methods. Process Window then automatically searches the process space and efficiently locates a defect free process zone suitable for production. Finally, the performance of the Process Window prototype is compared to the Matlab function fmins(), a multi-objective optimization which uses a Simplex search method.