Process Quality Optimization Model Based on ARM and Immune Principle

Facing with increasing international competition, how to reduce process variability is always one of the major concerns of manufacturing organization. The purpose of this paper is to investigate a data driven optimization model: Process Quality Optimization Model (PQOM) based on Association Rules Mining (ARM) and immune principle to support both staticand dynamic optimization of process quality. Realization of PQOM consists of two stages. First, ARM is used to analyze historical SPC data to explore the explicit, hidden process input-output mapping relations that affect final product quality. Then, based on excavated rules, negative selection algorithm inspired by natural immune system is introduced toanalyze on-line monitoring data for dynamic quality control. Two types of rule are utilized, namely Rules For Process Optimization (RFPO) and Rules For Exception Detection (RFED). RFED is used as detector to target process exception, while RFPO is used as immune antibody that reacts to process exception. The PQOM model is tested in a squeeze casting enterprise to verify its feasibility and correctness.