DefinePK

Abstract

Statistical process control (SPC) chart for variable is a powerful tool, which has been widely implemented for quality control in precision parts manufacturing. It is known to be effective in analyzing whether a manufacturing process lies within a stable or an unstable condition. In current practice, the conventional SPC chart will only detect an unstable process based on one point out-of-control. Unfortunately, this situation is too late for avoiding defective parts and leading to increase waste of materials. To overcome this issue, various studies have been focused on designing the SPC schemes based on control chart pattern recognition (CCPR) method. This advanced SPC scheme has improved the speed for detecting an unstable condition. Nevertheless, a broad of set studies in this area revealed that synthetic SPC samples have been utilized in analyzing the control chart patterns, which is limited to common causable patterns. In this research, a methodology to design a CCPR scheme using an original SPC data has been studied. Based on a case study in metal-stamping operation, the study involves (i) an identification of the unnatural variation for the critical-to-quality variables and (ii) an identification of the design elements for CCPR scheme. The sources of unnatural variation are investigated based on man, method, material, and machine. The CCPR scheme is designed using an artificial neural network (ANN) recognizer model. This methodology will be useful for industrial practitioners in identifying the root cause error in stamping-based operations based on its specific SPC chart pattern.