Applying emerging soft computing approaches to control chart pattern recognition for an SPC–EPC process

Abstract

One of the primary tasks for process personnel is to detect and identify the underlying process disturbances such that they can be quickly removed. Most research has concluded that the integration of statistical process control (SPC) and engineering process control (EPC) is an effective way to achieve this task. Although this integration may lead to many benefits, it could result in problems with control chart pattern recognition. The EPC adjustments could cause the underlying disturbance patterns to be embedded in the control chart, thus dramatically increasing the degree of difficulty to identify the behavior of process disturbances. This study considers a zero-order autoregressive and integrated moving average process (ARIMA) that contains five common process disturbances. In addition, the minimum mean squared error (MMSE) control actions serve as the role of the EPC. In contrast to using the conventional soft computing methods, this study proposes two emerging soft computing techniques, extreme learning machine (ELM) and random forest (RF), to address the difficulties for recognition of embedded disturbance patterns in the control charts. Experimental results revealed that the proposed approaches are able to effectively recognize various disturbance patterns of an SPC–EPC process.