An architecture for client-server based control chart patterns recognition system

Abstract

In any industrial process, regardless of how well designed or carefully maintained it is, a certain amount of process variation will always exist. One of the commonly used methods to monitor and diagnose the process variation is Statistical Process Control (SPC) chart. Some enhancements have been made by researchers to introduce the SPC chart system with patterns recognition feature. However, these SPC chart patterns recognition systems are limited to local and standalone system. There is a need to develop a client-server based system which would enable data sharing and remote access. The purpose of this study is to develop an architecture for a client-server based control chart patterns recognition (CCPR) system. Specific focus of this study is to investigate design and development aspects of the proposed client-server based CCPR system. A suitable client-server (CS) architecture and load passing procedure are investigated. Proper selection of CS architecture and load passing procedure are crucial to ensure each functional tier can be effectively integrated, system resources can be optimized and the system workload can be fairly distributed. The selection of design parameters are based on computer simulation studies using synthetic data. Candidate designs are evaluated based on calculation time, memory, processor usage and classification accuracy. The CS architecture with load passing procedure at server tier and patterns recognition at engine tier gave the best results among the investigated alternatives. This CS architecture is able to support more users and process more data. The load passing procedure with the combination of Nelson Run Rules and EWMA test resulted in better performance compared to Run Rules and CUSUM, CUSUM and EWMA and capable to prioritize the unstable processes to access the engine tier. The proposed CS architecture and load passing procedure can be used as a guideline for the development of a clientserver based CCPR system.