Process Variation Identification Using Small Recognition Window Size


  • Ibrahim Masood Universiti Tun Hussein Onn Malaysia
  • Azray Idiz Azizi


Statistical Process Control, Multilayer Perceptron, Recognition Accuracy, Statistical Features


Identification of unnatural mean shifts variation is challenging when involve with two or more correlated variables (bivariate). Statistical process control (SPC) is applied for improving product quality through statistical. The number of samples used is small window of size. Just-in-time (JIT) is applied because it required a small batch of sample to ensure the production is keep on running once have a demand. The scheme that used for identified the out-of-control variation is Statistical Features-Multilayer Perceptron (SF-MLP). Recognition accuracy (RA) was used as the performance measures. The best statistical features obtain from this study are Maximum and minimum values of each variable (Minmax), mean, slope, sinus, vector, and last value of exponentially weighted moving average (LEWMA) with recognition accuracy average in between (RA = 85 ~ 96 %).




How to Cite

Masood, I., & Azizi, A. I. (2021). Process Variation Identification Using Small Recognition Window Size. Research Progress in Mechanical and Manufacturing Engineering, 2(1), 48–55. Retrieved from