ABSTRACT
Unnatural patterns on manufacturing process control charts can reveal potential quality problems due to assignable causes at an early stage, helping prevent defects and improve quality performance. In recent years, artificial neural networks (ANNs) have been applied to the pattern recognition task for control charts. The results are promising. However, almost all researches in this area assume that the in-control process data in the control charts follow a normal distribution. This assumption often controverts the practical manufacturing situations. This research examines the effects of non-normality on the performance of ANN based control chart pattern recognition systems. Extensive simulation study with various non-normality was carried out to evaluate the performance of the ANN systems. The hypothesis tests were employed to evaluate the significance of the numerical results. Simulation results indicate that the larger the absolute value of skewness, the lower the recognition accuracy, and the larger the kurtosis, the higher the recognition accuracy. Compared to skewness, the influence rate of kurtosis on recognition accuracy is larger (about 1.3 times). Moreover, skewness will impair the stability of the recognition speed.