With the growing employment of automatic data-collection methods and the enhancements on computerised plotting on control charts, a demand exists to automate the analysis of process data. Computerised recognition techniques can provide an actual alternative to conventional methods for analysing control charts with little or no human intervention. In this paper, a neural network approach is discussed and applied to trend-pattern recognition on control charts. In the proposed approach the neural network is trained to recognise both "natural" and distribution of points. Experimental results are compared to a combined Shewhart-CUSUM approach in terms of Average Run Length (ARL).
A special-purpose neural network recogniser to detect non-random pattern on control charts
ANGLANI, Alfredo;PACELLA, Massimo;
2001-01-01
Abstract
With the growing employment of automatic data-collection methods and the enhancements on computerised plotting on control charts, a demand exists to automate the analysis of process data. Computerised recognition techniques can provide an actual alternative to conventional methods for analysing control charts with little or no human intervention. In this paper, a neural network approach is discussed and applied to trend-pattern recognition on control charts. In the proposed approach the neural network is trained to recognise both "natural" and distribution of points. Experimental results are compared to a combined Shewhart-CUSUM approach in terms of Average Run Length (ARL).I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
