A Novel Self-organising Neural Network for Control Chart Pattern Recognition

This paper describes a novel self-organising neural network. The network uses a new type of firing criterion that takes into account the individual components of the patterns to be clustered. To demonstrate the properties of the network, the results of applying it to the classification of control chart patterns arc presented. The training and test data sets for the network cover the 6 most common categories of control chart patterns, which are: Normal, Cyclic, Increasing Trend, Decreasing Trend, Upward Shift and Downward Shift. The classification results obtained show that the firing criterion adopted is superior to the usual Euclidean criterion.