Control Chart Pattern Recognition Using Artificial Neural Networks
Precise and fast control chart pattern (CCP) recognition is important for monitoring process environments to achieve appropriate control and to produce high quality products. CCPs can exhibit six types of pattern: normal, cyclic, increasing trend, decreasing trend, upward shift and downward shift. Except for normal patterns, all other patterns indicate that the process being monitored is not functioning correctly and requires adjustment. This paper describes a new type of neural network for speeding up the training process and to compare three training algorithms in terms of speed, performance and parameter complexity for CCP recognition. The networks are multilayered perceptrons trained with a resilient propagation, backpropagation (BP) and extended delta-bar-delta algorithms. The recognition results of CCPs show the BP algorithm is accurate and provides better and faster results.
Multilayered Perceptrons, Resilient Propagation, Backpropagation, Extended Delta-Bar-Delta, Control Chart Pattern Recognition.