Recognition of quality control chart patterns based on back propagation neural network

In the off-line monitoring of the control characteristics of manufactured parts/products has usually been performed by establishing suitable quality control charts. The control charts as maintained would usually show various patterns and the analysis of such patterns would require human expertise so that suitable remedial actions can be taken. In the present paper, a maiden venture has to be taken to recognize various observable quality control chart patterns through the application of back propagation neural network technique. Various patterns like normal, cyclic, increasing/decreasing trend, upward/ downward shift etc. have been taken into consideration. Various known patterns have also been generated through the simulation and induction approach. The developed module has been tested and observed to be quite fast, accurate and flexible as it can be able to cope up with varying quality control chart patterns as encountered in real time manufacturing environment.