Noise tolerance for pattern recognition for statistical process control

This paper describes an investigation into the performance of an algorithmi c pattern recognition system for statistical process control applications. The study, based on the application of simulation techniques, investigates whether the system can recognise patterns in the presence of noise, and how much noise the Pattern Recogniser can accept in relation to individual pat terns. The level of noise tolerance for each pattern is analysed using simu lated data. The simulation procedures are presented and the results are dis cussed. The pattern recognition system was found to be tolerant of noise, t he level of tolerance generally depending on the pattern specification.