Fuzzy Rule Based Prediction System of Surface Damage in the Laser Cleaning Process

During laser cleaning, surface damage on the substrate could easily occur by overexposure to a laser pulse of high fluence. Successful cleaning without surface damage can be achieved by skilled operators with long experience and good technique. This paper presents a fuzzy rule based expert system to predict surface damage during laser cleaning in the same way as a human expert. In this work, a fuzzy rule base was used to embed the acoustic information, which gives an indication of both progress in cleaning and the result. This is achieved by a wide-band microphone. An inference process was conducted to predict whether and how much surface damage would be induced on the substrate. Tests of the performance of the fuzzy system showed that the prediction of surface damage is well correlated with the actual results, independent of initial surface conditions. Based on the prediction of surface damage, the process control algorithm of laser cleaning has been developed.