Tool Wear Length Estimation with a Self-Learning Fuzzy Inference Algorithm in Finish Milling

The geometric accuracy and surface roughness are mainly affected by the flank wear at the minor cutting edge in finish machining. A genetic algorithm-based fuzzy estimator obtained by a fuzzy inference algorithm to evaluate the minor flank wear length in finish milling is introduced. The fuzzy inference rules are trained by genetic algorithms (GA) through practice. Fuzzy membership functions and rules are usually decided upon subjectively. In this paper, the performance of the fuzzy estimator may be improved if the fuzzy inference model is supplemented by a genetic-based learning mechanism. The features sensitive to minor flank wear are extracted from the dispersion analysis of a time series AR model of the feed directional acceleration of the spindle housing. Linguistic rules for fuzzy estimation are constructed using these features, and then fuzzy inferences are carried out with test data sets under various cutting conditions. The proposed system turns out to be effective for estimating minor flank wear length, and its mean error is less than 13%.