Tool condition monitoring for the chipboard drilling process using automatic, signal-based tool state evaluation

An automatic approach to tool condition monitoring is presented, with the best solution achieving overall accuracy of 94.33% and 9 misclassification errors. In wood industry, cutting tools need be evaluated periodically. This especially case when drills are concerned; since dulled, resulting poor-quality product may generate loss for manufacturing company, due discard it during quality control. Each can classified either as useful or useless, second type should exchanged fast possible. Manual evaluation time consuming, which results in production downtime. problem requires a faster, automated, precise work environment. response this issue, an ensemble algorithm was developed. Different signals were collected input data, including feed force, torque, noise, vibrations, acoustic emission. Based on those signals, set 152 initial features generated, while after feature selection 19 them used by classifiers. algorithms tested terms number The classifiers prepare solution, able classify accurately, very few errors between recognized classes.