Tool monitoring for drilling process applying enhanced neural networks

Detection of cutting tool wear is vital in automated manufacturing. It helps improving and increasing manufacturing productivity. This work considers the monitoring of cutting tool wear for drilling process investigating the use of genetic algorithms for identifying near optimal design parameters of diagnostic system that are based on artificial neural networks for condition monitoring of mechanical systems. Genetic Algorithms help identifying the most useful features for an efficient classification as opposed to using all features from all input sensors, leading to very high computational cost and is, consequently, not desirable. It is shown that GA can be used to select a smaller subset of features that together form a genetically fit family for successful fault identification and classification tasks. The objective of the improved system is to have a fast response time at a relatively cheap cost, while providing a warning in advance of potentially developing faults.