Use of artificial neural networks for the monitoring of screw insertions

The automation of screw insertions represents a highly desirable task. An important part of the automation process is the monitoring of the insertion, The paper presents an application of artificial neural networks for monitoring this common manufacturing procedure. The research focuses on the insertion of self-tapping screws. Artificial neural networks have been employed to distinguish between successful and failed insertions. The networks under investigation use radial basis functions for the computation of the data. A range of networks, differing in size, has been implemented and thoroughly tested. Results and evaluations of the networks from the experiments are presented.