At Hannover Messe 2019, Fabian Bause, Beckhoff’s TwinCat Product Manager, shows how Beckhoff has integrated machine learning into its TwinCat software. The company says it is currently the only automation technology supplier to integrate an inference engine into an industrial controller for machine learning. Bause noted that, because TwinCat machine learning uses the Open Neural Network Exchange (ONNX) format, files can be exchanged easily from any platform like Matlab, Pytorch or TensorFlow to TwinCat. In this video, Bause shows how data from a functional fan can be used to train TwinCat to recognize anomalies in fan function. Beyond basic anomaly detection, Bause said the algorithm can also be used to classify different kinds of faults to facilitate more precise troubleshooting. “You can even use it in motion control applications, to save energy in motion synchronization applications,” he added.
David Greenfield, editor in chief | Editor in Chief
David Greenfield joined Automation World in June 2011. Bringing a wealth of industry knowledge and media experience to his position, David’s contributions can be found in AW’s print and online editions and custom projects. Earlier in his career, David was Editorial Director of Design News at UBM Electronics, and prior to joining UBM, he was Editorial Director of Control Engineering at Reed Business Information, where he also worked on Manufacturing Business Technology as Publisher.
