TF3800 | TwinCAT 3 Machine Learning Inference Engine

The TF3800 TwinCAT 3 Function is a high-performance execution module (inference engine) for trained, conventional machine learning algorithms.

Beckhoff offers a machine learning (ML) solution that is seamlessly integrated into TwinCAT 3. This ensures ML applications can also benefit from the familiar advantages of system openness found in PC-based control thanks to the use of established standards. As an added bonus, the machine learning models are executed in real time, providing machine builders with the ideal foundations for improving machine performance.

The algorithms are trained in a wide variety of established frameworks, such as SciKit-Learn, libSVM, and XGBoots. The AI model created is exported from the learning environment as an ONNX file. ONNX (Open Neural Network Exchange) has asserted itself as an open standard for interoperability in machine learning, ensuring a clear distinction between the learning environment and execution environment of trained models.

The ONNX file can be read into TwinCAT 3 and supplemented with application-specific meta information, such as the model name, model version, and a brief description. The TwinCAT 3 PLC provides a function block that loads the AI model description file and executes it on a cycle-synchronous basis. These loading and execution processes are implemented as methods of the function block, which ensures the AI model is deeply integrated into the machine application.

The product supports the execution of a variety of different model types, ranging from k-means, SVM, and PCA through to decision trees and ensemble trees such as Random Forest, XGBoost, and LightGBM.

The range of applications for conventional machine learning algorithms is broad. They are often used for classification tasks, such as quality control, process monitoring, and anomaly detection.