Performance Boost in hydraulics thanks to machine learning

ATP Hydraulik and CSEM have achieved a breakthrough in the control of hydraulic systems. Machine learning can massively improve systems.

Hydraulic systems are widely used. However, the precise control - positioning in time and place - of the actuators is challenging. For example, the systems are very sensitive to temperature fluctuations. Initial setup also requires a recurring effort for each system due to manufacturing tolerances.

With our collaboration, CSEM has created a complete software solution for automated tuning and hydraulic system control. At the core are algorithms with machine learning. Control parameters are created with Bayesian optimization and Gauss process in real time. Precise synchronization of multi-axis systems also becomes possible. The VISARD software solution provides extensive and expandable configuration systems. VISARD also realizes the connection to the control via a Beckhoff system.

This Projekt was realized in collaboration with Industrie2025.ch and CSEM.