Neuronal Process Modeling

The deep learning infrastructure of the neuronal engine is tailored to specific sensor modalities to robustly identify knowledge relevant for efficient process realization. As streams of sensor data are fed into our AI engine, trained deep networks provide results on-site or in the private cloud. While desirable knowledge-based behaviors are monitored and optional corrective actions can be automatically triggered via workflow systems, delays in as-is processes can be avoided and quality targets met.

Neuronal Process Simulation

Our AI engine learns accurate predictive models of processes and "digital twins" directly from high-dimensional time series. The data can come directly from existing real-world processes and systems. In the high-fidelity, efficient and safe simulation environment, deep networks trained to mimic them yield valuable new insights into corrective actions applied in simulations and the performance to be expected. Here, third-party products can be integrated in addition. As inefficiencies in real-world design become visible, deep networks can be educated in a goal-oriented manner and best corrective actions can be identified.

Neuronal Process Optimization

Our AI engine trains neuronal networks to optimize the digital twin's inner and outer processes through Deep Supervised Learning, which results in new and robust process designs being on a higher, global performance level. These optimize economic criteria and minimize costs, save time, increase quality, etc. The new or modified to-be processes finally are deployed to target systems and improve real-world processes and systems.