Systems modelling and machine learning

Dynamic system modelling simulates a controllable system that may include hydraulics, mechanics, electronics, and almost anything else that can be mathematically modelled. By calibrating the model with data retrieved from the real environment, the control unit can be optimised and the functions tested through simulation. The method allows for holistic problem solving, where all the design teams can focus on finding the best possible result together.

Model-based software development shifts the emphasis from coding to design and development of the actual product. The algorithms and software architecture used in measuring and controlling can be generated directly from the design model.

Based on the models, simulation solutions can be built for services, business environments, versatile user groups, ecosystems, or a range of different services. These simulations enable testing and test automation even when the actual device is not available. This is often referred to as HIL (Hardware In the Loop) testing.

Machine learning allows for large amounts of data to be handled, it can teach the system to recognize structures and dependencies, and to spot anomalies. Machine learning can parse the space surrounding the machine and based on this, separate the high-level goals into a series of decisions, out of which actions can be created. The machine’s control unit will be designed in such a way that it can carry out these actions in real-time with a precision that requires lots of very fine adjustment techniques and detailed communication between the different control units.

All this will result in higher quality

• Digital prototype before the real prototype
• Test automation in a simulated environment
• Higher quality and higher safety levels through formalised design methods