Predicting coupling signals in a material flow real-time co-simulation with a Kalman filter

A hardware-in-the-loop simulation has to use the control cycle as simulation clock. This time-deterministic clocking can be difficult to meet for complex behavior models in a real-time simulation. To accelerate a real-time simulation, the calculation can be parallelized by co-simulation whereby multiple, differently detailed behavior models can be simulated parallelly. In these multi-scale models, a slower, detailed model guides a fast, abstract model. In this work, a Kalman filter is examined as a multi-rate method to predict the signals of a physical-based model for a material flow simulation in real-time. To improve the results, smoothing and boundary conditions are added.