Comparison and Application of Multi-Rate Methods for Real-Time Simulations of Production Systems

Abstract

A distributed simulation makes it possible to couple simulation tools and lays the foundation for the usage of multicore capabilities to decrease the calculation time. In consequence the simulation is partitioned on multiple simulation tasks. If simulation tasks with different integration step sizes are used, the configuration is called multi-rate simulation. In real-time simulations the tasks are calculated parallely, which means that fast tasks do not wait for the simulation result of slower tasks. A sequential approach where fast tasks wait for slower tasks would slow down the overall simulation and therefore tear the real-time requirements. In a real-time multi-rate approach, signal processing of the coupling signals between the tasks is required. For this signal processing, multi-rate methods are used. Easy multi-rate methods lead to stepped signals in the faster task, because the slower task does not provide a new calculated signal at each timestep of the faster task. In this work further methods are investigated in an industrial real-time simulation environment. The analysis contains continuous and discontinuous as well as energy conserving methods. It is shown how these different methods perform for various kinds of signals. The methods are compared and evaluated on signals with different characteristics, which allows a recommendation for the choice of a method in a specific simulation scenario. The application of the multi-rate methods is shown on an example virtual commissioning simulation of an industrial robot. It shows that the right choice of a multi-rate method has a big impact on the overall simulation result.