Automatic solver selection and solution for CPU based automotive HIL simulations with small step size requirements

Abstract:
More and more electronic control units (ECU) are being integrated in the rapidly evolving automotive industry. In order to simplify and shorten the testing of ECU algorithms, hardware-in-the-loop (HIL) testing is becoming increasingly popular in the industry. For several applications common CPU based high level simulation software is not able to achieve a suitable temporal resolution in real-time. This paper presents an easy to use model representation with an integrated automatic solver selection. Therefore it is possible to simulate models with ideal solver and step size. The selection algorithm uses an automatic parameter-dependent model and run time analysis and can also be applied to non-linear differential algebraic equation systems with switching conditions by local linearization of the complete state space. In an existing automotive HIL-simulation the presented procedure was proven to allow an easy implementation of multi-state, nonlinear, non-continuous, ordinary differential algebraic equation systems. With the implementation in C++ a step size of up to 5 microseconds von a low power compact computer could be achieved.