Real Time Simulation Using Non-causal Physical Models 2009-01-1021
As automotive electronics become more complex and more distributed, hardware in-the-loop simulation is now a widely adopted technique for performing controller software/hardware integration testing as well as controller/controller integration testing. Having real-time capable models that are correlated to physical hardware being controlled is key to successful implementation of hardware in-the-loop testing. Because models for hardware in-the-loop must be developed in a short amount of time and then stay in sync with the design through design changes, a best practice is to obtain such models from the system-level model used for requirements analysis and design trade offs. This way, one model can address the need of both requirements analysis and integration testing, reducing redevelopment of models and ensuring consistency between two process steps. While there has been significant progress made in recent years on real-time simulator technologies, including I/O accuracy, use of off-the-shelf hardware, acceleration using parallel processing, the process by which a system level simulation model is to be reused for hardware in-the-loop testing is not very well understood. This paper starts by examining options for developing a system level simulation model. When limited to causal modeling techniques, the process of creating models is often cumbersome and time-consuming. Many engineers find non-causal (or acausal) modeling methods to be much more intuitive. However, getting acausal models to run in real-time requires careful upfront planning and, when required, methodical reduction. The remaining sections of the paper deal with effective techniques for physical model development and reduction.