High-Speed Real-Time Simulation

Summary form only given. Real-time simulation, in which the behaviour of a simulation of a system develops in precisely the same time-scale as the corresponding behaviour of the system itself, has been used for many years for testing hardware (hardware-in-the-loop or HIL), software and for training (e.g. flight simulators, power-plant simulators). The two distinguishing features of real-time simulation are synchronization with the actual behaviour via a real-time clock and an interface system that allows the external systems to operate just as if they were connected to the real system. One of the key characteristics of a real-time simulation is the frame time which determines the discrete time increments by which a simulation is advanced each time the next state of the system is calculated. Many real-time simulations require frame times of the order of milliseconds, and modern realtime operating systems, which typically involve latencies of a few microseconds, are quite capable of supporting such applications. Some applications, however, such as in power electronics and automobile engines require frame times of less than 10 muS. Small arrays of digital signal processors on PCI cards offer an efficient way of meeting this requirement and have been demonstrated to support simple applications at frame times as low as 2 muS. Another promising approach for even shorter frame times is to use FPGAs, which promise sub-microsecond frame times. Many applications require that the high-speed real-time (HSRT) simulation is embedded in a larger simulation of an entire system with less stringent frame-time requirements, and multi-rate real-time simulation is another current research interest. The presentation will review these developments and ongoing research in this area