问题
I am looking for an alternative to Simulink that would allow to control the stepping mechanism. The same concern was raised in this question, that Simulink does not allow backwards stepping, so that complex systems x(t,u) where "changing variables" (t <- -t) requires weeks of work, cannot be controlled (u) correctly due to interdependance between past states and present decisions (u=f(x(s)),s<=t).
Conceptually, the trick simply consists in manually storing previous states in memory : for a Simulink user, this is difficult and error prone, whereas really simply stepping backwards in time would be simple and fit nicely in Simulink, from the developer's point of view.
来源：https://stackoverflow.com/questions/43070168/time-driven-simulation-backwards-steps