simulator.tex

由matlab开发的hybrid系统的描述语言

\chapter{Simulator}
\label{simulator_c}

 Given an initial state and a sequence of inputs,
the \MLD{} system determines the behavior of the hybrid system. In
this scenario, solving the \MLD{} system to get the output and the
next state is overkill. AD and DA items can be evaluated directly
using a \matlab{} `if then else' construct. The problem with this
approach is the execution order: In the \hysdel{} source the order
of the statements does not matter. Neither does it when
translating into a \MLD{} system. But when these statements are
directly evaluated, it is important that all auxiliary variables
on the right hand side are known when computing the left hand
variable value. Thus, we need an order in which the statements are
to be evaluated. \method{Symbol\_table}{find\_computable\_order}
finds such an order.

\method{Symbol\_table}{find\_computable\_order} sets the \code{computable\_order} attribute of \cppclass{Symbol}.
Variables with \code{computable\_order} zero are not computable (e.g. because they are not defined). Variables
with \code{computable\_order} one depend on no other variable and can be computed immediately. Variables with
\code{computable\_order} two can be computed as soon as all variables with computable order one are computed. A
translation into a \matlab{} simulator is only possible if all variables are computable.

\method{Symbol\_table}{find\_computable\_order} does not distinguish between state or output variables and
auxiliary variables. The simulator on the other hand has to; if an auxiliary variable depends on a state variable,
it depends on the previous state and can be computed before updating the state. Thus, the simulator first computes
the auxiliary variables in the order given by \code{computable\_order} and then computes all state and output
variables.

In every step the simulator checks that all constraints from the must section are fulfilled and that no bound (for
AD and DA items) is exceeded.