M .task/backlog.data M .task/completed.data M .task/pending.data M .task/undo.data A Presentations/ERLM/images/4_research_approach/strategy.png M Presentations/ERLM/main.aux M Presentations/ERLM/main.fdb_latexmk M Presentations/ERLM/main.fls
190 lines
5.5 KiB
TeX
190 lines
5.5 KiB
TeX
\begin{frame}{The goal of this research is to create verified autonomous control
|
|
systems}
|
|
|
|
\textbf{\underline{
|
|
If this research is successful, we will be able to do the following:
|
|
}}
|
|
\pause
|
|
|
|
\begin{enumerate}
|
|
|
|
\item \alert<2>{Translate written procedures into discrete control logic}
|
|
\pause
|
|
|
|
\item \alert<3>{Verify continuous control behavior across discrete mode
|
|
transitions}
|
|
\pause
|
|
|
|
\item \alert<4>{Demonstrate autonomous reactor startup with verifiable safety
|
|
guarantees}
|
|
|
|
\end{enumerate}
|
|
\end{frame}
|
|
|
|
\begin{frame}{First, we will formalize written procedures into logical
|
|
statements}
|
|
|
|
\begin{columns}[c]
|
|
\begin{column}{0.7\textwidth}
|
|
\includegraphics[
|
|
width=1.0\textwidth,
|
|
trim={1cm 22cm 1cm 0cm},
|
|
clip
|
|
]
|
|
{images/4_research_approach/procedure.png}
|
|
\scriptsize{
|
|
Westinghouse Technology Systems Manual, Section 19.0 - Plant Operations
|
|
}
|
|
\end{column}
|
|
|
|
\begin{column}{0.3\textwidth}
|
|
\only<2>{
|
|
\begin{alertblock}{FRET Specification}
|
|
\scriptsize
|
|
\texttt{INITIAL\_CONDITIONS shall satisfy:}
|
|
\begin{itemize}
|
|
\item[] \texttt{mode = MODE\_5}
|
|
\item[] \texttt{k\_eff < 0.99}
|
|
\item[] \texttt{power = 0}
|
|
\item[] \texttt{t\_avg < 200}
|
|
\item[] \texttt{...}
|
|
\end{itemize}
|
|
\end{alertblock}
|
|
}
|
|
|
|
\only<3->{
|
|
\begin{block}{FRET Specification}
|
|
\scriptsize
|
|
\texttt{INITIAL\_CONDITIONS shall satisfy:}
|
|
\begin{itemize}
|
|
\item[] \texttt{mode = MODE\_5}
|
|
\item[] \texttt{k\_eff < 0.99}
|
|
\item[] \texttt{power = 0}
|
|
\item[] \texttt{t\_avg < 200}
|
|
\item[] \texttt{...}
|
|
\end{itemize}
|
|
\end{block}
|
|
}
|
|
|
|
\only<3>{
|
|
\begin{alertblock}{LTL Formula}
|
|
\scriptsize
|
|
\texttt{$\Box$ (initial $\rightarrow$ (}
|
|
\begin{itemize}
|
|
\item[] \texttt{mode\_5\_active $\land$}
|
|
\item[] \texttt{k\_eff\_subcritical $\land$}
|
|
\item[] \texttt{zero\_power $\land$}
|
|
\item[] \texttt{temp\_safe $\land$}
|
|
\item[] \texttt{...))}
|
|
\end{itemize}
|
|
\end{alertblock}
|
|
}
|
|
\end{column}
|
|
|
|
\end{columns}
|
|
|
|
\end{frame}
|
|
|
|
\begin{frame}{Second, we will use reactive synthesis to convert the logical formulae to generate discrete
|
|
automata}
|
|
|
|
\begin{columns}
|
|
\begin{column}{0.5\textwidth}
|
|
\begin{tikzpicture}
|
|
\node (chess) at (0cm,0cm){\includegraphics[width=0.4\textwidth]{images/4_research_approach/strategy.png}};
|
|
\node[draw=textcolor, fill=secondary, left=0.1cm of chess, yshift=-0.8cm] {Plant};
|
|
\node[draw=textcolor, fill=primary!30!red, right=0.1cm of chess, yshift=0.4cm] {Controller};
|
|
\end{tikzpicture}
|
|
\end{column}
|
|
\begin{column}{0.5\textwidth}
|
|
\pause
|
|
\begin{tikzpicture}[
|
|
state/.style={circle, draw=textcolor, fill=secondary, minimum size=1.2cm},
|
|
transition/.style={->, >=stealth, thick}
|
|
]
|
|
|
|
% States arranged vertically
|
|
\coordinate (start) at (0,1.5);
|
|
\node[state] (s0) at (0,0) {\scriptsize MODE 5};
|
|
\node[state] (s1) at (0,-2) {\scriptsize MODE 4};
|
|
\node[state] (s2) at (1,-4) {\scriptsize MODE 3};
|
|
\node[state] (s3) at (3,-5.) {\scriptsize MODE 2};
|
|
\node[state] (s4) at (5,-6) {\scriptsize MODE 1};
|
|
\node[state] (s5) at (3,-1.5) {\scriptsize SCRAM};
|
|
|
|
% Transitions
|
|
\draw[transition] (start) -- node[midway, right] {\scriptsize Start} (s0);
|
|
\draw[transition] (s0) -- (s1);
|
|
\draw[transition] (s1) -- (s2);
|
|
\draw[transition] (s2) -- (s3);
|
|
\draw[transition] (s3) -- (s4);
|
|
\draw[transition] (s2) to[bend right] (s5);
|
|
\draw[transition] (s1) to[bend right] (s5);
|
|
\draw[transition] (s3) to[bend right] (s5);
|
|
\draw[transition] (s4) to[bend right] (s5);
|
|
|
|
\draw[transition] (s0) to[loop left] (s0);
|
|
\draw[transition] (s1) to[loop left] (s1);
|
|
\draw[transition] (s2) to[loop left] (s2);
|
|
\draw[transition] (s3) to[loop below] (s3);
|
|
\draw[transition] (s4) to[loop below] (s4);
|
|
|
|
\end{tikzpicture}
|
|
\end{column}
|
|
\end{columns}
|
|
|
|
\end{frame}
|
|
|
|
\begin{frame}{Finally, we will build continuous controllers to move between
|
|
discrete states}
|
|
|
|
\begin{columns}
|
|
\begin{column}{0.5\textwidth}
|
|
\includegraphics[height=0.7\textheight]{bouncing_ball_hybrid.png}
|
|
|
|
\end{column}
|
|
|
|
\begin{column}{0.5\textwidth}
|
|
\only<2>{
|
|
\begin{alertblock}{Key Challenge}
|
|
\small
|
|
Verify continuous control behavior across discrete mode transitions
|
|
\end{alertblock}
|
|
}
|
|
\only<3->{
|
|
\begin{block}{Key Challenge}
|
|
\small
|
|
Verify continuous control behavior across discrete mode transitions
|
|
\end{block}
|
|
}
|
|
|
|
\vspace{0.3cm}
|
|
|
|
\only<3>{
|
|
\begin{alertblock}{Reachable Set}
|
|
\small
|
|
$\mathcal{R}(t) = \{x(t) \mid x(0) \in X_0, \dot{x} = f(x)\}$
|
|
\end{alertblock}
|
|
}
|
|
\only<4->{
|
|
\begin{block}{Reachable Set}
|
|
\small
|
|
$\mathcal{R}(t) = \{x(t) \mid x(0) \in X_0, \dot{x} = f(x)\}$
|
|
\end{block}
|
|
}
|
|
|
|
\vspace{0.3cm}
|
|
|
|
\only<4>{
|
|
\begin{alertblock}{Barrier Certificate}
|
|
\small
|
|
$B(x) > 0 \land \nabla B \cdot f(x) \leq 0 \implies x \in \text{Safe}$
|
|
\end{alertblock}
|
|
}
|
|
\end{column}
|
|
\end{columns}
|
|
|
|
\end{frame}
|
|
|
|
|