Getting started, How optquest works, The optquest user interface – Rockwell Automation Arena OptQuest Users Guide User Manual

Page 11: How optquest works the optquest user interface

Advertising
background image

7

2

2 • Getti

ng Star

ted

Getting Started

How OptQuest works

Recent developments in the area of optimization have allowed for the creation of
intelligent search methods capable of finding optimal or near optimal solutions to
complex problems involving elements of uncertainty. Often, optimal solutions can be
found among large sets of possible solutions even when exploring only a small fraction of
them. OptQuest is the result of implementing these search technologies in combination
with simulation models built for Arena.

Once the optimization problem is described (by means of selecting controls, the objective,
and possibly imposing constraints), Arena is called every time a different set of control
values needs to be evaluated. The optimization method used by OptQuest evaluates the
responses from the current simulation run, analyzes and integrates these with responses
from previous simulation runs, and determines a new set of values for the controls, which
are then evaluated by running the Arena model. This is an iterative process that
successively generates new sets of values for the controls, not all of them improving, but
which, over time, provides a highly efficient trajectory to the best solutions. The process
continues until some termination criterion is satisfied—usually stopping after a number of
simulations or when the OptQuest determines the objective value has stopped improving.
Its ultimate goal is to find the solution that optimizes (maximizes or minimizes) the value
of the model’s objective.

Once OptQuest exits, the controls in the Arena model are returned to their original default
values. The Arena model is completely unaffected by OptQuest.

The OptQuest user interface

OptQuest for Arena now has a tree-structured user interface that displays the optimization
model components (controls, responses, constraints, objectives, suggested solutions, and
options) as nodes in the tree structure in the leftmost pane. When selected, each node
displays its summary grid in the righthand pane. Entries in the tree containing a
plus/minus (+/-) sign before the descriptor name may be expanded or collapsed to reveal
or hide the sub-categories. Selecting the main level displays the summary sheet, while
selecting the sub-category displays the editing window for the selected node.

Some individual nodes in the tree will display a right-click context menu option. Each
represents an action that is specific to the tree item (not all nodes have a context menu).
For example, a right-click on Controls displays either Expand or Collapse, depending on

Advertising