©2000 - 2002 Dipl.-Ing. Tobias Jockenhövel. All rights reserved.
Introduction

[OptControlCentre] [Offline Menu] [Edit Menu] [Online Menu] [Sim. Annealing] [System Requirements]

In the offline menu, the user has a convenient interface to OCOMA to generate the FORTRAN code of the model, link it against the available solvers and execute the optimization problem. The user has also access to a couple of very useful features such as integration error calculation, auto increase time steps, make starting values (from previous solution or from dynamic simulation).

The following commands can be executed within the offline menu:

Generating and running the optimization problem:

  • “Generate Fortran Code” invokes the code generator OCOMA
  • “Make Executable” compiles and links the generated FORTRAN code against the NLP solvers
  • “Run Executable” runs the optimization problem
  • “Load Data” reads the optimization results, they can be viewed in the plot section.
  • “Auto Code Generation and Run” does automatically steps 1 to 3.
  • “Auto Increase Timesteps” executes a user given series of increasing number of time elements for the same model. Useful to get very fine solutions automatically from a coarse grid. If a dynamic models exists, the (decreasing) integration error over the number of collocation points is plotted. The latter is very useful to decide how many collocation points should be chosen.

Start guesses:

  • “Make Start Values” produces good start values from a previous solution.

Interaction with dynamic model:

  • “Make Start Values with Simulation” asks for a guess of the control profile and invokes a Simulink model of the process to generate a good starting point. This feature has been proven to be very effective, because the solver starts from a feasible point.
  • “Calculate Integration Error” takes the optimization results and invokes a Simulink simulation of the process. The overall integration error is calculated and a plot for every system variable at every collocation points is given. This feature has been proven to find even the slightest differences in the Simulink model and the optimization model. It can be used to verify results and models.

Viewing the results:

  • “Show Optimization Results (Plot)” plots the results in external windows.
  • “Show Optimization Results (in Editor)” opens the result file in the MATLAB Editor.
  • “Show Lagrangian Multipliers” plots in external windows the multiplier convergence for each system equation.
  • “Show Inzidenz Matrix” plots the occurrence Matrix of the optimization problem.

Other:

  • “Save result file ...” renames the result file for later use.
  • “Save executable file ...” renames the executable of the optimization problem for later use.
  • “Edit Model in Editor ...” opens the MATLAB Editor with the OCOMA optimization file.
  • “Edit Start Values in Editor ...” opens the file with the start guesses for the NLP solver. Starting values can be manually edited.

The Offline Menu was intended to initially just control the code generator OCOMA together with the MATLAB help files “makestart.m” (make start values from previous solution), “showme.m” (show all variables in a plot) written by Jens Bausa. The offline menu developed since in a ‘complete’ solution for offline optimization.