Prescan Runner

Prescan Runner is a plugin for OpenSBT to apply search-based testing to a Simulink-based system simulated in Prescan.

Prerequisites

This plugin is compatible with Python 3.8/3.7. Installation has been tested with Windows 10/11 and Prescan 2021.3.0.

Getting Started

Python Wheel

The easiest way to get the Prescan Runner up and running is to build it as a Python package and install it. To build the package, run python -m build in the repository's root directory (npms build package has to be installed before). Once completed, install the *.whl package found in the newly created dist folder via python -m pip install /path/to/the/package.whl

Assumptions

The current implementation of Prescan Runner requires that following is assured:

• The scenario is parametrized using an input.json file.
• A ChangeModel.m script is provided to update the .pb experiment file using a file called input.json file.
• The simulink model contains blocks to write traces of actors of interest into trace_online.csv. The format of the output can be adapted here.
• The prescan experiment has been compiled, i.e. simulation is triggered via running an executable. The name of the executable can be set here.

Before Running Experiment

Execute the following steps before running experiments with OpenSBT:

(Note, that step 1 is only required once.)

1. Import the matlab engine into OpenSBT:

   cd MATLAB_DIR/extern/engines/python
   py -3.8 setup.py install --prefix="C:\Path\To\Project\venv"
   

   Further description/options are available here.

2. Start MATLAB via the Prescan Process Manager.

3. Share the MATLAB engine to be accessable by OpenSBT by executing from the MATLAB terminal:

   matlab.engine.shareEngine

4. Make sure all files have been added to the MATLAB path which a required for executing the simulink SUT, as done in this example startup script.