This code is accompanying my Bachelor Thesis at the Ludwig-Maximilians-University of Munich. It simulates a basic Smart Factory and tries to optimize the positions of workstations in its layout. This optimization is accomplished by using methods following a basic Genetic Algorithm.
The Thesis compares the use of several diversity metrics induced in the fitness formula of the parent selection phase of the Genetic Algorithm. The use of diversity measurements itself stems from intentional unexpected changes in the environment, namely the breakdown (deactivation) of workstations in the middle of the simulation process. The hypothesis is that high diversity within the population reduces the negative effect of environmental changes to the generated solutions. Population in this setting is defined as a specified number of factory layouts.
This project uses Carthage as a dependency manager. To run the code you have to install Carthage by following their guide. Afterwards the following commands have to be executed in the Terminal, where PATH/TO/PROJECT/ is your actual path to the project:
cd PATH/TO/PROJECT/
carthage bootstrap --platform macOS
This project comes equipped with a few simulation modes to choose from:
- Development
- Test Phase 1-4
- Production
Those modes where the ones used in the process of development and as the basis of my research. Adding customized simulation modes is one of the benefits of the clear code structure in this project. To add your own mode, simply go to SimulationMode.swift and add a seventh case to the Enumeration. Building the project then will fail and Xcode will point out all the lines where the simulation parameters of your newly created mode have to be added. Make sure to fill them all according to the documentation comments above each parameter. As a last step the new simulation mode has to be selected inside SimulationSetting.swift along with the path to a folder where you want the generated statistics after each simulation.
This project has a modularized structure of phases for the Genetic Algorithm. This structure is enforced by a protocol called GAPhase. Every phase has to conform to GAPhase, which requires a method called execute(on generation: inout Generation). This method will be called each time the phase is executed and should modify the current generation (current solutions of the Genetic Algorithm). Adding a custom phase therefore only requires two steps:
- Write a struct that implements
GAPhase - Add your phase to the list of phases for your desired simulation mode. You can find this as
var phasesin SimulationMode.swift