It is a library and a GUI application for Classic Genetic Algorithm on the topic PID tuning. It includes functions for analyzing systems, simulating a system consists of PID controller and a plant, and of course Classic Genetic Algorithm.
Classic Genetic Algorithm is a evolutionary programming methodology that does cross-over, mutation and inversion into the each generation to find a solution for particular problems. In our program, problem is finding a controller for given plant. The program firstly creates a initial population from random PID triples. After selection of parents, it applies crossover, mutation and inversion.
Selection of parent chromosomes is done by Elite Selection Method. It is a way to select using a wheel of luck, and put the best ones into the next generation. This method is useful for not losing the best solution that we have.
It is using the functionality of application interface, with embedded functions inside the GUI code. It is possible to watch how GUI looks in YouTube. Here you can see the features:
- First-order inertia plant creation,
- Inputs for population size, reference value, ending conditions, and chances of biological functions,
- An option to stop simulation,
- Create a text file as a report,
- Save the simulation, and import it to continue later.
Please note, "Max Same Generation" text-box is an input for the maximum generation of being same "best" individual. So, if your simulation won't stop after X amount of generation, don't surprise. It is going to be finished after the same individual became best for X generations.
You can find the API reference in the doc directory. To start using it, check out a fast-start example in below:
% Create the Plant Model plant_model = CreatePlantObject(1.5, 12, 4); % Find the PID Tuning [PIDCoeff, FitnessVal, GenCount] = ClassicGeneticAlgorithm(plant_model, 20, 5, -0.4, 25, 0.5, 0.5, 0.5); % Simulate the Controller [y_data, x_data] = SimulatePIDSystem(5, PIDCoeff, plant_model); plot(x_data, y_data);
Application and its functions have licensed under MIT. You can directly change the code, update it, or fix something, and publish it. Please contribute to this project to make it better, and free for everyone.