Genetic Algorithm for Job Scheduling

This repository contains MATLAB code for solving a job scheduling problem using a genetic algorithm. The script initializes the task parameters and runs a genetic algorithm to optimize the job scheduling with considerations for processing times, transportation times, and penalties for earliness and tardiness.

Result as Gantt Chart

Usage

1. Configure the settings in the main.m script:

Task Parameters

• processingTimes1: A list of processing times for the first stage of each job.
• transportationTimes: A list of transportation times between stages for each job.
• processingTimes2: A list of processing times for the second stage of each job.
• transCapacity: The transportation capacity, indicating how many jobs can be transported at once.
• returnTime: The time it takes for the transportation resources to return.
• dueDates: A list of due dates for each job to be completed.
• earlinessPenalty: A list of penalties for completing each job earlier than its due date.
• tardinessPenalty: A list of penalties for completing each job later than its due date.
• w1: Weight parameter for the fitness function, influencing the importance of penalties.
• w2: Weight parameter for the fitness function, influencing the importance of penalties.

Genetic Algorithm Parameters

• numGenerations: The number of generations for the genetic algorithm to run.
• popSize: The size of the population in the genetic algorithm.
• crossoverRate: The probability of crossover between pairs of individuals.
• mutationRate: The probability of mutation in an individual.
• use_NEH: Boolean flag to determine whether to use the NEH heuristic for initialization.
• use_local_search: Boolean flag to determine whether to apply local search for optimization.

2. Run the main.m script:

   main

Functions

Main Script

• main.m: Initializes parameters, runs the genetic algorithm, and prints results.

Helper Functions

• initializePopulationNEH: Initializes population using the NEH heuristic.
• initializePopulation: Initializes a random population.
• calculateFitness: Evaluates the fitness of a solution.
• rouletteWheelSelection: Selects parents for crossover based on fitness.
• pmxCrossover: Performs partially mapped crossover on two parents.
• insertionMutation: Mutates a solution by inserting an element at a new position.
• localSearch: Improves a solution using local search methods.
• reshape_2D: Reshapes a solution into a 2D matrix for easy interpretation.
• plotGanttChart: Plots a Gantt chart to visualize the scheduling.

License

This project is licensed under the MIT License - see the LICENSE file for details.