The Heat Treatment Optimisation Project contains a series of datasets which are currently being examined by Dr Stathis Tingas at Edinburgh Napier University which explores what happens when certain processes are applied to materials. The project itself consists of five datasets (Plastic Anisotropy, Tensile Strength, Total Elongation, Vickers Hardness and Yield Strength) where the data contained within them is represented in three dimensions. The first is time, the second is temperature, and the third is a value that has been normalised between 1 and 0, detailed within the dataset.
The objective function is the euclidean distance of the time (t) and temperature (T), which is given as below.
The evolutionary algorithm was developed in Java 19 and the project can be compiled by using Maven to have the code deployed on to a computer of higher specification. For ease, it is recommended that the project is opened in an Integrated Development Environment (IDE) such as IntelliJ Idea provided by JetBrains.
Most notably, the project contains a rather handy Parameters file, which controls how the code will execute upon compilation. There is an abridged version of this file below in the following code block.
/* Control how the evolutionary algorithm initialises the population.
* Possible values:
* - AUGMENTED
* - RANDOM
*/
public static Initialisation INITIALISATION = Initialisation.AUGMENTED;
/* Control how the evolutionary algorithm controls the selection process.
* Possible values:
* - RANDOM
* - ROULETTE
* - ROUTE_RANK
* - TOURNAMENT
*/
public static Selection SELECTION = Selection.ROULETTE;
/* Control how the evolutionary algorithm controls the crossover process.
* Possible values:
* - ARITHMETIC
* - ONE_POINT
* - TWO_POINT
* - UNIFORM
*/
public static Crossover CROSSOVER = Crossover.UNIFORM;
/* Control how the evolutionary algorithm controls the mutation process.
* Possible values:
* - CONSTRAINED
* - STANDARD
*/
public static Mutation MUTATION = Mutation.CONSTRAINED;
/* Control how the evolutionary algorithm controls the replacement process.
* Possible values:
* - TOURNAMENT
* - WORST
*/
public static Replacement REPLACEMENT = Replacement.WORST;
/* Control how the evolutionary algorithm evaluates the chromosomes.
* Possible values:
* - MINIMISATION
* - MAXIMISATION
*/
public static ProblemType PROBLEM = ProblemType.MINIMISATION;
// The tournament size for the selection and replacement process.
public static int TOURNAMENT_SIZE = 10;
// The file that is to be read in by the evolutionary algorithm.
public static String FILE_NAME = "plastic-anisotropy.csv";
// The default population size.
public static int POPULATION_SIZE = 300;
// The augmented population size when using augmented initialisation.
public static int AUGMENTED_POPULATION_SIZE = 15750;
// The maximum number of evaluations/generations.
public static int MAXIMUM_EVALUATIONS = 20000;
// Parameters for mutation
// Rate = probability of changing a gene
// Change = the maximum +/- adjustment to the gene value
public static double MUTATION_RATE = 0.90; // mutation rate for mutation operator
public static double MUTATION_CHANGE = 1.50; // delta change for mutation operatorPull requests are welcome! All we ask is, credit goes to the authors and contributors of this project:
Dr Stathis Tingas: [email protected], Lecturer of Mathematics
Dr Neil Urquhart: [email protected], Lecturer of Computer Science
Stewart Anderson: [email protected], PhD Candidate
Robert Galloway: [email protected], Research Assistant
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent