The Train Delay Model Developed by the Genetic Programming Algorithm

Algorithm 1

Used GPA-ES algorithm structure.

(1)	FOR ALL individuals DO Initialize() END FOR;
(2)	FOR ALL individuals DO Evaluate()=>fitness END FOR;
(3)	Sort(individuals according to fitness);
(4)	IF terminal condition is met THEN STOP END IF;
(5)	FOR ALL individuals DO
	SELECT Rand() OF
CASE a DO Mutate()=> new_individuals;
CASE b DO Symmetric_crossover(i-th, (i+1)th individuals) => new_individuals;
CASE c DO One_point_crossover(i-th, (i+1)th individuals) => new_individuals;
CASE d DO Re-generating() => new_individuals;
END SELECT;
END FOR;
(6)	FOR ALL new_individuals DO
New ES_algorithm_object with related ES_individuals and ES_fitness arrays
//for each GPA individual new independent parameter optimizer is created
FOR ALL ES_individuals DO Initialize() END FOR;
	Evaluate() => ES_fitness;
FOR ALL ES_cycles DO
FOR ALL ES_individuals DO
	Evaluate() => ES_fitness;
END FOR;
FOR ALL ES_individuals DO
	Intelligent_crossover() => new_ES_individuals
	Evaluate() => new_ES_fitness;
END FOR;
FOR ALL ES_individuals DO
	IF new_ES_fitness < ES_fitness THEN
	ES_individual = new_ES_individual;
	ES_fitness = new_ES_fitness;
END IF;
END FOR;
Sort(ES_individuals, ES_fitness);
END FOR;
new_individual = ES_individual[0]; new_fitness = ES_fitness[0];
END FOR;
(7)	FOR ALL individuals DO IF new_fitness < fitness THEN
	individual = new_individual;
	fitness = new_fitness;
	END IF;
	END FOR;
(8)	GOTO 3);