def Genetic_al(model, data, i, stud1):
ga = GeneticAlgorithm(data,
population_size=100,
generations=100,
crossover_probability=0.01,
mutation_probability=0.01,
elitism=False,
maximise_fitness=False)
def create_individual(data):
return [
float(1.0 + 5.0 * random.uniform(0, 1)), # Drop time actual
float(1.0 + 15.0 * random.uniform(0, 1)), # Drop time actual_EWMA
float(1.0 + 15.0 * random.uniform(0, 1)), # Drop time difference
float(1.0 +
15.0 * random.uniform(0, 1)), # Drop time difference_EWMA
# float((random.randint(300, 4000) + random.uniform(0, 1))),# Energy
# float((random.randint(300, 4000) + random.uniform(0, 1))), # Energy_EWMA
float(1.0 + 2.0 * random.uniform(0, 1)), # Lift Height actual
float(1.0 + 2.0 * random.uniform(0, 1)), # Lift Height actual_EWMA
# float((random.randint(0, 2) + random.uniform(0, 1))), # lift height ref
# float((random.randint(0, 2) + random.uniform(0, 1))), # lift height ref_EWMA
float(15.0 + 40.0 *
random.uniform(0, 1)), # Main Weldcurrent voltage actual
float(15.0 + 4.0 * random.uniform(0, 1)
), # Main Weldcurrent voltage actual_EWMA
# ((random.randint(25, 50) + random.uniform(0, 1))),# Main Weldcurrent voltage Max
# ((random.randint(25, 50) + random.uniform(0, 1))), # Main Weldcurrent voltage Max_EWMA
# ((random.randint(8, 28) + random.uniform(0, 1))), # Main Weldcurrent voltage Min
# ((random.randint(8, 28) + random.uniform(0, 1))), # Main Weldcurrent voltage Min_EWMA
# (-(random.randint(1, 3) + random.uniform(0, 1))), # Penetration Max
# (-(random.randint(1, 3) + random.uniform(0, 1))), # Penetration Max_EWMA
# (-(random.randint(0, 3) + random.uniform(0, 1))), # Penetration Min
# (-(random.randint(0, 3) + random.uniform(0, 1))), # Penetration Min_EWMA
# (-(random.randint(0, 3) + random.uniform(0, 1))),# Penetratiom Ref
# (-(random.randint(0, 3) + random.uniform(0, 1))), # Penetratiom Ref_EWMA
float(8.0 + 35.0 *
random.uniform(0, 1)), # Pilot Weldcurrent Arc Voltage Act
float(8.0 + 35.0 * random.uniform(0, 1)
), # Pilot Weldcurrent Arc Voltage Act_EWMA
# ((random.randint(20, 50) + random.uniform(0, 1))),# Pilot Weldcurrent Arc Voltage Max
# ((random.randint(20, 50) + random.uniform(0, 1))), # Pilot Weldcurrent Arc Voltage Max_EWMA
# ((random.randint(5, 20) + random.uniform(0, 1))),# Pilot Weldcurrent Arc Voltage Min
# ((random.randint(5, 20) + random.uniform(0, 1))), # Pilot Weldcurrent Arc Voltage Min_EWMA
float(1.5 + 8.0 * random.uniform(0, 1)), # Stickout
float(1.5 + 8.0 * random.uniform(0, 1)), # Stickout_EWMA
float(500.0 + 1000.0 *
random.uniform(0, 1)), # Weldcurrent actual Positive
float(500.0 + 1000.0 *
random.uniform(0, 1)), # Weldcurrent actual Positive_EWMA
float(-1500.0 + 1500.0 *
random.uniform(0, 1)), # Weldcurrent actual Negative
float(-1500.0 + 1500.0 *
random.uniform(0, 1)), # Weldcurrent actual Negative_EWMA
float(10.0 + 100.0 * random.uniform(0, 1)), # Weld time actual
float(10.0 + 100.0 * random.uniform(0, 1))
] # Weld time actual_EWMA
# ((random.randint(10, 100) + random.uniform(0, 1))), # Weldtime ref
# ((random.randint(10, 100) + random.uniform(0, 1)))] # Weldtime ref_EWMA
ga.create_individual = create_individual
def eval_fitness(individual, data):
array = np.array(individual)[np.newaxis]
error_array = []
error = (model.predict(array, batch_size=1) + 2)**2
error_array.append(individual)
print('Evaluating... error: ' + str(error))
return error
ga.fitness_function = eval_fitness
ga.run()
Gen1 = pd.DataFrame(ga.last_generation())
filepath = "C:\\Users\PHLEGRA\Desktop\MASTER\Data_intersection\Prescribed_parameters\\new"
filename = str(stud1) + "_" + str(i) + '_predictions.csv'
Gen1.to_csv(filepath + '\\' + filename, index=False)
print('Please see file. Process Complete')
parser.add_argument('--turbines', default=2, type=int, help='enter an integer number of turbines to start with')
parser.add_argument('--grid_size', default=36, type=int, help='enter an perfect square integer representing grid size (e.g. 6x6 is 36)')
args = parser.parse_args()
given_crossover_probability = args.crossover_probability
given_mutation_probability = args.starting_mutation_probability
given_adaptive_mutation_factor = args.adaptive_mutation_factor
given_clipping_threshold = args.clipping_threshold
given_turbines = args.turbines
grid_size= args.grid_size
input_data = [0]*grid_size # length of this list is the # of grid cells in layout
easyga = GeneticAlgorithm(input_data, crossover_probability =given_crossover_probability , mutation_probability=given_mutation_probability, adaptive_mutation_factor = given_adaptive_mutation_factor, clipping_threshold = given_clipping_threshold, num_turbines = given_turbines)
easyga.fitness_function = fitness
easyga.run()
print(easyga.best_individual())
def fitness (individual, data):
freestream_velocity = 15
individual_2D_list = [[individual[0], individual[1]],[individual[2], individual[3]]]
individual_np = np.asarray(individual_2D_list)
velocities = np.zeros(individual_np.shape)
alpha = .9 # entrainment constant
x = 150 # distance between perpendicular turbine centers
r = 40 # radius of turbine
for rowIndex in range(np.shape(individual_np)[0]):
mutate_index2], individual[mutate_index1]
ga.mutate_function = mutate
#Рулеточная селекция
def selection(population):
return random.choice(population)
ga.selection_function = selection
#Реализация функции приспособленности
#Ищет такой х, при котором значение функции максимально
def fitness(individual, data):
fitness = -float("inf")
x = -4
for item in individual:
if item != 1:
x += 0.01
else:
fitness = 1 / x
break
return fitness
ga.fitness_function = fitness
ga.run()
print(ga.best_individual()) #Выводит самую приспособленную особь
rnd = np.random.normal(0, 0.5)
if rnd < -0.5:
rnd = -0.5
elif rnd > 0.5:
rnd = 0.5
individual[0] = individual[0] + rnd
ga.mutate_function = mutate
def selection(population): #Функция селекции (турнирный отбор)
ind1 = random.choice(population)
ind2 = random.choice(population)
if ind1.fitness > ind2.fitness:
return ind1
else:
return ind2
ga.selection_function = selection
def fitness (individual, data): #Целевая функция
if individual[0] < -4.0: #Ограничения переменных ЦФ
individual[0] = -4.0
if individual[0] >= 0:
individual[0] = 1e-10
return 1 / individual[0] #Значение целевой функции
ga.fitness_function = fitness
ga.run() #Запуск ГА
print("\nBest individual: ", ga.best_individual(), "\n") #Вывод лучшей особи популяции (Решение)
for individual in ga.last_generation(): #Вывод всех особей популяции в последнем поколении
print(individual)