""" State consists of (in order): machine params: pending jobs in last epoch machine age last epoch params: free labour next epoch demand: avg due after delay penalty reward = objfun, cost """ env = IndustrySim(machines=[machine1,machine2, machine3, machine4, machine5, machine6], epoch_length=epoch_length, max_labor=max_labor, wages=wages, job_demand=job_demand, delay_penalty=delay_penalty, state_size=state_size) start = time.time() env.reset() res = EpochResult(None, None, None) nn = NN(dim_input=state_size, dim_hidden_layers=nn_arch, dim_output=action_size, do_dropout=True) states = np.zeros((max_epochs, state_size)) actions = np.zeros((max_epochs, action_size)) rewards = np.zeros(max_epochs) state = np.zeros(state_size) # hyper params e = 0.2 training_passes = 20000 start = time.time()
from entities import Policy, Machine, MaintenanceTask, EpochResult, PeriodicPolicy
from engine import IndustrySim
from tools import e_greedy, NN
import numpy as np
import time
from industry import epoch_length, max_epochs, max_labor, wages, num_machines, job_demand
from industry import delay_penalty, mt_fixed_cost, mt_RF, mt_ttr, mt_labor, beta, age
from industry import compatible_jobs, machine_names, mt_task1, mt_task2, mt_task3
from industry import machine1, machine2, machine3
env = IndustrySim(machines=[machine1, machine2, machine3],
epoch_length=epoch_length,
max_labor=max_labor,
wages=wages,
job_demand=job_demand,
delay_penalty=delay_penalty)
start = time.time()
env.reset()
res = EpochResult(None, None, None)
experiments = 300
start = time.time()
avg_obj = 0
high_jobs = {'FnC1': 0, 'Lathe': 0, 'Milling': 0}
low_jobs = {'FnC1': 0, 'Lathe': 0, 'Milling': 0}
for exp in range(experiments):
pp = PeriodicPolicy(machine_names=machine_names, epoch_interval=5)
for i in range(max_epochs):
epoch_result = env.run_epoch(pp.get_policy())
res = env.get_result()
print(res)