def run_world(num_steps=number_of_steps):
world = World(initialization_year=2018,
market_time_splices=MARKET_TIME_SPLICES,
data_folder="test_new",
number_of_steps=number_of_steps,
scenario_file=scenario_RL_few_agents,
total_demand=5000,
number_of_agents=3,
client_rl=client)
for i in range(num_steps):
world.step()
def eval_world(individual):
# time_start = time.time()
# for i in range(1000):
# pass
# t1 = time.time()
#
# time_taken = t1-time_start
# return [1], time_taken
# return ([1]),
# individual = np.array([0,0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0])
prices_individual = np.array(individual[:-3]).reshape(-1, 2).tolist()
MARKET_TIME_SPLICES = 8
YEARS_TO_RUN = 18
number_of_steps = YEARS_TO_RUN * MARKET_TIME_SPLICES
scenario_2018 = "{}/../run/beis_case_study/scenario/reference_scenario_2018.py".format(ROOT_DIR)
world = World(initialization_year=2018, scenario_file=scenario_2018, market_time_splices=MARKET_TIME_SPLICES, data_folder="best_run_beis_comparison", number_of_steps=number_of_steps, long_term_fitting_params=prices_individual, highest_demand=63910, nuclear_subsidy=individual[-3], future_price_uncertainty_m=individual[-2], future_price_uncertainty_c=individual[-1])
time_start = time.perf_counter()
timestamp_start = time.time()
for _ in range(YEARS_TO_RUN):
for i in range(MARKET_TIME_SPLICES):
try:
results_df, over_invested = world.step()
except:
return [[99999999], 0, 0, 0, 0]
if over_invested:
return [[99999999], 0, 0, 0, 0]
_, cumulative_diff = get_projection_difference_sum(results_df, world.year_number)
# print("cumulative diff: {}".format(cumulative_diff))
if cumulative_diff > 1.5:
joined, total_difference = get_projection_difference_sum(results_df)
else:
pass
time_end = time.perf_counter()
timestamp_end = time.time()
time_taken = time_end-time_start
joined, total_difference = get_projection_difference_sum(results_df)
print("max_demand : dif: {} :x {}".format(individual, total_difference))
# print(joined.simulated)
# print("input: {} {}, returns: {}, {}, {}".format(individual[0], individual[1], [total_difference], time_taken, joined.simulated))
# print("input: {} {}, returns: {}, {}, {}".format(individual[0], individual[1], [total_difference], time_taken, timestamp_start, timestamp_end, joined.simulated))
try:
return [total_difference], time_taken, timestamp_start, timestamp_end, joined.simulated
except AttributeError:
return [total_difference], time_taken, timestamp_start, timestamp_end, 0
def __init__(self, config):
print("trying to init")
self.step_number_env = 0
self.action = None
self.config = config
self.max_number_of_steps = config['max_number_of_steps']
self.world = World(initialization_year=2018, scenario_file=config['scenario_file'], data_folder=config['data_folder'], number_of_steps=self.max_number_of_steps)
self.action_space = Box(
0.0, 250.0, shape=(1, ), dtype=np.float32)
self.observation_space = Box(
-10000.0, 0.0, shape=(2, ), dtype=np.float32)
def reset(self):
self.step_number_env=0
self.world = World(initialization_year=2018, scenario_file=self.config['scenario_file'], data_folder=self.config['data_folder'], number_of_steps=self.max_number_of_steps)
self.world.step_number -= 1
resultant_observation = self.world.step(17)
mean_electricity_price = resultant_observation[0]
carbon_emitted = resultant_observation[1]
obs = np.array([mean_electricity_price, carbon_emitted])
# obs = -abs(mean_electricity_price)-abs(carbon_emitted)
# obs = np.random.uniform(low=-100, high=1, size=(2,))
obs = np.array(obs)
# return np.expand_dims(obs, axis=0)
return obs
class WorldEnvironment(gym.Env):
# def __init__(self, scenario_file=None, max_number_of_steps=32, data_folder="reinforcement_learning"):
def __init__(self, config):
print("trying to init")
self.step_number_env = 0
self.action = None
self.config = config
self.max_number_of_steps = config['max_number_of_steps']
self.world = World(initialization_year=2018, scenario_file=config['scenario_file'], data_folder=config['data_folder'], number_of_steps=self.max_number_of_steps)
self.action_space = Box(
0.0, 250.0, shape=(1, ), dtype=np.float32)
self.observation_space = Box(
-10000.0, 0.0, shape=(2, ), dtype=np.float32)
def reset(self):
self.step_number_env=0
self.world = World(initialization_year=2018, scenario_file=self.config['scenario_file'], data_folder=self.config['data_folder'], number_of_steps=self.max_number_of_steps)
self.world.step_number -= 1
resultant_observation = self.world.step(17)
mean_electricity_price = resultant_observation[0]
carbon_emitted = resultant_observation[1]
obs = np.array([mean_electricity_price, carbon_emitted])
# obs = -abs(mean_electricity_price)-abs(carbon_emitted)
# obs = np.random.uniform(low=-100, high=1, size=(2,))
obs = np.array(obs)
# return np.expand_dims(obs, axis=0)
return obs
def step(self, action):
self.action = action
self.step_number_env += 1
resultant_observation = self.world.step(action)
mean_electricity_price = resultant_observation[0]
carbon_emitted = resultant_observation[1]
ob = -abs(mean_electricity_price)-abs(carbon_emitted)
reward = ob
done = self.world.step_number > self.max_number_of_steps
print("step number: {}, action: {}, reward: {}, mean_electricity_price: {}, carbon_emitted: {}".format(self.world.step_number,action, reward, mean_electricity_price, carbon_emitted))
ob = np.array([mean_electricity_price, carbon_emitted])
return ob, reward, done, {}
def render(self):
return False
def run_scenario(gencos_rl_bidding, port_number):
print("Running scenario with: {}".format(gencos_rl_bidding))
time.sleep(60)
beis_params = [0.00121256259168, 46.850377392563864, 0.0029982421515, 28.9229765616468, 0.00106156336814,
18.370337670063762, 0.00228312539654, 0.0, 0.0024046471141100003, 34.43480109190594, 0.0,
-20.88014916953091, 0.0, 8.15032953348701, 0.00200271495761, -12.546185375581802, 0.00155518243668,
39.791132970522796, 0.00027449937576, 8.42878689508516, 0.00111989525697, 19.81640207212787,
0.00224091998324, 5.26288570922149, 0.00209189353332, -5.9117317131295195, 0.00240696026847,
-5.0144941135222, 0.00021183142492999999, -1.29658413335784, 0.00039441444392000004,
-11.41659250225168, 0.00039441444392000004, -11.41659250225168, 120.21276910611674, 0.0,
0.00059945111227]
prices_individual = np.array(beis_params[:-3]).reshape(-1, 2).tolist()
MARKET_TIME_SPLICES = 8
YEARS_TO_RUN = 1
number_of_steps = YEARS_TO_RUN * MARKET_TIME_SPLICES
scenario_2018 = "../scenario/reference_scenario_2018.py".format(ROOT_DIR)
carbon_df = pd.read_csv('linear_data_exploded.csv'.format(ROOT_DIR))
carbon_list = carbon_df.x.tolist()
# result_distributions_object = pickle.load(open(
# "{}/run/market_forecasting_comparison/run/Compare_worlds/result_distributions_object.p".format(ROOT_DIR),
# "rb"))
#
# resultant_dist = '{}'
#
# dist_class = eval(list(result_distributions_object[resultant_dist].fitted_param.keys())[0] + ".rvs")
# dist_object = dist_class(*list(result_distributions_object[resultant_dist].fitted_param.values())[0],
# size=50000).tolist()
while True:
world = World(carbon_price_scenario=carbon_list, initialization_year=2018, scenario_file=scenario_2018,
market_time_splices=MARKET_TIME_SPLICES, data_folder="compare_ml_accuracy",
number_of_steps=number_of_steps, long_term_fitting_params=prices_individual, highest_demand=63910,
nuclear_subsidy=beis_params[-3], future_price_uncertainty_m=beis_params[-2],
future_price_uncertainty_c=beis_params[-1], dropbox=False, gencos_rl=gencos_rl_bidding,
write_data_to_file=True, rl_port_number=port_number)
for _ in range(YEARS_TO_RUN):
for i in range(MARKET_TIME_SPLICES):
# try:
if i / 8 == 0:
print('end of year')
world.step()
def world_eval(individual, carbon_policy):
beis_params = [0.00121256259168, 46.850377392563864, 0.0029982421515, 28.9229765616468, 0.00106156336814, 18.370337670063762, 0.00228312539654, 0.0, 0.0024046471141100003, 34.43480109190594, 0.0, -20.88014916953091, 0.0, 8.15032953348701, 0.00200271495761, -12.546185375581802, 0.00155518243668, 39.791132970522796, 0.00027449937576, 8.42878689508516, 0.00111989525697, 19.81640207212787, 0.00224091998324, 5.26288570922149, 0.00209189353332, -5.9117317131295195, 0.00240696026847, -5.0144941135222, 0.00021183142492999999, -1.29658413335784, 0.00039441444392000004, -11.41659250225168, 0.00039441444392000004, -11.41659250225168, 120.21276910611674, 0.0, 0.00059945111227]
# beis_params = [0.00121256259168, 46.850377392563864, 0.0029982421515, 28.9229765616468, 0.00106156336814, 18.370337670063762, 0.00228312539654, 0.0, 0.0024046471141100003, 34.43480109190594, 0.0, -20.88014916953091, 0.0, 8.15032953348701, 0.00200271495761, -12.546185375581802, 0.00155518243668, 39.791132970522796, 0.00027449937576, 8.42878689508516, 0.00111989525697, 19.81640207212787, 0.00224091998324, 5.26288570922149, 0.00209189353332, -5.9117317131295195, 0.00240696026847, -5.0144941135222, 0.00021183142492999999, -1.29658413335784, 0.00039441444392000004, -11.41659250225168, 0.0021988838824299997, 12.633572943294599, 120.21276910611674, 0.0, 0.00059945111227]
prices_individual = np.array(beis_params[:-3]).reshape(-1, 2).tolist()
MARKET_TIME_SPLICES = 8
YEARS_TO_RUN = 17
number_of_steps = YEARS_TO_RUN * MARKET_TIME_SPLICES
scenario_2018 = "{}/../run/beis_case_study/scenario/reference_scenario_2018.py".format(ROOT_DIR)
# if individual[0] == 1:
# individual = [individual[1]*i + individual[2] for i in range(1, 20)]
# else:
# individual = [individual[3]*i ** individual[4] + individual[2] for i in range(1, 20)]
individual = [individual[0]*i + individual[1] for i in range(1, 20)]
print(individual)
world = World(carbon_price_scenario=individual[:-1], initialization_year=2018, scenario_file=scenario_2018, market_time_splices=MARKET_TIME_SPLICES, data_folder="{}_data".format(carbon_policy), number_of_steps=number_of_steps, long_term_fitting_params=prices_individual, highest_demand=63910, nuclear_subsidy=beis_params[-3], future_price_uncertainty_m=beis_params[-2], future_price_uncertainty_c=beis_params[-1])
for _ in range(YEARS_TO_RUN):
for i in range(MARKET_TIME_SPLICES):
# try:
if i/8 == 0:
print('end of year')
average_electricity_price, carbon_emitted = world.step()
# except Exception as e:
# print("try catch error: \n{}".format(e))
# return 889999999999999999, 889999999999999999
# if carbon_emitted is bool:
if isinstance(carbon_emitted, bool):
print("carbon emitted error")
return 999999999999999999, 999999999999999999
if not isinstance(average_electricity_price, float):
print("average_electricity_price {}, type {}".format(average_electricity_price, type(average_electricity_price)))
average_electricity_price = 779999999999999999
if not isinstance(carbon_emitted, float):
print("carbon_emitted {}, type {}".format(carbon_emitted, type(carbon_emitted)))
carbon_emitted = 779999999999999999
print("average_electricity_price: {}, carbon_emitted: {}".format(average_electricity_price, carbon_emitted))
return average_electricity_price, carbon_emitted
def eval_world_parallel(individual):
prices_individual = np.array(individual[:-3]).reshape(-1, 2).tolist()
# print(prices_individual)
MARKET_TIME_SPLICES = 8
YEARS_TO_RUN = 18
number_of_steps = YEARS_TO_RUN * MARKET_TIME_SPLICES
scenario_2018 = "{}/../run/beis_case_study/scenario/reference_scenario_2018.py".format(ROOT_DIR)
world = World(initialization_year=2018, scenario_file=scenario_2018, market_time_splices=MARKET_TIME_SPLICES, data_folder="best_run_beis_comparison", number_of_steps=number_of_steps, long_term_fitting_params=prices_individual, highest_demand=63910, nuclear_subsidy=individual[-3], future_price_uncertainty_m=individual[-2], future_price_uncertainty_c=individual[-1])
for j in range(YEARS_TO_RUN):
for i in range(MARKET_TIME_SPLICES):
try:
# print("j:{}, i: {}".format(j, i))
results_df, over_invested = world.step()
except:
return 99999, 0
del world
return individual, results_df
def run_scenario(ignore):
print(ignore)
world = World(carbon_price_scenario=individual[:-1],
initialization_year=2018,
scenario_file=scenario_2018,
market_time_splices=MARKET_TIME_SPLICES,
data_folder="UK_carbon_scenario_data",
number_of_steps=number_of_steps,
long_term_fitting_params=prices_individual,
highest_demand=63910,
nuclear_subsidy=beis_params[-3],
future_price_uncertainty_m=beis_params[-2],
future_price_uncertainty_c=beis_params[-1])
for _ in range(YEARS_TO_RUN):
for i in range(MARKET_TIME_SPLICES):
# try:
if i / 8 == 0:
print('end of year')
average_electricity_price, carbon_emitted = world.step()
# except Exception as e:
# print("try catch error: \n{}".format(e))
# return 889999999999999999, 889999999999999999
# if carbon_emitted is bool:
if isinstance(carbon_emitted, bool):
print("carbon emitted error")
# return 999999999999999999, 999999999999999999
if not isinstance(average_electricity_price, float):
# print("average_electricity_price {}, type {}".format(average_electricity_price, type(average_electricity_price)))
print("average_electricity_price")
average_electricity_price = 779999999999999999
if not isinstance(carbon_emitted, float):
print("carbon_emitted {}, type {}".format(
carbon_emitted, type(carbon_emitted)))
carbon_emitted = 779999999999999999
print("average_electricity_price: {}, carbon_emitted: {}".format(
average_electricity_price, carbon_emitted))
def eval_world(individual):
# time_start = time.time()
# for i in range(1000):
# pass
# t1 = time.time()
#
# time_taken = t1-time_start
# return [1], time_taken
# return ([1]),
# MARKET_TIME_SPLICES = 8
MARKET_TIME_SPLICES = 8
YEARS_TO_RUN = 6
number_of_steps = YEARS_TO_RUN * MARKET_TIME_SPLICES
scenario_2013 = "{}/../run/validation-optimisation/scenario_file/scenario_2013.py".format(
ROOT_DIR)
world = World(
initialization_year=2013,
scenario_file=scenario_2013,
market_time_splices=MARKET_TIME_SPLICES,
data_folder="best_run_all_dat_yearly_time_step_retired_plants",
number_of_steps=number_of_steps,
fitting_params=[individual[0], individual[1]],
highest_demand=63910)
time_start = time.perf_counter()
timestamp_start = time.time()
for i in range(number_of_steps):
results_df = world.step()
time_end = time.perf_counter()
timestamp_end = time.time()
time_taken = time_end - time_start
#
# contributed_results = results_df.filter(regex='contributed_').tail(MARKET_TIME_SPLICES)
# contributed_results *= 1/24
#
# # print("contributed_results: {}".format(contributed_results))
# contributed_results = contributed_results.rename(columns={'contributed_PV': "contributed_solar"})
# cluster_size = pd.Series([22.0, 30.0, 32.0, 35.0, 43.0, 53.0, 68.0, 82.0])
#
# # contributed_results['cluster_size'] = [22.0, 30.0, 32.0, 35.0, 43.0, 53.0, 68.0, 82.0]
#
# results_wa = contributed_results.apply(lambda x: np.average(x, weights=cluster_size.values)).to_frame()
#
# results_wa.index = results_wa.index.str.split("_").str[1].str.lower()
# # print("results_wa: {}".format(results_wa))
# offshore = results_wa.loc["offshore"].iloc[0]
# onshore = results_wa.loc["onshore"].iloc[0]
# # print("offshotre: {}".format(offshore))
# # results_wa = results_wa.append(pd.DataFrame({"wind", offshore+onshore}))
# results_wa.loc['wind'] = [offshore+onshore]
# # print("results_wa: {}".format(results_wa))
#
# electricity_mix = pd.read_csv("{}/data/processed/electricity_mix/energy_mix_historical.csv".format(ROOT_DIR))
# actual_mix_2018 = electricity_mix[electricity_mix.year == 2018]
#
#
# actual_mix_2018 = actual_mix_2018.set_index("variable")
# # print(actual_mix_2018)
#
# joined = actual_mix_2018[['value']].join(results_wa, how='inner')
# # print("joined: \n{}".format(joined))
#
# joined = joined.rename(columns={'value':'actual', 0:'simulated'})
#
# joined = joined.loc[~joined.index.str.contains('biomass')]
#
# # print("joined: \n{}".format(joined))
#
# joined['actual_perc'] = joined['actual']/joined['actual'].sum()
# joined['simulated_perc'] = joined['simulated']/joined['simulated'].sum()
#
# # print("joined: \n{}".format(joined))
#
# total_difference_col = joined['actual_perc'] - joined['simulated_perc']
# print(total_difference_col)
# total_difference = total_difference_col.abs().sum()
# # print("max_demand : dif: {} :x {}".format(individual, total_difference))
# # print(joined.simulated)
# # print("input: {} {}, returns: {}, {}, {}".format(individual[0], individual[1], [total_difference], time_taken, joined.simulated))
# # print("input: {} {}, returns: {}, {}, {}".format(individual[0], individual[1], [total_difference], time_taken, timestamp_start, timestamp_end, joined.simulated))
# return [total_difference], time_taken, timestamp_start, timestamp_end, joined.simulated
return 0, 0, 0, 0, 0
# # replace "/path/to/module/file.py" with "module/file.py"
# filename = os.sep.join(frame.filename.split(os.sep)[-2:])
# print("#%s: %s:%s: %.1f KiB"
# % (index, filename, frame.lineno, stat.size / 1024))
# line = linecache.getline(frame.filename, frame.lineno).strip()
# if line:
# print(' %s' % line)
#
# other = top_stats[limit:]
# if other:
# size = sum(stat.size for stat in other)
# print("%s other: %.1f KiB" % (len(other), size / 1024))
# total = sum(stat.size for stat in top_stats)
# print("Total allocated size: %.1f KiB" % (total / 1024))
#
# class TestWorld:
# def test_world_initialization(self):
# with PyCallGraph(output=GraphvizOutput()):
MARKET_TIME_SPLICES = 8
YEARS_TO_RUN = 40
number_of_steps = YEARS_TO_RUN * MARKET_TIME_SPLICES
for _ in range(1000):
world = World(initialization_year=2018,
market_time_splices=MARKET_TIME_SPLICES,
data_folder="test_new",
number_of_steps=number_of_steps)
for i in range(number_of_steps):
world.step()
import sys
# import scenario_file
# sys.modules['elecsim'].scenario.scenario_data=scenario_file
from elecsim.model.world import World
import logging
# logging.basicConfig(level=logging.INFO)
if __name__ == "__main__":
world = World(2018,
log_level="info",
market_time_splices=8,
number_of_steps=8)
for day in range(8):
world.step()
"""
File name: run_reference_scenario.py
Date created: 22/08/2019
Feature: #Enter feature description here
"""
__author__ = "Alexander Kell"
__copyright__ = "Copyright 2018, Alexander Kell"
__license__ = "MIT"
__email__ = "*****@*****.**"
# MARKET_TIME_SPLICES = 8
MARKET_TIME_SPLICES = 8
YEARS_TO_RUN = 17
number_of_steps = YEARS_TO_RUN * MARKET_TIME_SPLICES
scenario_2013 = "{}/../run/beis_case_study/scenario/reference_scenario_2018.py".format(
ROOT_DIR)
for _ in range(100):
world = World(initialization_year=2018,
scenario_file=scenario_2013,
market_time_splices=MARKET_TIME_SPLICES,
data_folder="best_run_beis_comparison",
number_of_steps=number_of_steps,
fitting_params=[0.001644, 11.04157],
highest_demand=63910)
for _ in range(number_of_steps):
world.step()
from elecsim.constants import ROOT_DIR
import logging
logger = logging.getLogger(__name__)
"""
File name: run_2013
Date created: 09/07/2019
Feature: #Enter feature description here
"""
__author__ = "Alexander Kell"
__copyright__ = "Copyright 2018, Alexander Kell"
__license__ = "MIT"
__email__ = "*****@*****.**"
pd.set_option('display.max_rows', 4000)
logging.basicConfig(level=logging.INFO)
MARKET_TIME_SPLICES = 8
YEARS_TO_RUN = 8
number_of_steps = YEARS_TO_RUN * MARKET_TIME_SPLICES
scenario_2013 = "{}/../run/validation-optimisation/scenario_file/scenario_2013.py".format(ROOT_DIR)
world = World(initialization_year=2013, scenario_file=scenario_2013, market_time_splices=MARKET_TIME_SPLICES, data_folder="runs_2013", number_of_steps=number_of_steps, fitting_params=[0.002, 0])
for i in range(number_of_steps):
world.step()