def get_data():
"""
Returns all possible simulations
"""
conf = parse_config()
datafile = conf["datafile"]
data = pd.read_csv(datafile)
if "P1" in data.columns:
PL = (data.P1 + data.P2).to_numpy()
else:
PL = data.PL.to_numpy()
PV = data.PV.to_numpy()
if "Spot_pris" in data.columns:
grid_buy = grid_sell = data.Spot_pris.to_numpy()
else:
grid_buy = grid_sell = 1.5
if conf["perfect_predictions"] or "PV_pred" not in data.columns:
PV_pred = PV.copy()
PL_pred = PL.copy()
else:
PV_pred = data.PV_pred.to_numpy()
PL_pred = data.PL_pred.to_numpy()
return PV, PV_pred, PL, PL_pred, grid_buy, grid_sell
def __init__(self, T, N):
self.T = T
self.N = N
conf = parse_config()
conf_system = conf["system"]
# Get system constants
self.C_MAX = conf_system["C_MAX"]
self.nb_c = conf_system["nb_c"]
self.nb_d = conf_system["nb_d"]
self.x_min = conf_system["x_min"]
self.x_max = conf_system["x_max"]
self.x_ref = conf_system["x_ref"]
self.Pb_max = conf_system["Pb_max"]
self.Pg_max = conf_system["Pg_max"]
self.battery_cost = conf_system["battery_cost"]
self.grid_cost = conf_system["grid_cost"]
self.ref_cost = conf_system["ref_cost"]
self.verbose = conf_system["verbose"]
self.states = struct_symSX([entry("SOC")])
self.inputs = struct_symSX(
[
entry("Pbc"),
entry("Pbd"),
entry("Pgb"),
entry("Pgs"),
]
)
self.w = struct_symSX(
[
entry("states", struct=self.states, repeat=self.N),
entry("inputs", struct=self.inputs, repeat=self.N - 1),
]
)
self.data = struct_symSX([entry("pv"), entry("l"), entry("E")])
self.all_data = struct_symSX([entry("data", struct=self.data, repeat=self.N)])
self.E = SX.sym("E", self.N)
# Initialize system properties
self.ode = self.build_ode()
self.L = None
self.F = None
self.solver = None
# Keep optimal solutions
self.SOC = np.asarray([])
self.Pbc = np.asarray([])
self.Pbd = np.asarray([])
self.Pgs = np.asarray([])
self.Pgb = np.asarray([])
def __init__(self, actions_per_hour=6):
self.logger = logging.getLogger("Metrics")
f_handler = logging.FileHandler("./logs/metrics.log")
s_handler = logging.StreamHandler()
f_handler.setLevel(logging.DEBUG)
s_handler.setLevel(logging.INFO)
s_format = logging.Formatter("%(message)s")
f_format = logging.Formatter("%(message)s")
s_handler.setFormatter(s_format)
f_handler.setFormatter(f_format)
self.logger.addHandler(s_handler)
self.logger.addHandler(f_handler)
self.conf = parse_config()
self.actions_per_hour = actions_per_hour
self.C_max = self.conf["battery"]["C_MAX"]
self.battery_deg = self.conf["system"]["battery_cost"]
self.peak_cost = self.conf["system"]["peak_cost"]
self.old_grid_fee = self.conf["system"]["old_grid_fee"]
assert (
self.peak_cost *
self.old_grid_fee == 0), "Either peak or old cost has to be zero"
self.logger.info("\n \n")
self.logger.info("Run started at {}".format(datetime.now()))
self.logger.info(
"Sim-time: {}, Pred-horizon: {}, N Scenarios: {}, Perfect Predictions: {}"
.format(
str(self.conf["simulation_horizon"]),
str(self.conf["prediction_horizon"]),
str(self.conf["N_scenarios"]),
str(self.conf["perfect_predictions"]),
))
self.logger.info("-" * 100)
self.grid_cost = 0
self.battery_cost = 0
self.grid_max = 0
self.old_system_cost = 0
self.primary_cost = 0
self.pv_rmse = 0
self.accumulated_error = 0
self.computational_time = 0
self.worst_case = 0
self.steps = 0
def scenario_mpc():
"""
Main function for mpc-scheme with receding horizion.
"""
np.random.seed(1)
conf = utils.parse_config()
testfile = conf["testfile"]
trainfile = conf["trainfile"]
logpath = None
loggerpath = "./logs/all_logs.log"
foldername = conf["foldername"]
if foldername:
logpath = utils.create_logs_folder(conf["logpath"], foldername)
loggerpath = logpath + "logs.log"
logging.basicConfig(
filename=loggerpath,
filemode="a",
format="%(name)s - %(levelname)s - %(message)s",
level=logging.INFO,
)
perfect_predictions = conf["perfect_predictions"]
actions_per_hour = conf["actions_per_hour"]
horizon = conf["simulation_horizon"]
T = conf["prediction_horizon"]
N = conf["prediction_horizon"] * actions_per_hour
N_scenarios = conf["N_scenarios"]
simulation_horizon = horizon * actions_per_hour
assert N_scenarios in [1, 3, 7, 9], "Only 1, 3, 7 or 9 branches allowed"
# Get data
observations = pd.read_csv(testfile, parse_dates=["date"]).fillna(0)
observations = observations[observations["date"] >= datetime(
conf["start_year"], conf["start_month"], conf["start_day"])]
assert (simulation_horizon + N <
observations.shape[0]), "Not enough data for simulation"
# Read forecast file
solcast_forecasts = pd.read_csv(conf["solcast_file"],
parse_dates=["time",
"collected"]).fillna(0)
# Read weighting files
load_weights = pd.read_csv("./data/load_weights.csv")
pv_weights = pd.read_csv("./data/pv_weights.csv")
current_time = observations.date.iloc[0]
forecast = solcast_forecasts[solcast_forecasts["collected"] ==
current_time - timedelta(minutes=60)]
obs = observations[
(observations["date"] >= current_time)
& (observations["date"] <= current_time + timedelta(minutes=10 * N))]
# Initialize components
E = pd.read_csv(conf["price_file"], parse_dates=["time"])
L = Load(N, testfile, "L", current_time)
B = Battery(T, N, **conf["battery"])
PV = LinearPhotovoltaic(trainfile)
sys_metrics = metrics.SystemMetrics()
# Define variables
Pbc = []
Pbd = []
Pgs = []
Pgb = []
Pgb_p = 0
Pgb_p_all = [Pgb_p]
primary_Pgb = 0
primary_Pgs = 0
prob = [1]
pv_measured = []
l_measured = []
errors = []
E_measured = []
time_stamps = [current_time]
prediction_time = 0
solver_time = 0
filter_ = [str(i) for i in range(N)]
# Initialize and build optimal control problem
ocp = ScenarioOCP(T, N, N_scenarios)
s_data = ocp.s_data(0)
s0, lbs, ubs, lbg, ubg = ocp.build_scenario_ocp()
# Initialize plots and progressbar
fig1, (ax1, ax2) = plt.subplots(2, 1, figsize=(10, 7))
bar = progressbar.ProgressBar(
maxval=simulation_horizon,
widgets=[
progressbar.Bar("=", "[", "]"),
" Finished with",
progressbar.Percentage(),
],
)
bar.start()
# MPC loop
for step in range(simulation_horizon):
step_start = time.time()
# Get measurements
pv_true = obs["PV"].values[0]
l_true = obs["L"].values[0]
# Get new forecasts every hour
if current_time.minute == 30:
new_forecast = solcast_forecasts[solcast_forecasts["collected"] ==
current_time -
timedelta(minutes=30)]
if new_forecast.empty:
print("Could not find forecast, using old forecast")
else:
forecast = new_forecast
ref = forecast[
(forecast["time"] > current_time)
& (forecast["time"] <= current_time + timedelta(minutes=10 * (N)))]
E_prediction = E[(E.time > current_time)
& (E.time <= current_time +
timedelta(minutes=10 * (N)))].price.values
# Get predictions
if perfect_predictions:
pv_prediction = obs["PV"].values[1:]
l_prediction = obs["L"].values[1:]
else:
pred_time = time.time()
pv_prediction = PV.predict(ref.temp.values, ref.GHI.values,
obs["PV"].iloc[0])
l_prediction, l_lower, l_upper = L.get_statistic_scenarios(
current_time, step)
l_prediction = L.linear_mixture(l_prediction, l_true)
prediction_time += time.time() - pred_time
if N_scenarios == 1:
pv_scenarios = [pv_prediction]
l_scenarios = [l_prediction]
else:
# Construct scenarios and weights
l_lower, l_upper = L.get_minmax_day(current_time, step)
l_probs = np.asarray([
utils.get_probability_from_file(load_weights, step, i, filter_)
for i in range(3)
])
pv_probs = np.asarray([
utils.get_probability_from_file(pv_weights, step, i, filter_)
for i in range(3)
])
pv_upper = PV.predict(ref.temp.values, ref.GHI90.values)
pv_lower = PV.predict(ref.temp.values, ref.GHI10.values)
pv_scenarios = np.asarray([pv_upper, pv_prediction, pv_lower])
l_scenarios = np.asarray([l_lower, l_prediction, l_upper])
prob = np.multiply(pv_probs, l_probs)
prob /= np.sum(prob) # Scale to one
if N_scenarios == 7:
pv_scenarios = [
pv_upper,
pv_upper,
pv_prediction,
pv_prediction,
pv_prediction,
pv_lower,
pv_lower,
]
l_scenarios = [
l_lower,
l_prediction,
l_lower,
l_prediction,
l_upper,
l_prediction,
l_upper,
]
prob = np.multiply(
np.asarray([
pv_probs[0],
pv_probs[0],
pv_probs[1],
pv_probs[1],
pv_probs[1],
pv_probs[2],
pv_probs[2],
]),
np.asarray([
l_probs[0],
l_probs[1],
l_probs[0],
l_probs[1],
l_probs[2],
l_probs[1],
l_probs[2],
]),
)
elif N_scenarios == 9:
pv_scenarios = np.repeat(pv_scenarios, 3, axis=0)
l_scenarios = np.tile(l_scenarios, (3, 1))
prob = np.multiply(np.repeat(pv_probs, 3), np.tile(l_probs, 3))
prob /= np.sum(prob)
# Plot scenarios and predictions
if N_scenarios <= 3 and step % 18 == 0:
colors = [i for i in get_cmap("tab10").colors]
t_1 = [current_time + timedelta(minutes=10 * i) for i in range(N)]
if step == 0:
label_s = ["Upper", "Prediction", "Lower"]
label_obs = "Observation"
else:
label_s = [None] * 3
label_obs = None
for i in range(len(pv_scenarios)):
ax1.plot(
t_1,
pv_scenarios[i],
label=label_s[1],
color=colors[1],
)
ax2.plot(
t_1,
l_scenarios[i],
label=label_s[1],
color=colors[1],
)
ax1.plot(t_1, obs.PV[1:], label=label_obs, color=colors[0])
ax2.plot(t_1, obs.L[1:], label=label_obs, color=colors[0])
# Update parameters
for i in range(N_scenarios):
s0["scenario" + str(i), "states", 0, "SOC"] = B.get_SOC()
lbs["scenario" + str(i), "states", 0, "SOC"] = B.get_SOC()
ubs["scenario" + str(i), "states", 0, "SOC"] = B.get_SOC()
s0["scenario" + str(i), "states", 0, "Pgb_p"] = Pgb_p
lbs["scenario" + str(i), "states", 0, "Pgb_p"] = Pgb_p
ubs["scenario" + str(i), "states", 0, "Pgb_p"] = Pgb_p
for k in range(N):
s_data["scenario" + str(i), "data", k,
"pv"] = pv_scenarios[i][k]
s_data["scenario" + str(i), "data", k, "l"] = l_scenarios[i][k]
s_data["scenario" + str(i), "data", k,
"E"] = (E_prediction[k] / actions_per_hour)
s_data["scenario" + str(i), "data", k, "prob"] = prob[i]
# Solve OCP
sol_time = time.time()
xk_opt, Uk_opt = ocp.solve_nlp([s0, lbs, ubs, lbg, ubg], s_data,
np.argmax(prob))
solver_time += time.time() - sol_time
# Simulate the system after disturbances
current_time += timedelta(minutes=10)
obs = observations[(observations["date"] >= current_time)
& (observations["date"] <= current_time +
timedelta(minutes=10 * N))]
pv_measured.append(pv_true)
l_measured.append(l_true)
E_measured.append(E_prediction[0])
time_stamps.append(current_time)
Uk_temp = np.copy(Uk_opt)
e, uk = utils.primary_controller(B.get_SOC(), Uk_opt,
obs["PV"].values[0],
obs["L"].values[0])
Pgb_p = np.max([Pgb_p_all[-1], uk[2]])
Pgb_p_all.append(Pgb_p)
errors.append(e)
Pbc.append(uk[0])
Pbd.append(uk[1])
Pgb.append(uk[2])
Pgs.append(uk[3])
# Calculate price from primary controller
temp_sale_diff = uk[3] - Uk_temp[3]
temp_buy_diff = uk[2] - Uk_temp[2]
if temp_buy_diff >= 0:
primary_Pgb = np.abs(
((uk[2] - Uk_temp[2]) *
E[E.time == current_time].price.values[0] / actions_per_hour))
primary_Pgs = 0
if temp_sale_diff >= 0:
primary_Pgs = np.abs(
((uk[3] - Uk_temp[3]) *
E[E.time == current_time].price.values[0] / actions_per_hour))
primary_Pgb = 0
B.simulate_SOC(xk_opt, [uk[0], uk[1]])
step_time = time.time() - step_start
sys_metrics.update_metrics(
[Pbc[-1], Pbd[-1], Uk_temp[2], Uk_temp[3]],
E[E.time == current_time].price.values[0] / actions_per_hour,
e,
Pgb_p_all[-1],
primary_Pgb,
primary_Pgs,
step_time,
)
bar.update(step)
sys_metrics.print_metrics(B.x_sim, E_measured)
df = utils.create_datafile(
[
time_stamps[1:],
100 * np.asarray(B.x_sim[1:]),
100 * np.asarray(B.x_opt[1:]),
np.asarray(Pbc) - np.asarray(Pbd),
np.asarray(Pgb) - np.asarray(Pgs),
ocp.Pbc - ocp.Pbd,
ocp.Pgb - ocp.Pgs,
np.asarray(errors),
pv_measured,
l_measured,
E_measured,
Pgb_p_all[1:],
],
[
"date",
"SOC_sim",
"SOC_opt",
"Pb_sim",
"Pg_sim",
"Pb_opt",
"Pg_opt",
"Errors",
"PV",
"Load",
"Spot_prices",
"P_peak",
],
logpath=logpath,
).set_index("date")
# Plotting
u = np.asarray(
[np.asarray(Pbc) - np.asarray(Pbd),
np.asarray(Pgb) - np.asarray(Pgs)])
p.plot_from_df(
df,
["Pb_sim"],
title="Battery Action",
upsample=True,
legends=["Battery"],
logpath=logpath,
)
p.plot_from_df(
df,
["Pg_sim", "Pg_opt", "P_peak"],
title="Grid Control Actions",
upsample=True,
legends=["Primary Control", "Optimal", "Peak Power"],
logpath=logpath,
)
p.plot_from_df(
df,
["SOC_sim"],
ylabel="SOC [%]",
title="State of Charge",
logpath=logpath,
)
p.plot_from_df(
df,
["PV", "Load"],
title="Measured PV and Load",
upsample=True,
legends=["PV", "Load"],
logpath=logpath,
)
p.plot_from_df(
df,
["Spot_prices"],
title="Spot Prices",
ylabel="Price [NOK]",
upsample=True,
logpath=logpath,
)
p.format_figure(
fig1,
ax1,
df.index,
title="PV Scenarios",
logpath=logpath,
)
p.format_figure(
fig1,
ax2,
df.index,
title="Load Scenarios",
logpath=logpath,
)
if conf["plot"]:
fig1.tight_layout()
if logpath:
fig1.savefig(logpath + "scenarios" + ".pdf", format="pdf")
plt.show(block=True)
def __init__(self, T, N, N_scenarios):
self.T = T
self.N = N
self.N_scenarios = N_scenarios
conf = parse_config()
conf_system = conf["system"]
# Get system constants
self.C_MAX = conf["battery"]["C_MAX"]
self.nb_c = conf["battery"]["nb_c"]
self.nb_d = conf["battery"]["nb_d"]
self.x_min = conf_system["x_min"]
self.x_max = conf_system["x_max"]
self.x_ref = conf_system["x_ref"]
self.Pb_max = conf_system["Pb_max"]
self.Pg_max = conf_system["Pg_max"]
self.battery_cost = conf_system["battery_cost"]
self.peak_cost = conf_system["peak_cost"]
self.old_grid_fee = conf_system["old_grid_fee"]
self.terminal_cost = conf_system["terminal_cost"]
self.verbose = conf_system["verbose"]
self.states = struct_symSX([entry("SOC"), entry("Pgb_p")])
self.inputs = struct_symSX([
entry("Pbc"),
entry("Pbd"),
entry("Pgb"),
entry("Pgs"),
])
self.slacks = struct_symSX([
entry("us"),
entry("ls"),
entry("s1"),
entry("s2"),
])
self.w = struct_symSX([
entry("states", struct=self.states, repeat=self.N),
entry("inputs", struct=self.inputs, repeat=self.N - 1),
])
self.data = struct_symSX([
entry("pv"),
entry("l"),
entry("E"),
entry("prob"),
])
self.all_data = struct_symSX(
[entry("data", struct=self.data, repeat=self.N)])
scenarios = []
s_data = []
for k in range(self.N_scenarios):
scenarios.append(entry("scenario" + str(k), struct=self.w))
s_data.append(entry("scenario" + str(k), struct=self.all_data))
self.s = struct_symSX(scenarios)
self.s_data = struct_symSX(s_data)
self.E = SX.sym("E", self.N)
# Initialize system properties
self.ode = self.build_ode()
self.L = None
self.F = None
self.solver = None
# Keep optimal solutions
self.SOC = np.asarray([])
self.Pbc = np.asarray([])
self.Pbd = np.asarray([])
self.Pgs = np.asarray([])
self.Pgb = np.asarray([])
def main():
"""
Main function for mpc-scheme with receding horizion.
"""
conf = utils.parse_config()
logpath = None
log = input("Do you wish to log this run? ")
if log in ["y", "yes", "Yes"]:
foldername = input("Do you wish to name logfolder? (enter to skip)")
logpath = utils.create_logs_folder(conf["logpath"], foldername)
openloop = False
predictions = conf["predictions"]
print("Using {} predictions.".format(predictions))
actions_per_hour = conf["actions_per_hour"]
horizon = conf["simulation_horizon"]
simulation_horizon = horizon * actions_per_hour
start_time = time.time()
step_time = start_time
PV, PV_pred, PL, PL_pred, grid_buy, grid_sell = utils.load_data()
T = conf["prediction_horizon"]
N = conf["prediction_horizon"] * actions_per_hour
xk = conf["x_inital"]
xk_sim = conf["x_inital"]
x_opt = np.asarray([xk])
x_sim = np.asarray([xk])
u0 = np.asarray([])
u1 = np.asarray([])
u2 = np.asarray([])
u3 = np.asarray([])
solver = OptiSolver(N)
x, lbx, ubx, lbg, ubg = solver.build_nlp(
T,
N,
)
net_cost_grid = 0
net_cost_bat = 0
J = 0
pv_preds = [PV[0]]
pl_preds = [PL[0]]
pv_error = []
pl_error = []
plt.figure()
if predictions in ["arima", "best"]:
pv_model = Arima("PV", order=(3, 1, 2))
pl_model = Arima("PL", order=(1, 1, 4), seasonal_order=(0, 0, 0, 0))
for step in range(simulation_horizon - N):
# Update NLP parameters
x[0] = xk
lbx[0] = xk
ubx[0] = xk
PV_true = PV[step:step + N]
PL_true = PL[step:step + N]
if predictions == "constant": # Predicted values equal to measurement
pv_ref = np.ones(N) * PV[step]
pl_ref = np.ones(N) * PL[step]
elif predictions == "arima": # Estimate using ARIMA
pv_model.update(PV[step])
pl_model.update(PL[step])
pv_ref = pv_model.predict(T)
pl_ref = pl_model.predict(T)
elif predictions == "data":
pv_ref = PV_pred[step:step + N]
pl_ref = PL_pred[step:step + N]
elif predictions == "scaled_mean":
pv_ref = (PV[step] / PV_pred[step]) * PV_pred[step:step + N]
pl_ref = (PL[step] / PL_pred[step]) * PL_pred[step:step + N]
elif predictions == "best":
pv_model.update(PV[step])
pv_ref = pv_model.predict(T)
pl_ref = (PL[step] / PL_pred[step]) * PL_pred[step:step + N]
else: # Use true predictions
pv_ref = PV_true
pl_ref = PL_true
pv_preds.append(pv_ref[1])
pl_preds.append(pl_ref[1])
pv_error.append(metrics.rmse(PV_true[0:4], pv_ref[0:4]))
pl_error.append(metrics.rmse(PL_true[0:4], pl_ref[0:4]))
plt.plot(range(step, step + N), pv_ref, c="b")
plt.plot(range(step, step + N), PV_true, c="r")
xk_opt, Uk_opt, J_opt = solver.solve_nlp([x, lbx, ubx, lbg, ubg],
vertcat(pv_ref, pl_ref))
J += J_opt
x_opt = np.append(x_opt, xk_opt[1])
xk_sim, Uk_sim = simulate_SOC(
xk_sim,
Uk_opt,
PV[step],
PL[step],
solver.F,
)
x_sim = np.append(x_sim, xk_sim)
if openloop:
xk = xk_opt[1] # xk is optimal
else:
xk = xk_sim
uk = [u[0] for u in Uk_opt]
u0 = np.append(u0, uk[0])
u1 = np.append(u1, uk[1])
u2 = np.append(u2, uk[2])
u3 = np.append(u3, uk[3])
net_cost_grid += metrics.net_spending_grid(uk, 1.5, actions_per_hour)
net_cost_bat += metrics.net_cost_battery(
uk, conf["system"]["battery_cost"], actions_per_hour)
if step % 50 == 0:
print("\nFinshed iteration step {}. Current step took {}s".format(
step, np.around(time.time() - step_time, 2)))
print("xsim {}%, x_opt {}%".format(np.around(xk_sim, 2),
np.around(xk_opt[1], 2)))
step_time = time.time()
peak_power = np.around(np.max(u2), 2) * 70
E_start = conf["x_inital"] * conf["system"]["C_MAX"]
E_end = xk * conf["system"]["C_MAX"]
battery_change = np.around(grid_buy * (E_end - E_start), 2)
print()
print("Error PV prediction:", np.mean(pv_error))
print("Error PL prediction:", np.mean(pl_error))
print("Net spending grid: {} kr".format(np.around(net_cost_grid, 2)))
print("Peak power cost: {} kr".format(peak_power))
print("Net spending battery: {} kr".format(np.around(net_cost_bat, 2)))
print("Grid + battery spending: {} kr".format(
np.around(net_cost_grid + net_cost_bat, 2), ))
print("Change in battery energy {} kr".format(battery_change))
print("Total spending:",
net_cost_grid + net_cost_bat - battery_change + peak_power)
# Plotting
u = np.asarray([-u0, u1, u2, -u3])
u_bat = np.asarray([-u0, u1])
u_grid = np.asarray([u2, -u3])
p.plot_control_actions(u, horizon - T, actions_per_hour, logpath)
p.plot_control_actions(
u_bat,
horizon - T,
actions_per_hour,
logpath,
title="Battery Controls",
legends=["Battery Charge", "Battery Discharge"],
)
p.plot_control_actions(
u_grid,
horizon - T,
actions_per_hour,
logpath,
title="Grid Controls",
legends=["Grid Buy", "Grid Sell"],
)
p.plot_SOC(x_sim, horizon - T, logpath)
p.plot_data(
[x_opt, x_sim],
logpath=logpath,
legends=["SOC optimal", "SOC simulated"],
title="Simulated vs optimal SOC",
)
p.plot_data(
[PV[:simulation_horizon - N], PL[:simulation_horizon - N]],
logpath=logpath,
legends=["PV Production", "Load Demands"],
title="PV Production & Load Demands",
)
p.plot_SOC_control_subplots(x_sim, u, horizon - T, logpath=logpath)
stop = time.time()
print("\nFinished optimation in {}s".format(np.around(
stop - start_time, 2)))
utils.save_datafile(
[x_opt, x_sim, u0, u1, u2, u3, PV, PV_pred, PL, PL_pred],
names=[
"x_opt",
"x_sim",
"u0",
"u1",
"u2",
"u3",
"PV",
"PV_pred",
"PL",
"PL_pred",
],
logpath=logpath,
)
print("One-step PV RMSE:", metrics.rmse_predictions(PV, pv_preds))
print("One-step Load RMSE:", metrics.rmse_predictions(PL, pl_preds))
if conf["plot_predictions"]:
p.plot_predictions_subplots(PV, pv_preds, PL, pl_preds, logpath)
plt.show(block=True)
plt.ion()
plt.close("all")
def nominel_mpc():
"""
Main function for mpc-scheme with receding horizion.
"""
conf = utils.parse_config()
datafile = conf["datafile"]
loads_trainfile = conf["loads_trainfile"]
logpath = None
log = input("Log this run? ")
if log in ["y", "yes", "Yes"]:
foldername = input("Enter logfolder name? (enter to skip) ")
logpath = utils.create_logs_folder(conf["logpath"], foldername)
openloop = conf["openloop"]
perfect_predictions = conf["perfect_predictions"]
actions_per_hour = conf["actions_per_hour"]
horizon = conf["simulation_horizon"]
simulation_horizon = horizon * actions_per_hour
T = conf["prediction_horizon"]
N = conf["prediction_horizon"] * actions_per_hour
start_time = time.time()
step_time = start_time
# Get data
observations = pd.read_csv(datafile, parse_dates=["date"])
# observations = observations[observations["date"] >= datetime(2021, 3, 11)]
solcast_forecasts = pd.read_csv(
conf["solcast_file"], parse_dates=["time", "collected"]
)
current_time = observations.date.iloc[0]
forecast = solcast_forecasts[
solcast_forecasts["collected"] == current_time - timedelta(minutes=60)
]
obs = observations[observations["date"] == current_time]
l = Load(N, loads_trainfile, "L", groundtruth=observations["L"])
E = np.ones(2000) # get_spot_price()
B = Battery(T, N, **conf["battery"])
PV = Photovoltaic()
Pbc = []
Pbd = []
Pgs = []
Pgb = []
pv_measured = []
l_measured = []
errors = []
ocp = NominelMPC(T, N)
sys_metrics = metrics.SystemMetrics()
x, lbx, ubx, lbg, ubg = ocp.build_nlp()
for step in range(simulation_horizon - N):
# Get measurements
x["states", 0, "SOC"] = B.get_SOC(openloop)
lbx["states", 0, "SOC"] = B.get_SOC(openloop)
ubx["states", 0, "SOC"] = B.get_SOC(openloop)
pv_true = obs["PV"].values[0]
l_true = obs["L"].values[0]
pv_measured.append(pv_true)
l_measured.append(l_true)
# Get new forecasts every hour
if current_time.minute == 30:
new_forecast = solcast_forecasts[
solcast_forecasts["collected"] == current_time - timedelta(minutes=30)
]
if new_forecast.empty:
print("Could not find forecast, using old forecast")
else:
forecast = new_forecast
ref = forecast[
(forecast["time"] >= current_time)
& (forecast["time"] < current_time + timedelta(minutes=10 * (N + 1)))
]
# Create predictions for next period
if perfect_predictions:
pv_ref = pv[step + 1 : step + N + 1]
l1_ref = l1.perfect_pred(step)
E_ref = E[step : step + N]
else:
pv_ref = PV.predict(ref.temp.values, ref.GHI.values)
l_ref = l.scaled_mean_pred(l_true, step % 126)
E_ref = E[step : step + N]
forecasts = ocp.update_forecasts(pv_ref, l_ref, E_ref)
xk_opt, Uk_opt = ocp.solve_nlp([x, lbx, ubx, lbg, ubg], forecasts)
# Simulate the system after disturbances
current_time += timedelta(minutes=10)
obs = observations[observations["date"] == current_time]
e, uk = utils.calculate_real_u(
xk_opt, Uk_opt, obs["PV"].values[0], obs["L"].values[0]
)
errors.append(e)
Pbc.append(uk[0])
Pbd.append(uk[1])
Pgb.append(uk[2])
Pgs.append(uk[3])
B.simulate_SOC(xk_opt, [uk[0], uk[1]])
sys_metrics.update_metrics(
[Pbc[step], Pbd[step], Pgb[step], Pgs[step]], E[step], e
)
utils.print_status(step, [B.get_SOC(openloop)], step_time, every=50)
step_time = time.time()
sys_metrics.calculate_consumption_rate(Pgs, pv_measured)
sys_metrics.calculate_dependency_rate(Pgb, l_measured)
sys_metrics.print_metrics()
# Plotting
u = np.asarray(
[np.asarray(Pbc) - np.asarray(Pbd), np.asarray(Pgb) - np.asarray(Pgs)]
)
if openloop:
p.plot_control_actions(
np.asarray([ocp.Pbc - ocp.Pbd, ocp.Pgb - ocp.Pgb]),
horizon - T,
actions_per_hour,
logpath,
legends=["Battery", "Grid"],
title="Optimal Control Actions",
)
else:
p.plot_control_actions(
u,
horizon - T,
actions_per_hour,
logpath,
legends=["Battery", "Grid"],
title="Simulated Control Actions",
)
p.plot_data(
np.asarray([B.x_sim, B.x_opt]),
title="State of charge",
legends=["SOC", "SOC_opt"],
)
p.plot_data([np.asarray(errors)], title="Errors")
p.plot_data(
np.asarray([pv_measured, l_measured]), title="PV & Load", legends=["PV", "L"]
)
# p.plot_data(np.asarray([E]), title="Spot Prices", legends=["Spotprice"])
stop = time.time()
print("\nFinished optimation in {}s".format(np.around(stop - start_time, 2)))
plt.ion()
if True:
plt.show(block=True)
def scenario_mpc():
"""
Main function for mpc-scheme with receding horizion.
"""
conf = utils.parse_config()
datafile = conf["datafile"]
loads_trainfile = conf["loads_trainfile"]
data = pd.read_csv("./data/data_oct20.csv",
parse_dates=["date"]).iloc[::10]
logpath = None
log = input("Log this run? ")
if log in ["y", "yes", "Yes"]:
foldername = input("Enter logfolder name? (enter to skip) ")
logpath = utils.create_logs_folder(conf["logpath"], foldername)
openloop = conf["openloop"]
perfect_predictions = conf["perfect_predictions"]
actions_per_hour = conf["actions_per_hour"]
horizon = conf["simulation_horizon"]
simulation_horizon = horizon * actions_per_hour
T = conf["prediction_horizon"]
N = conf["prediction_horizon"] * actions_per_hour
Nr = conf["robust_horizon"]
branch_factor = conf["branch_factor"]
N_scenarios = branch_factor**Nr
start_time = time.time()
step_time = start_time
# Get data
observations = pd.read_csv(datafile, parse_dates=["date"])
# observations = observations[observations["date"] >= datetime(2021, 3, 11)]
solcast_forecasts = pd.read_csv(conf["solcast_file"],
parse_dates=["time", "collected"])
current_time = observations.date.iloc[0]
forecast = solcast_forecasts[solcast_forecasts["collected"] ==
current_time - timedelta(minutes=60)]
obs = observations[observations["date"] == current_time]
l = Load(N, loads_trainfile, "L", groundtruth=observations["L"])
E = np.ones(2000) # get_spot_price()
B = Battery(T, N, **conf["battery"])
PV = Photovoltaic()
Pbc = []
Pbd = []
Pgs = []
Pgb = []
pv_measured = []
l_measured = []
errors = []
# Build reference tree
tree, leaf_nodes = build_scenario_tree(N, Nr,
branch_factor, np.ones(N + 1), 0,
np.ones(N + 1), 0)
ocp = ScenarioOCP(T, N, N_scenarios)
s_data = ocp.s_data(0)
s0, lbs, ubs, lbg, ubg = ocp.build_scenario_ocp()
sys_metrics = metrics.SystemMetrics()
for step in range(simulation_horizon - N):
# Get measurements
pv_true = obs["PV"].values[0]
l_true = obs["L"].values[0]
pv_measured.append(pv_true)
l_measured.append(l_true)
# Get new forecasts every hour
if current_time.minute == 30:
new_forecast = solcast_forecasts[solcast_forecasts["collected"] ==
current_time -
timedelta(minutes=30)]
if new_forecast.empty:
print("Could not find forecast, using old forecast")
else:
forecast = new_forecast
ref = forecast[(forecast["time"] >= current_time)
& (forecast["time"] < current_time +
timedelta(minutes=10 * (N + 1)))]
# Get predictions
pv_ref = PV.predict(ref.temp.values, ref.GHI.values)
l_ref = l.scaled_mean_pred(l_true, step % 126)
root, leaf_nodes = build_scenario_tree(N, Nr, branch_factor, pv_ref,
0.5, l_ref, 0.5)
pv_scenarios = get_scenarios(leaf_nodes, "pv")
l_scenarios = get_scenarios(leaf_nodes, "l")
# Update parameters
for i in range(N_scenarios):
s0["scenario" + str(i), "states", 0, "SOC"] = B.get_SOC(openloop)
lbs["scenario" + str(i), "states", 0, "SOC"] = B.get_SOC(openloop)
ubs["scenario" + str(i), "states", 0, "SOC"] = B.get_SOC(openloop)
for k in range(N):
s_data["scenario" + str(i), "data", k,
"pv"] = pv_scenarios[i][k]
s_data["scenario" + str(i), "data", k, "l"] = l_scenarios[i][k]
s_data["scenario" + str(i), "data", k, "E"] = 1
s_data["scenario" + str(i), "data", k, "prob"] = 1
xk_opt, Uk_opt = ocp.solve_nlp([s0, lbs, ubs, lbg, ubg], s_data)
# Simulate the system after disturbances
current_time += timedelta(minutes=10)
obs = observations[observations["date"] == current_time]
e, uk = utils.calculate_real_u(xk_opt, Uk_opt, obs["PV"].values[0],
obs["L"].values[0])
errors.append(e)
Pbc.append(uk[0])
Pbd.append(uk[1])
Pgb.append(uk[2])
Pgs.append(uk[3])
B.simulate_SOC(xk_opt, [uk[0], uk[1]])
sys_metrics.update_metrics([Pbc[-1], Pbd[-1], Pgb[-1], Pgs[-1]], E[-1],
e)
utils.print_status(step, [B.get_SOC(openloop)], step_time, every=50)
step_time = time.time()
sys_metrics.calculate_consumption_rate(Pgs, pv_measured)
sys_metrics.calculate_dependency_rate(Pgb, l_measured)
sys_metrics.print_metrics()
# Plotting
u = np.asarray(
[np.asarray(Pbc) - np.asarray(Pbd),
np.asarray(Pgb) - np.asarray(Pgs)])
p.plot_control_actions(
np.asarray([ocp.Pbc - ocp.Pbd, ocp.Pgb - ocp.Pgb]),
horizon - T,
actions_per_hour,
logpath,
legends=["Battery", "Grid"],
title="Optimal Control Actions",
)
p.plot_control_actions(
u,
horizon - T,
actions_per_hour,
logpath,
legends=["Battery", "Grid"],
title="Simulated Control Actions",
)
p.plot_data(
np.asarray([B.x_sim, B.x_opt]),
title="State of charge",
legends=["SOC", "SOC_opt"],
)
p.plot_data([np.asarray(errors)], title="Errors")
p.plot_data(
np.asarray([pv_measured, l_measured]),
title="PV and Load",
legends=["PV", "Load"],
)
# p.plot_data(np.asarray([E]), title="Spot Prices", legends=["Spotprice"])
stop = time.time()
print("\nFinished optimation in {}s".format(np.around(
stop - start_time, 2)))
plt.ion()
if True:
plt.show(block=True)
