Dev/skforecast bump optim dynamics

CHANGELOG.md

@@ -1,5 +1,13 @@
 # Changelog
 
+## Unreleased
+### Improvement
+- Adding new constraints to limit the dynamics (kW/sec) of deferrable loads and battery power. The LP formulation works correctly and a work should be done on integrating the user input parameters to control this functionality.
+- Added new constraint to avoid battery discharging to the grid
+### Fix
+- Bumped version of skforecast from 0.6.0 to 0.8.0. Doing this mainly implies changing how the exogenous data is passed to fit and predict methods.
+- Fixed wrong path for csv files when using load cost and prod price forecasts.
+
 ## [0.4.10] - 2023-05-21
 ### Fix
 - Fixed wrong name of new cost sensor.

config_emhass.yaml

@@ -48,6 +48,10 @@ optim_conf:
   - lp_solver: 'PULP_CBC_CMD' # set the name of the linear programming solver that will be used
   - lp_solver_path: 'empty' # set the path to the LP solver
   - set_nocharge_from_grid: False # avoid battery charging from the grid
+  - set_nodischarge_to_grid: True # avoid battery discharging to the grid
+  - set_battery_dynamic: False # add a constraint to limit the dynamic of the battery power in power per time unit
+  - battery_dynamic_max: 0.9 # maximum dynamic positive power variation in percentage of battery maximum power
+  - battery_dynamic_min: -0.9 # minimum dynamic negative power variation in percentage of battery maximum power
 
 plant_conf:
   - P_grid_max: 9000 # The maximum power that can be supplied by the utility grid in Watts

docs/conf.py

@@ -22,7 +22,7 @@
 author = 'David HERNANDEZ'
 
 # The full version, including alpha/beta/rc tags
-release = '0.4.10'
+release = '0.4.11'
 
 # -- General configuration ---------------------------------------------------
 

docs/requirements.txt

@@ -12,7 +12,7 @@ pulp>=2.4
 pyyaml>=5.4.1
 netcdf4>=1.5.3
 tables==3.7.0
-skforecast==0.6.0
+skforecast==0.8.0
 markupsafe==2.1.2
 Jinja2<3.2
 sphinx==5.3.0

requirements.txt

@@ -12,4 +12,4 @@ pulp>=2.4
 pyyaml>=5.4.1
 netcdf4>=1.5.3
 tables==3.7.0
-skforecast==0.6.0
+skforecast==0.8.0

requirements_webserver.txt

@@ -12,7 +12,7 @@ pulp>=2.4
 pyyaml>=5.4.1
 netcdf4>=1.5.3
 tables==3.7.0
-skforecast==0.6.0
+skforecast==0.8.0
 flask>=2.0.3
 waitress>=2.1.1
 plotly>=5.6.0

scripts/use_cases_analysis.py

@@ -43,6 +43,7 @@ def get_forecast_optim_objects(retrieve_hass_conf, optim_conf, plant_conf,
 if __name__ == '__main__':
     get_data_from_file = False
     params = None
+    save_figures = False
     retrieve_hass_conf, optim_conf, plant_conf = get_yaml_parse(pathlib.Path(root+'/config_emhass.yaml'), use_secrets=True)
     rh = retrieve_hass(retrieve_hass_conf['hass_url'], retrieve_hass_conf['long_lived_token'], 
                         retrieve_hass_conf['freq'], retrieve_hass_conf['time_zone'],
@@ -71,25 +72,28 @@ def get_forecast_optim_objects(retrieve_hass_conf, optim_conf, plant_conf,
     fig_inputs1.update_yaxes(title_text = "Powers (W)")
     fig_inputs1.update_xaxes(title_text = "Time")
     fig_inputs1.show()
-    fig_inputs1.write_image(root + "/docs/images/inputs_power.svg", 
-                            width=1080, height=0.8*1080)
+    if save_figures:
+        fig_inputs1.write_image(root + "/docs/images/inputs_power.svg", 
+                                width=1080, height=0.8*1080)
 
     fig_inputs2 = df_input_data[['unit_load_cost',
                                  'unit_prod_price']].plot()
     fig_inputs2.layout.template = template
     fig_inputs2.update_yaxes(title_text = "Load cost and production sell price (EUR)")
     fig_inputs2.update_xaxes(title_text = "Time")
     fig_inputs2.show()
-    fig_inputs2.write_image(root + "/docs/images/inputs_cost_price.svg", 
-                            width=1080, height=0.8*1080)
+    if save_figures:
+        fig_inputs2.write_image(root + "/docs/images/inputs_cost_price.svg", 
+                                width=1080, height=0.8*1080)
 
     fig_inputs_dah = df_input_data_dayahead.plot()
     fig_inputs_dah.layout.template = template
     fig_inputs_dah.update_yaxes(title_text = "Powers (W)")
     fig_inputs_dah.update_xaxes(title_text = "Time")
     fig_inputs_dah.show()
-    fig_inputs_dah.write_image(root + "/docs/images/inputs_dayahead.svg", 
-                               width=1080, height=0.8*1080)
+    if save_figures:
+        fig_inputs_dah.write_image(root + "/docs/images/inputs_dayahead.svg", 
+                                   width=1080, height=0.8*1080)
 
     # Let's first perform a perfect optimization
     opt_res = opt.perform_perfect_forecast_optim(df_input_data, days_list)
@@ -98,8 +102,9 @@ def get_forecast_optim_objects(retrieve_hass_conf, optim_conf, plant_conf,
     fig_res.update_yaxes(title_text = "Powers (W)")
     fig_res.update_xaxes(title_text = "Time")
     fig_res.show()
-    fig_res.write_image(root + "/docs/images/optim_results_PV_defLoads_perfectOptim.svg", 
-                        width=1080, height=0.8*1080)
+    if save_figures:
+        fig_res.write_image(root + "/docs/images/optim_results_PV_defLoads_perfectOptim.svg", 
+                            width=1080, height=0.8*1080)
 
     print("System with: PV, two deferrable loads, perfect optimization, profit >> total cost function sum: "+\
         str(opt_res['cost_profit'].sum()))
@@ -113,8 +118,9 @@ def get_forecast_optim_objects(retrieve_hass_conf, optim_conf, plant_conf,
     fig_res_dah.update_yaxes(title_text = "Powers (W)")
     fig_res_dah.update_xaxes(title_text = "Time")
     fig_res_dah.show()
-    fig_res_dah.write_image(root + "/docs/images/optim_results_PV_defLoads_dayaheadOptim.svg", 
-                        width=1080, height=0.8*1080)
+    if save_figures:
+        fig_res_dah.write_image(root + "/docs/images/optim_results_PV_defLoads_dayaheadOptim.svg", 
+                                width=1080, height=0.8*1080)
 
     print("System with: PV, two deferrable loads, dayahead optimization, profit >> total cost function sum: "+\
         str(opt_res_dah['cost_profit'].sum()))
@@ -132,8 +138,9 @@ def get_forecast_optim_objects(retrieve_hass_conf, optim_conf, plant_conf,
     fig_res_dah.update_yaxes(title_text = "Powers (W)")
     fig_res_dah.update_xaxes(title_text = "Time")
     fig_res_dah.show()
-    fig_res_dah.write_image(root + "/docs/images/optim_results_defLoads_dayaheadOptim.svg", 
-                        width=1080, height=0.8*1080)
+    if save_figures:
+        fig_res_dah.write_image(root + "/docs/images/optim_results_defLoads_dayaheadOptim.svg", 
+                                width=1080, height=0.8*1080)
 
     print("System with: two deferrable loads, dayahead optimization, profit >> total cost function sum: "+\
         str(opt_res_dah['cost_profit'].sum()))
@@ -152,15 +159,17 @@ def get_forecast_optim_objects(retrieve_hass_conf, optim_conf, plant_conf,
     fig_res_dah.update_yaxes(title_text = "Powers (W)")
     fig_res_dah.update_xaxes(title_text = "Time")
     fig_res_dah.show()
-    fig_res_dah.write_image(root + "/docs/images/optim_results_PV_Batt_defLoads_dayaheadOptim.svg", 
-                        width=1080, height=0.8*1080)
+    if save_figures:
+        fig_res_dah.write_image(root + "/docs/images/optim_results_PV_Batt_defLoads_dayaheadOptim.svg", 
+                                width=1080, height=0.8*1080)
     fig_res_dah = opt_res_dah[['SOC_opt']].plot()
     fig_res_dah.layout.template = template
     fig_res_dah.update_yaxes(title_text = "Battery State of Charge (%)")
     fig_res_dah.update_xaxes(title_text = "Time")
     fig_res_dah.show()
-    fig_res_dah.write_image(root + "/docs/images/optim_results_PV_Batt_defLoads_dayaheadOptim_SOC.svg", 
-                        width=1080, height=0.8*1080)
+    if save_figures:
+        fig_res_dah.write_image(root + "/docs/images/optim_results_PV_Batt_defLoads_dayaheadOptim_SOC.svg", 
+                                width=1080, height=0.8*1080)
 
     print("System with: PV, Battery, two deferrable loads, dayahead optimization, profit >> total cost function sum: "+\
         str(opt_res_dah['cost_profit'].sum()))

setup.py

@@ -19,7 +19,7 @@
 
 setup(
     name='emhass',  # Required
-    version='0.4.10',  # Required
+    version='0.4.11',  # Required
     description='An Energy Management System for Home Assistant',  # Optional
     long_description=long_description,  # Optional
     long_description_content_type='text/markdown',  # Optional (see note above)
@@ -57,7 +57,7 @@
         'pyyaml>=5.4.1',
         'netcdf4>=1.5.3',
         'tables==3.7.0',
-        'skforecast==0.6.0',
+        'skforecast==0.8.0',
     ],  # Optional
     entry_points={  # Optional
         'console_scripts': [

src/emhass/command_line.py

@@ -229,7 +229,8 @@ def dayahead_forecast_optim(input_data_dict: dict, logger: logging.Logger,
         input_data_dict['df_input_data_dayahead'],
         method=input_data_dict['fcst'].optim_conf['load_cost_forecast_method'])
     df_input_data_dayahead = input_data_dict['fcst'].get_prod_price_forecast(
-        df_input_data_dayahead, method=input_data_dict['fcst'].optim_conf['prod_price_forecast_method'])
+        df_input_data_dayahead, 
+        method=input_data_dict['fcst'].optim_conf['prod_price_forecast_method'])
     opt_res_dayahead = input_data_dict['opt'].perform_dayahead_forecast_optim(
         df_input_data_dayahead, input_data_dict['P_PV_forecast'], input_data_dict['P_load_forecast'])
     # Save CSV file for publish_data

src/emhass/forecast.py

@@ -658,7 +658,7 @@ def get_load_forecast(self, days_min_load_forecast: Optional[int] = 3, method: O
         return P_Load_forecast
 
     def get_load_cost_forecast(self, df_final: pd.DataFrame, method: Optional[str] = 'hp_hc_periods',
-                               csv_path: Optional[str] = "/data/data_load_cost_forecast.csv") -> pd.DataFrame:
+                               csv_path: Optional[str] = "data_load_cost_forecast.csv") -> pd.DataFrame:
         r"""
         Get the unit cost for the load consumption based on multiple tariff \
         periods. This is the cost of the energy from the utility in a vector \
@@ -671,7 +671,7 @@ def get_load_cost_forecast(self, df_final: pd.DataFrame, method: Optional[str] =
             and 'csv' to load a CSV file, defaults to 'hp_hc_periods'
         :type method: str, optional
         :param csv_path: The path to the CSV file used when method = 'csv', \
-            defaults to "/data/data_load_cost_forecast.csv"
+            defaults to "data_load_cost_forecast.csv"
         :type csv_path: str, optional
         :return: The input DataFrame with one additionnal column appended containing
             the load cost for each time observation.

src/emhass/machine_learning_forecaster.py

@@ -97,6 +97,17 @@ def neg_r2_score(y_true, y_pred):
         """The negative of the r2 score."""
         return -r2_score(y_true, y_pred)
 
+    @staticmethod
+    def generate_exog(data_last_window, periods, var_name):
+        """Generate the exogenous data for future timestamps."""
+        forecast_dates = pd.date_range(start=data_last_window.index[-1]+data_last_window.index.freq, 
+                                       periods=periods, 
+                                       freq=data_last_window.index.freq)
+        exog = pd.DataFrame({var_name:[np.nan]*periods},
+                            index=forecast_dates)
+        exog = mlforecaster.add_date_features(exog)
+        return exog
+
     def fit(self, split_date_delta: Optional[str] = '48h', perform_backtest: Optional[bool] = False
             ) -> Tuple[pd.DataFrame, pd.DataFrame]:
         r"""The fit method to train the ML model.
@@ -119,7 +130,7 @@ def fit(self, split_date_delta: Optional[str] = '48h', perform_backtest: Optiona
         # train/test split
         self.date_train = self.data_exo.index[-1]-pd.Timedelta('5days')+self.data_exo.index.freq # The last 5 days
         self.date_split = self.data_exo.index[-1]-pd.Timedelta(split_date_delta)+self.data_exo.index.freq # The last 48h
-        self.data_train = self.data_exo.loc[:self.date_split,:]
+        self.data_train = self.data_exo.loc[:self.date_split-self.data_exo.index.freq,:]
         self.data_test  = self.data_exo.loc[self.date_split:,:]
         self.steps = len(self.data_test)
         # Pick correct sklearn model
@@ -143,7 +154,7 @@ def fit(self, split_date_delta: Optional[str] = '48h', perform_backtest: Optiona
                             exog=self.data_train.drop(self.var_model, axis=1))
         self.logger.info(f"Elapsed time for model fit: {time.time() - start_time}")
         # Make a prediction to print metrics
-        predictions = self.forecaster.predict(steps=self.steps, exog=self.data_train.drop(self.var_model, axis=1))
+        predictions = self.forecaster.predict(steps=self.steps, exog=self.data_test.drop(self.var_model, axis=1))
         pred_metric = r2_score(self.data_test[self.var_model],predictions)
         self.logger.info(f"Prediction R2 score of fitted model on test data: {pred_metric}")
         # Packing results in a DataFrame
@@ -187,18 +198,18 @@ def predict(self, data_last_window: Optional[pd.DataFrame] = None
         :rtype: pd.Series
         """
         if data_last_window is None:
-            predictions = self.forecaster.predict(steps=self.num_lags, exog=self.data_train.drop(self.var_model, axis=1))
+            predictions = self.forecaster.predict(steps=self.num_lags, exog=self.data_test.drop(self.var_model, axis=1))
         else:
-            data_last_window = mlforecaster.add_date_features(data_last_window)
-            data_last_window = data_last_window.interpolate(method='linear', axis=0, limit=None)
             if self.is_tuned:
+                exog = mlforecaster.generate_exog(data_last_window, self.lags_opt, self.var_model)
                 predictions = self.forecaster.predict(steps=self.lags_opt, 
                                                       last_window=data_last_window[self.var_model],
-                                                      exog=data_last_window.drop(self.var_model, axis=1))
+                                                      exog=exog.drop(self.var_model, axis=1))
             else:
+                exog = mlforecaster.generate_exog(data_last_window, self.num_lags, self.var_model)
                 predictions = self.forecaster.predict(steps=self.num_lags, 
                                                       last_window=data_last_window[self.var_model],
-                                                      exog=data_last_window.drop(self.var_model, axis=1))
+                                                      exog=exog.drop(self.var_model, axis=1))
         return predictions
 
     def tune(self, debug: Optional[bool] = False) -> pd.DataFrame:
@@ -223,11 +234,11 @@ def tune(self, debug: Optional[bool] = False) -> pd.DataFrame:
         if self.sklearn_model == 'LinearRegression':
             if debug:
                 def search_space(trial):
-                    search_space  = {'fit_intercept': trial.suggest_categorical('fit_intercept', ['True'])} 
+                    search_space  = {'fit_intercept': trial.suggest_categorical('fit_intercept', [True])} 
                     return search_space
             else:
                 def search_space(trial):
-                    search_space  = {'fit_intercept': trial.suggest_categorical('fit_intercept', ['True', 'False'])} 
+                    search_space  = {'fit_intercept': trial.suggest_categorical('fit_intercept', [True, False])} 
                     return search_space
         elif self.sklearn_model == 'ElasticNet':
             if debug:
@@ -276,7 +287,7 @@ def search_space(trial):
         )
         self.logger.info(f"Elapsed time: {time.time() - start_time}")
         self.is_tuned = True
-        predictions_opt = self.forecaster.predict(steps=self.num_lags, exog=self.data_train.drop(self.var_model, axis=1))
+        predictions_opt = self.forecaster.predict(steps=self.num_lags, exog=self.data_test.drop(self.var_model, axis=1))
         freq_hours = self.data_exo.index.freq.delta.seconds/3600
         self.lags_opt = int(np.round(len(self.optimize_results.iloc[0]['lags'])))
         self.days_needed = int(np.round(self.lags_opt*freq_hours/24))