updating load and renewable generation with new numpy random generator

chronix2grid/generation/consumption/consumption_utils.py

@@ -63,8 +63,8 @@ def compute_residential(locations, Pmax, temperature_noise, params, weekly_patte
     nb_sec_per_day =  24. * 60. * 60.
     nb_sec_per_year = (365. * nb_sec_per_day)
     year_pattern = 2. * np.pi / nb_sec_per_year
-    #seasonal_pattern = 1.5 / 7. * np.cos(year_pattern * (t + start_min - 45 * nb_sec_per_day))  # min of the load is 15 of February so 45 days after beginning of year
-    seasonal_pattern = 1.5 / 7. * np.cos(year_pattern * (t - start_min - 30 * nb_sec_per_day))
+    seasonal_pattern = 1.5 / 7. * np.cos(year_pattern * (t + start_min - 45 * nb_sec_per_day))  # min of the load is 15 of February so 45 days after beginning of year
+    #seasonal_pattern = 1.5 / 7. * np.cos(year_pattern * (t - start_min - 30 * nb_sec_per_day)) #older version to be removed
     seasonal_pattern += 5.5 / 7.
 
     # Get weekly pattern
@@ -142,11 +142,11 @@ def create_csv(prng, dict_, path, forecasted=False, reordering=True, noise=None,
 
     df_reactive_power = 0.7 * df
     if noise is not None:
-        #df *= prng.lognormal(mean=0.0,sigma=noise, size=df.shape)
-        df *= np.random.lognormal(mean=0.0, sigma=noise, size=df.shape)
-        #df_reactive_power *= prng.lognormal(mean=0.0, sigma=noise, size=df.shape)
-        df_reactive_power *= np.random.lognormal(mean=0.0, sigma=noise,
-                                                 size=df.shape)
+        df *= prng.lognormal(mean=0.0,sigma=noise, size=df.shape)
+        #df *= np.random.lognormal(mean=0.0, sigma=noise, size=df.shape) #older version to be removed
+        df_reactive_power *= prng.lognormal(mean=0.0, sigma=noise, size=df.shape)
+        #df_reactive_power *= np.random.lognormal(mean=0.0, sigma=noise,
+        #                                         size=df.shape) #older version to be removed
 
     if write_results:
         file_extension = '_forecasted' if forecasted else ''

chronix2grid/generation/consumption/generate_load.py

@@ -40,7 +40,7 @@ def main(scenario_destination_path, seed, params, loads_charac, load_weekly_patt
 
     # Set random seed of scenario
     prng = default_rng(seed)
-    np.random.seed(seed)
+    #np.random.seed(seed) #older version - to be removed
 
     # Define reference datetime indices
     datetime_index = pd.date_range(
@@ -57,7 +57,7 @@ def main(scenario_destination_path, seed, params, loads_charac, load_weekly_patt
         y_plus = int(y // dy_corr + 1)
         add_dim = max(y_plus, add_dim)
         add_dim = max(x_plus, add_dim)
-        add_dim=0
+        #add_dim=0 #to get back to when this parameter did not exist - to be removed
 
     # Generate GLOBAL temperature noise
     print('Computing global auto-correlated spatio-temporal noise for thermosensible demand...') ## temperature is simply to reflect the fact that loads is correlated spatially, and so is the real "temperature". It is not the real temperature.

chronix2grid/generation/generation_utils.py

@@ -65,8 +65,8 @@ def generate_coarse_noise(prng, params, data_type, add_dim):
     Nt_comp = int(T // dt_corr + 1) + add_dim
 
     # Generate gaussian noise input·
-    output = np.random.normal(0, 1, (Nx_comp, Ny_comp, Nt_comp))
-    #output = prng.normal(0, 1, (Nx_comp, Ny_comp, Nt_comp))
+    #output = np.random.normal(0, 1, (Nx_comp, Ny_comp, Nt_comp))
+    output = prng.normal(0, 1, (Nx_comp, Ny_comp, Nt_comp))
 
     return output
 

chronix2grid/generation/renewable/generate_solar_wind.py

@@ -44,7 +44,7 @@ def main(scenario_destination_path, seed, params, prods_charac, solar_pattern, w
     """
 
     prng = default_rng(seed)
-    np.random.seed(seed)
+    #np.random.seed(seed) #olver version - to be removed
     smoothdist = params['smoothdist']
 
     # Define datetime indices
@@ -81,10 +81,6 @@ def main(scenario_destination_path, seed, params, prods_charac, solar_pattern, w
         add_dim = max(y_plus, add_dim)
         add_dim = max(x_plus, add_dim)
 
-        ##########
-        #Getting back to old model
-        add_dim=0
-        ##########
     solar_noise = utils.generate_coarse_noise(prng, params, 'solar', add_dim=add_dim)
     long_scale_wind_noise = utils.generate_coarse_noise(prng, params, 'long_wind', add_dim=add_dim)
     medium_scale_wind_noise = utils.generate_coarse_noise(prng, params, 'medium_wind', add_dim=add_dim)

chronix2grid/generation/renewable/solar_wind_utils.py

@@ -50,9 +50,9 @@ def compute_wind_series(prng, locations, Pmax, long_noise, medium_noise, short_n
     signal = (0.7 + 0.3 * seasonal_pattern) * (0.3 + std_medium_wind_noise * medium_scale_signal + std_long_wind_noise * long_scale_signal)
     signal += std_short_wind_noise * short_scale_signal
     signal = 1e-1 * np.exp(4 * signal)
-    # signal += prng.uniform(0, SMOOTHDIST/Pmax, signal.shape)
-    #signal += prng.uniform(0, smoothdist, signal.shape)
-    signal += np.random.uniform(0, smoothdist, signal.shape)
+    #signal += prng.uniform(0, SMOOTHDIST/Pmax, signal.shape)
+    signal += prng.uniform(0, smoothdist, signal.shape)
+    #signal += np.random.uniform(0, smoothdist, signal.shape) #older version - to be removed
 
     # signal *= 0.95
     signal[signal < 0.] = 0.
@@ -80,14 +80,14 @@ def compute_solar_series(prng, locations, Pmax, solar_noise, params, solar_patte
         mean_solar_pattern = 0.75
 
     signal = solar_pattern * (mean_solar_pattern + std_solar_noise * final_noise)
-    # signal += prng.uniform(0, smoothdist/Pmax, signal.shape)
-    signal += np.random.uniform(0, smoothdist / Pmax, signal.shape)
+    #signal += prng.uniform(0, smoothdist/Pmax, signal.shape) #to be revised: since smmothdist/PMax is very small, the added noise compared to the previous sinal was unsignificant
+    #signal += np.random.uniform(0, smoothdist / Pmax, signal.shape) #older version - to be removed
     # signal[signal > 1] = 1
     signal[signal < 0.] = 0.
     signal = smooth(signal)
     solar_series = Pmax * signal
     # solar_series[np.isclose(solar_series, 0.)] = 0
-    #solar_series[solar_series > 0.95 * Pmax] = 0.95 * Pmax
+    solar_series[solar_series > 0.95 * Pmax] = 0.95 * Pmax
     return solar_series
 
 def compute_solar_pattern(params, solar_pattern):
@@ -178,8 +178,8 @@ def create_csv(prng, dict_, path, reordering=True, noise=None, shift=False,
         new_ordering = [x for _ ,x in sorted(zip(value ,list(df)))]
         df = df[new_ordering]
     if noise is not None:
-        #df *= ( 1 +noise * prng.normal(0, 1, df.shape))
-        df *= (1 + noise * np.random.normal(0, 1, df.shape))
+        df *= ( 1 +noise * prng.normal(0, 1, df.shape))
+        #df *= (1 + noise * np.random.normal(0, 1, df.shape)) #older version - to be removed
     if shift:
         df = df.shift(-1)
         df = df.fillna(0)