Spaces:
Sleeping
Sleeping
| import json | |
| import joblib | |
| import pandas as pd | |
| from sklearn.preprocessing import StandardScaler | |
| from pickle import load | |
| import numpy as np | |
| class RTUPipeline: | |
| scaler1 = None # RTU 1,2 | |
| scaler2 = None # RTU 3,4 | |
| def __init__(self, rtus=[1, 2, 3, 4], scaler1_path=None,scaler2_path=None): | |
| outputs = [ | |
| "sa_temp", | |
| "oadmpr_pct", | |
| "ra_temp", | |
| "oa_temp", | |
| "ma_temp", | |
| "sf_vfd_spd_fbk_tn", | |
| "rf_vfd_spd_fbk_tn", | |
| ] | |
| self.output_col_names = [ | |
| "hp_hws_temp", | |
| ] | |
| for rtu in rtus: | |
| for output in outputs: | |
| self.output_col_names.append(f"rtu_00{rtu}_{output}") | |
| self.input_col_names = [] | |
| for rtu in rtus: | |
| self.input_col_names.append(f"rtu_00{rtu}_sat_sp_tn") | |
| self.input_col_names = self.input_col_names + [ | |
| "air_temp_set_1", | |
| "air_temp_set_2", | |
| "dew_point_temperature_set_1d", | |
| "relative_humidity_set_1", | |
| "solar_radiation_set_1", | |
| ] | |
| self.num_inputs = len(self.input_col_names)-2 | |
| self.num_outputs = len(self.output_col_names)-14 | |
| self.column_names = self.output_col_names + self.input_col_names | |
| if scaler1_path: | |
| self.scaler1 = self.get_scaler(scaler1_path) | |
| if scaler2_path: | |
| self.scaler2 = self.get_scaler(scaler2_path) | |
| self.df = pd.DataFrame(columns=self.column_names) | |
| def get_scaler(self, scaler_path): | |
| return joblib.load(scaler_path) | |
| def get_window(self, df): | |
| len_df = len(df) | |
| print(len_df) | |
| if len_df > 30: | |
| df = df.rolling(window=30,min_periods=1).mean() | |
| return df[len_df - 31 : len_df].astype("float32") | |
| else: | |
| return None | |
| def transform_window(self, df_window): | |
| columns_scaler1 = [0] + list(range(1,15)) + [29,30] + list(range(33, 38)) | |
| columns_scaler2 = [0] + list(range(15, 29)) + [31,32] + list(range(33, 38)) | |
| return self.scaler1.transform(df_window.iloc[:, columns_scaler1]),self.scaler2.transform(df_window.iloc[:, columns_scaler2]) | |
| def prepare_input(self, df_trans): | |
| return df_trans[:30, :].reshape((1, 30, len(self.column_names)-16)) | |
| def extract_data_from_message(self, message): | |
| payload = json.loads(message.payload.decode()) | |
| len_df = len(self.df) | |
| k = {} | |
| for col in self.column_names: | |
| k[col] = payload[col] | |
| self.df.loc[len_df] = k | |
| return self.df | |
| def fit(self, message): | |
| df = self.extract_data_from_message(message) | |
| df_window = self.get_window(df) | |
| if df_window is not None: | |
| df_trans1,df_trans2 = self.transform_window(df_window) | |
| df_new1 = self.prepare_input(df_trans1) | |
| df_new2 = self.prepare_input(df_trans2) | |
| else: | |
| df_new1 = None | |
| df_trans1 = None | |
| df_new2 = None | |
| df_trans2 = None | |
| return df_new1, df_trans1, df_new2, df_trans2 | |