Short-term global load forecasting model for smart meters (LightGBM)
A forecasting service for predicting the positive active energy (in kWh) consumption of prosumers in the Italian city of Terni. The dataset was provided by ASM and the service makes use of a LightGBM model. (AIExperiments Asset)
This is a forecasting service for predicting the positive active energy (in kWh) consumption of prosumers in the Italian city of Terni. The dataset was provided by ASM. The core of the service is a global LightGBM model, which has been trained on the positive active energy time series of all smart meters for the time period 2023-06-08 to 2023-10-07. Temperature and radiation have been used as covariates, and the model can use them by calling an API during inference. The service is served as a docker container and a client script is also provided to help the user form their inference requests. The model is totally configurable in terms of:
- Predicted smart meter (client input entrypoint ts_id_pred): Chose the smart meter code that needs to be predicted. Example files have been provided for all supported smart meters. Smart meter codes supported are:
- BB6020
- BB6030
- BB6052
- BB6062
- BB6065
- BB6103
- BB6150
- BB6152
- BB6154
- BB6155
- BB6156
- BB6157
- BB6158
- BB6159
- BB6160
- BB6161
- BB6162
- BB6163
- BB6164
- BB6166
- BB6167
- BB6168
- BB6169
- BB6170
- BB6171
- BB6173
- BB6174
- BB6175
- BB6176
- BB6177
- BB6178
- BB6179
- BB6180
- BB6181
- BB6182
- BB6183
- BB6184
- BB6185
- BB6186
- BB6188
- BB6189
- BB6190
- BB6191
- BB6192
- BB6195
- BB6197
- BB6198
- BB6199
- Provided ground truth data points (client input entrypoint series_uri): The client can update the existing model with the desired length of new data points that have been observed.
- Forecast horizons (client input entrypoint hours_ahead): The client can request a forecast horizon of their preference.
This model has been developed within I-NERGY project. This asset has been published in AIExperiments.