Predicting energy consumption and generation using Machine learning and Deep learning

Exploring household energy data by consumption, by generating (solar panel) by hourly, weekly, monthly, yearly, winter, summer.

• Keyword explanations (from PDF file with the data)

  • Generator Capacity: Solar panel capacity recorded on the application for connection for each customer.

  • Consumption Category:

    • GC:General Consumption for electricity supplied all the time excluding solar generation and controlled load supply
    • CL:Controlled Load Consumption (only in Australia) We will add CL and GG to find final energy generation.
    • GG:Gross Generation for electricity generated by the solar system with a gross metering configuration, measured separately to household loads

  • Half-hourly data: Kilowatt hours (kWh) of electrical energy consumed or generated in the half hour ending at 0:30 (eg. between 0:00 and 0:30).
    The value is positive regardless of whether it is consumption or generation.

Modeling:

• For energy generation:
  • Features:weather information(UV Index, Temperature, Humidity, Wind Speed, Wind direction, Rain chance, etc)
  • Using models such as **Random Forest, XGBoost, Logistic regression
• For energy consumption:
  • Features: time (will add other features)
  • Using time-series models such as **ARIMA, SARIMAX, Prophet and naive LSTM(DL)

More information on slide:Predicting energy consumption and generation.pdf