Electric load is constantly changing under the influence of consumer habits, weather conditions, temperature, seasonality, weekdays, weekends and holidays. Behind every hour of electricity consumption lies a large amount of data which, if properly analysed, can support better planning of power system operations.
This lecture will introduce the basic concepts, methods and steps in short-term load forecasting (STLF), with a particular focus on the application of statistical methods, machine learning and artificial intelligence.
The lecture will cover the following topics:
- • what electric load forecasting is and why it matters;
• what short-term load forecasting involves;
• how daily, weekly and annual seasonality affect electricity consumption;
• why historical load is one of the most important predictors;
• how weather conditions, temperature and calendar factors influence forecasting;
• how data is prepared for STLF models;
• what feature engineering means in time series;
• which statistical, AI and machine learning methods are used;
• how model accuracy is evaluated;
• why interpretability is important in energy systems;
• what a modern short-term load forecasting system looks like.
The aim of the lecture is to show that STLF is not only a matter of choosing the best model, but a complete process that includes understanding data, the power system, seasonality, predictors, evaluation and real-world application.
Participants will gain a clear introduction to the field of short-term load forecasting and understand how data and artificial intelligence can be used for more reliable planning, more efficient management and a safer electricity supply.