Slovenské elektrárne supercharges weather predictions with Azure and Power BI analytics

Reliable weather forecasting plays a crucial role in the business of green energy. It helps predict electricity demand and anticipate pricing trends. “Some countries build a lot of wind and solar power plants that provide cheaper electricity than fossil fuels. However, weather changes can lead to constantly fluctuating prices, making it a challenge to stay ahead of electricity trading,” explains Matúš Medžo, Sales Strategy Manager at Slovenské elektrárne (SEAS).

SEAS—a leading electricity provider in Slovakia—wanted to gain an edge in the industry by using an efficient weather prediction system. “We needed to predict how much electricity was produced by renewable power plants depending on weather changes, which ultimately affects the pricing of electricity,” says Medžo. “Our competitors also use forecasting systems, but we wanted a solution that could do the process quickly and accurately. This was important to stay ahead of the market and react to changes accordingly.”

Powering data-driven business decisions

SEAS wanted to leverage the potential of artificial intelligence (AI) models to provide accurate data to its users. The company’s legacy system took too much time and didn’t enable its employees to work efficiently. Analysts used to download data onto desktops, use scripts, and process everything manually.

“Running the analyses by hand meant we couldn’t do them periodically without wasting time. It was also tedious to download the data, then modulate and train models on local computers to process the information,” shares Ľubomir Franko, Sales Strategy Analyst at SEAS. “The entire process took nearly two hours and had to be repeated four times a day.”

To streamline its forecasting system, SEAS sought out AI-driven Microsoft solutions. “We arranged a hackathon to create a prototype from scratch because we didn’t have any existing machine learning solution at that time,” says Michal Krásny, Applications Architect at SEAS. “We created a proof of concept (PoC), including ingestion, data transformation, and aggregation to target BI Report using Azure data services in just three days.”

The platform offers several functions working together. The first part focuses on moving information to be processed—this is when Azure Data Factory transports meteorological data from servers and copies it to Azure Data Lake Analytics, where it is applied to a specific geographical area.

Then, the analysts at SEAS overlap the information with historical data such as wind speed, pressure, and temperature in order to train the model in Azure Databricks (AI). Finally, Power BI generates an easy-to-read report predicting how much energy will be produced for a specific area based on the weather information it received.

The new solution sped up the projection process dramatically. “Instead of spending hours doing forecasts, now it only takes 15 minutes to get accurate results—all without human intervention,” Franko adds.

Being able to move at a faster pace enables SEAS to stay on top of its game. The company now gets the predictions earlier than external providers, which means its sales team is able to react quicker to market changes in electricity trading. “We have estimated financial savings of around €100,000 a year. Our work is also smoother and employees can use their time more efficiently,” says Medžo.

Moving forward, SEAS wants to further explore the use of AI in other parts of its business. It is in process of building a modern central data warehouse platform based on Azure Synapse Analytics technology, while also developing numerous other BI solutions. “We want to use Azure services extensively as there are still lots of ways to use our data to drive better business decisions,” sums up Maroš Bubán, Data Architect at SEAS.