Master's lecture in Computer Science - Guðmundur Smári Guðmundsson | University of Iceland Skip to main content

Master's lecture in Computer Science - Guðmundur Smári Guðmundsson

When 
Wed, 03/06/2020 - 13:00 to 13:45
Where 
Further information 
Free admission

The lecture will be held via Zoom: https://zoom.us/j/65202558564

Master's student: Guðmundur Smári Guðmundsson

Title: Prediction of time series for electricity generation

 ___________________________________________

Faculty: Faculty of Industrial Engineering, Mechanical Engineering and Computer Science

Advisor:  Helmut Neukirchen, Professor at the Faculty of Industrial Engineering, Mechanical Engineering and Computer Science

Other members of the masters committee: Morris Riedel, Assistant Professor at the Faculty of Industrial Engineering, Mechanical Engineering and Computer Science
Ólafur Pétur Pálsson, Professor at the Faculty of Industrial Engineering, Mechanical Engineering and Computer Science

Examiner:  Sebastian Lührs, Member of the Division Application Support at Jülich Supercomputing Centre, Forschungszentrum Jülich research centre

Abstract

Electric energy meters are used for measuring how much energy is generated per hour in a power station. These measurements are time series which are typically only available at the end of each month, nevertheless the data needs to be avail-able as soon as possible.  In this thesis, using near real-time data, two methods are presented for time series predictions: a ratio method and a deep learning Long Short-Term Memory (LSTM) neural network method. The results from these methods are compared by two error metrics, Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) with an emphasis is on a lower RMSE. Both methods are applied to three hydro power stations: Ljósafoss, Hrauneyjafoss, and Fljótsdalur. The best acquired RMSE value for each station is: 0.066, 1.651, and 2.667 respectively. While the ratio method achieves a low RMSE for one station, the LSTM method achieves the lowest RMSE for all three power stations. The results conclude the LSTM method to be a good choice for time series predictions for other hydro power stations, improving speed of a data analysis by making data predictions available innear real-time.