Ebrahim is breaking new ground with machine learning methods for power quality analysis

Unipower is offering a new product together with the startup Eneryield – an analytics report which automates the process of finding root-causes of power quality disturbances. The underlying technology builds on research made by Ebrahim Balouji (Ph.D Electrical Engineering) and Karl Bäckström (Ph.D.c Computer Science) at Chalmers University of Technology. Ebrahim has eight years of experience within data mining of power quality analytics, grid modeling and compensation of power quality problems. He believes there are great opportunities to modernize the power grids.

The world is becoming more digitized and power systems are following the trend. As more sensors are deployed in the power grids, the amount of data collected is also increasing. This creates great opportunities to add value by analysing data in more sophisticated ways, for example by using novel machine learning methods. These have become suited to analyze power grid data as the availability of computational power, data storage and communication have improved recently. Benefits created revolve around energy efficiency, reliability and predictive maintenance.

Ebrahim Balouji has developed new, state-of-the-art deep learning and machine learning methods to analyze signals related current and voltage. He is currently doing his second PhD within Signal Processing at the Electrical Engineering department at Chalmers University of Technology.

“Machine learning makes it possible to utilize the large amounts of data available on the power grids to extract valuable insights, which can give power grids a higher performance and make them more effective,” Ebrahim says.

By using supervised and unsupervised deep learning techniques, Ebrahim’s methods can classify power quality events and earth faults to identify their root-cause and the direction with a very high accuracy. This process usually requires manual effort and expertise in the field, but Ebrahim’s work shows that this process could be automated with improved accuracy.

This is one important cornerstone in Eneryield’s work to deliver an accurate model of the power grid at any location and point in time. By pinpointing weaknesses in power systems, it is possible to strengthen them and improve security of supply and energy efficiency. Ebrahim has also classified and analyzed partial discharges with impressive results, which is used to estimate the lifetime of electronic devices, such as motors, to understand when maintenance is needed.

The ability to predict power quality events and variations, like harmonics and flicker, is another one of Ebrahim’s findings. This research has been performed together with his fellow colleague Karl Bäckström, who also doctorates at Chalmers. In order to predict power quality, special pre-processing methods for the signal are used. The developed method resulted in the possibility of predicting power quality events and variations upto desired horizons. Applying this innovation in for example active power filters, devices used to compensate power quality disturbances, could reduce their current delays. This would make them act in real time with a higher efficiency and improve the power quality in affected areas.

If you want to read about Ebrahim’s research in detail, it is possible to find his articles here: https://ieeexplore.ieee.org/author/37085455267

Unipower and the startup Eneryield (www.eneryield.com) are now offering a new report for power quality analytics based on Ebrahim’s research. The machine learning based insights delivered in the report makes it possible for power utilities to improve the reliability and efficiency of their power systems.

Comments are closed.

By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.

Close