Mining historical data enables power companies to predict and prevent power outages

Since 2014 Hassan Nemati’s research focus has been on estimating the reliability of underground power cables by using historical data. This means, estimating how reliable a cable is without installing additional sensors in a power grid. On September 29, Hassan Nemati will present the research during his licentiate seminar at Halmstad University.

”My research project is a collaborative work with Halmstad Energi och Miljö AB (HEM Nät), the electricity distribution company which operates in and around Halmstad city. The aim of this project is to reduce the number and duration of electricity outages at HEM’s electricity grid. We investigate and implement data mining techniques to discover patterns in previous outages, and model power cables’ lifetime”, says Hassan Nemati.

Hassan Nemati is a doctoral student at the Embedded and Intelligent Systems Industrial Graduate School (EISIGS) at Halmstad University. As an EISIGS-student, he works both at the University and at the industrial research partner – in this case HEM Nät.

”I spend one day per week at the company but this will probably be increased in the future. This collaborative work is very efficient for my research since I get the chance to meet and discuss with experts at the company regularly”, says Hassan Nemati.

The method is spreading to other Swedish power companies

Headshot of a man against a grey background

Hassan Nemati

When we met Hassan the project team was having a monthly meeting in a modern office building at HEM. Advisors from HEM, and supervisors from Halmstad University were gathered to see the latest results in Hassan Nemati’s research. On September 29, Hassan Nemati will present his licentiate thesis which is based on the research that he started in 2014.

In his licentiate thesis “Data-Driven Methods for Reliability Evaluation of Power Cables in Smart Distribution Grids”, historical data is used for failure pattern discovery and reliability evaluation of power cables. The method he proposed for failure pattern discovery got attention from other electricity distribution companies in Sweden such as Öresundskraft, Göteborg Energy, and Växjö Energy. These companies asked if their data also could be analysed by Hassan and, after receiving the outcomes, the companies claim that the results were very beneficial.

The method that Hassan Nemati has proposed for reliability evaluation of power cables is currently being evaluated by HEM Nät as a method to prevent failures of cables. According to the method, the reliability is estimated based on different factors such as previous failures, number of cable joint, and geographical position. The cables are then ranked for prioritising maintenance actions.

Research that will improve services for the customers

Alexander Örning and Peter Addicksson, both Electrical Engineering alumni from Halmstad University, are Hassan’s supervisors at HEM. They believe that the research collaboration is good for both the company and their customers.

”We could clearly see that this work, now and in the future, can help us in the maintenance work”, says Alexander Örning, manager and engineer at HEM Nät.

The future work of this project focuses on deviation detection of power components by using data collected from smart meters.

”Since 2009 we use smart meters that give us different types of data, such as power quality, active and reactive power, and alarms. We receive a huge amount of data and now, through the collaboration with Halmstad University, we can use it and try to predict problems before they occur”, says Alexander Örning.

”We have several ongoing collaborations with Halmstad University, using research to increase sustainability and improve services for our customers. We can´t sit and wait for others to come up with the innovations that we need. Hopefully we, in our turn, can spread Hassan’s knowledge, and also give him all the help he needs to reach great results”, says Alexander Örning.

Hassan Nemati will continue his research towards a PhD:

”My goal for the coming years is to investigate smart meters’ data. Mining and analysing these data can provide valuable information which could be used to improve reliability and increase efficiency.”

Text: Lotta Andersson and Louise Wandel
Photo: Magnus Karlsson

About the licentiate seminar

2017-09-29 at 10:00–12:00 Haldasalen, House J, Halmstad University

Licentiate thesis: "Data-Driven Methods for Reliability Evaluation of Power Cables in Smart Distribution Grids"

Examiner: Tomas Nordström, Halmstad University

Opponent: Niklas Lavesson, Blekinge Institute of Technology

Supervisor: Slawomir Nowaczyk and Anita Sant´Anna, Halmstad University

About the Embedded and Intelligent Systems Industrial Graduate School (EISIGS)

The goal EISIGS is to provide the right environment for producing qualified, independent doctoral students that understand, advance, and champion embedded and intelligent systems research. This goal is aligned with the largest research environment at the Halmstad University, Halmstad Embedded and Intelligent Systems Research (EIS) as well as with the University’s major research initiative, Research for Innovation, with long-term funding from The Knowledge Foundation. Funding for the school itself is from the Knowledge Foundation complemented by funding from industrial partners as well as from Halmstad University.

About the Knowledge Foundation Environment

Halmstad University is one of five universities in Sweden that has qualified to become a Knowledge Foundation Environment. This means that the foundation, on a long-term basis, invests in the University's key areas. Crucial to all research projects within the Knowledge Foundation Environment is the collaboration with business and commerce, known as co-production. This usually means that companies are engaged in research projects together with the University's researches and account for part of the financing. The environment is called Research for Innovation since Halmstad University's profile is The Innovation Driven University.