Event under the auspices of the Ministry of Economy of Slovak Republic

Estimation of overhead conductor temperature using artificial neural networks

Klemen Deželak
University of Maribor, Faculty of Electrical Engineering and

Jože Pihler
University of Maribor, Faculty of Electrical Engineering and Computer Science

     Last modified: 2017-06-13

Dynamic thermal ratings of overhead lines could represent a promising approach to increase the transmission capacity by considering a weather dependent thermal ratings of overhead power lines in real time instead of using constant ratings. Knowledge about aforementioned is not only needed in real time observation but as well in a day-ahead basis within a network operational planning and in order to assess network security [1].
The key elements in the development of a dynamic-line rating system are the conductor type, ambient temperature, solar radiation and wind speed with its direction in a real time [2]. Joule and magnetic heating refer to the conductor heating due to its resistance and due to cyclic magnetic flux which causes heating by eddy currents, hysteresis and magnetic viscosity (magnetic heating) [3], while for precise evaluation of solar heating, numerous parameters are relevant, e.g. the conductor’s inclination to the horizontal or the intensity of the direct as well as diffuse solar radiation [4]. On the cooling side the forced and the natural convective cooling (wind speed and its direction) are important, while further heat losses occur by a thermal radiation, depending on conductor surface temperature [4].

This paper aimed to use the artificial neural network to estimate the conductor temperature of an overhead power line. Using both, the measured and calculated data, the accuracy of the artificial neural network will be evaluated. In that manner, to train the artificial neural network, ambient temperature, solar radiation and wind speed values are an inputs, and the conductor temperature is an output. So, the artificial neural network is used to estimate the conductor temperature, while the feed forward artificial neural network will be trained by using a different algorithms. The input learning signal (pattern) of the artificial neural network in the learning process is the value of the aforementioned variables. According to the conductor temperature, the output learning signal of the presented solution (target) will be set to the value zero or one.

In the paper will be shown that application of artificial neural network, as part of aforementioned problem, could represent a promising solution.

[1] Ringelband, Tilman, Philipp Schäfer, and Albert Moser. ''Probabilistic ampacity forecasting for overhead lines using weather forecast ensembles.'' Electrical Engineering 95.2 (2013).

[2] Kim, S. D., and Medhat M. Morcos. ''An application of dynamic thermal line rating control system to up-rate the ampacity of overhead transmission lines.'' IEEE Transactions on power delivery 28.2 (2013): 1231-1232.

[3] Staszewski, L., and W. Rebizant. ''The differences between IEEE and CIGRE heat balance concepts for line ampacity considerations.'' Modern Electric Power Systems (MEPS), 2010 Proceedings of the International Symposium. IEEE, 2010.

[4] Krontiris, Thanos, Andreas Wasserrab, and Gerd Balzer. ''Weather-based loading of overhead lines—Consideration of conductor's heat capacity.'' Modern Electric Power Systems (MEPS), 2010 Proceedings of the International Symposium. IEEE, 2010.


Proceedings Indexing

Recent proceedings have been recognized by Thomson Reuters for including in Conference Proceedings Citation Index - Science (CPCI-S).

View More

Original Research Reporting

We recognize international reporting standards. Authors of reports of original research should present an accurate account of the work performed as well as an objective discussion of its significance. .

View More

Call for Papers Announcement

We call for papers that address theoretical, experimental, systems-related and power network management issues...

View More