Prediction of power cable conductor temperature in heating cycle test by machine learning methods
Electric Power Systems Research, cilt.259, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 259
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.epsr.2026.113231
- Dergi Adı: Electric Power Systems Research
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, Environment Index, INSPEC, Academic Search Ultimate (EBSCO), Engineering Source (EBSCO)
- Anahtar Kelimeler: Heating cycle, Induction heating, Machine learning, Power cable, Reduced error pruning tree, Thermal aging
- Kırklareli Üniversitesi Adresli: Evet
Özet
Medium-voltage (MV) power cables utilized in distribution systems undergo various types of electrical and non-electrical tests. In accordance with IEC 60502-2, a sequence of tests should be performed, comprising a heating cycle test as well as tests conducted both before and after it. For the heating cycle test, the cable was stripped to the conductor to monitor the temperature. This situation requires working on two cable samples to apply the tanδ, partial discharge, impulse, and voltage tests after the heating cycle test with an undeformed cable. Because performing subsequent tests with a deformed cable leads to inaccurate test results. To alleviate this procedure and enable testing with a single cable, this study has demonstrated that the prediction of cable conductor temperature can be achieved using different machine learning (ML) methods. The handled data for prediction were obtained by measurements taken from certain intervals in heating cycles carried out 20 times for a cable with a voltage level of 35 kV and a cross-section of 400 mm2. With an average prediction error of 3.53% using the Reduced Error Pruning Tree (REPTree) method, the results indicate that predictive methods can be incorporated into heating cycle tests. The performance of machine learning predictions is compared with the temperature values calculated with the standards of IEC 60287-1-1 and IEC 60853-2. As a result, incorporating machine learning methods as an alternative to the relevant standard and MV cable test procedures in the future could provide significant advantages in terms of cost and time savings.