FisPro based interpretable fuzzy inference system for dual circuit extra high voltage transmission line fault classification and fault distance estimation
نویسندگان
1 Department of Electrical and Electronics Engineering, Institute of Aeronautical Engineering, Hyderabad, India.
2 Department of Electronics and Communication Engineering, KS Institute of Technology, Bengaluru, India.
3 Department of Space Engineering, Ajeenkya DY Patil University, Pune, India.
4 Department of Aerospace Engineering, Toronto Metropolitan University, Toronto, Canada.
5 Department of CSE (Data Science), Institute of Aeronautical Engineering, Hyderabad, India.
6 Department of Electrical and Electronics Engineering, Vasavi College of Engineering, Hyderabad, India.
7 Department of Electrical, Electronics and Communication Engineering, GITAM, Bengaluru, India.
8 Department of Information Technology, University of the Cumberlands, Williamsburg, USA.
doi
10.22105/jfea.2025.452949.1447چکیده
Accurate fault classification and precise fault distance estimation play a critical role in reliable, stable and optimal operation of electrical power systems. Especially Dual Circuit Extra High Voltage Transmission Line (DCEHVTL) fault diagnosis is a challenging task using conventional algorithms. Early fault recognition is crucial for DCEHVTL performance. Due to its effectiveness in classification and forecasting, the interpretable Mamdani Fuzzy Inference System (MFIS) is more suitable. Haar wavelet is efficient in any signal behavior evaluation. Therefore, this paper proposed a robust and intelligent Fuzzy Inference System Professional (FisPro) based interpretable MFIS with Haar wavelets transform for fault classification and fault distance estimation in DCEHVTL using single-side data. The dual circuit three-phase currents of two cycles are recorded from DCEHVTL. Haar wavelet transforms are utilized to estimate the behaviour and characteristics of the current signals in terms of high-frequency components. The captured fault currents from the DCEHVTL single side are used as inputs for the hierarchical MFIS framework. Here, the fault classification is followed by the faulty distance estimation in MFIS. The test results support the MFIS consistency under extensive changes in fault factors. Moreover, the extensive performance comparison study with the state of art fault diagnosis methods reiterates the FisPro based MFIS effectiveness in successful fault diagnosis. The obtained results indicate that MFIS has a fast processing time (in 1ms), high accuracy (above 99.8%), less fault location error (Within 0.0024%) and is useful for studying the system stability in the electricity field. The MFIS is proven to be successful for the fault diagnosis without any communication channel between the source and receiving end.
