Investigation of partial discharges in medium voltage cables with advanced signal processing methods

In the solid and liquid insulators used in medium voltage and high voltage, physical damages occur in a long process due to partial discharges and reach destructive levels. It is stated in the standards that all faults at high and medium voltages are caused by insulation problems due to partial discharges of 80%. These damages can cause the high voltage system to become damaged and fail to work, as well as lead to fatal accidents. Therefore, the classification of these ejaculation events, determining the possible malfunction type and location in advance and taking the necessary precautions accordingly are of great importance in terms of both uninterrupted operation of the system and preventing possible unwanted work accidents. Within the scope of the doctoral thesis study, the practices of determining the types and locations of partial discharges in the medium voltage cables, which are among the most important components of the medium voltage distribution network and which are of great importance in terms of cost for the system, have been carried out with innovative methods. In this context, as a first step, artificial damages have been created on cross-linked polyethylene insulated (XLPE) cables in order to observe different discharge events. In the second stage, many different experiments were carried out by applying high voltages in various amplitudes to the cable samples produced in the first stage and damaged at certain points in their insulation. During the experiments, partial discharge data were measured and recorded on cables. Experiments were carried out in the high voltage laboratory of Afyon Kocatepe University Electrical Engineering. Statistical calculations, Fourier transforms and statistical features of these transformations, phase resolution partial discharge (PRPD) analyzes were applied on the obtained data. Current and accepted methods in the field such as support vector machines, neural networks, nearest neighborhoods, wavelet transforms have been applied to partial discharge data to classify the type and location of the fault and successful results have been obtained. In the last step of the study, for the first time, within the scope of this thesis, the performance of Mycielski-Markov hybrid approach in the classification of XLPE cables with different defects and modeling of the discharge occurrence in XLPE cables has been tried and successful results have been obtained.