Machine learning methods are effective for the diagnosis of power transformer faults. However, influenced by uncertainty and noise in data, machine-learning-based diagnostic methods are still in the initial phase of certain assets in power systems. To mitigate this gap, a deep noisy filtering diagnostic model is proposed for accurate and rapid evaluations of power transformer faults using noisy vibration signals. A balanced isolation forest method is employed to detect abnormal data from the original vibration signals. Two deep noisy filter networks suppress the level of noise, based on which contrastive learning obtains the transformer fault states. Datasets collected from a 10-kV real power transformer validate the proposed model. The results demonstrate that the proposed method acquires a higher fault diagnostic accuracy with respect to the compared algorithms, showing the superiority and efficacy of the proposed model.

According to published papers on Rogowski coils, the sensitivity of Rogowski coils can be increased by using high-density turns and multilayer turns. Low-density turns can increase the bandwidth of the coil but can also reduce the anti-interference ability of the coil. The shielding layer can also increase the anti-interference ability of Rogowski coils. However, for Rogowski coils with high-density turns, using a shielding layer or multilayer turns will reduce their measurement bandwidth. To increase the application flexibility of a Rogowski coil, this article proposes a method capable of increasing the bandwidth without sacrificing size, sensitivity, anti-interference, and measurement accuracy. In this method, coils are segmented without reducing the number of turns, and the induced voltage of the segmented coils is output through the voltage followers and joined together by an adder. Seventeen Rogowski coils are wound by an automatic coil winding machine. These 17 coils consist of single-layer coils, double-layer coils, coils without a shielding layer, and coils with a shielding layer. The method is verified for use with Rogowski coils when measuring the sinusoidal current and the turn-on/off pulse current of an high voltage-insulated gate bipolar transistor (HV-IGBT) collector. The measurement errors of this method and the error compensation method are also explored both theoretically and experimentally.

There are Rogowski coils of various shapes in the on-site measurement, and it is difficult to calculate the electrical quantities of Rogowski coils of curved skeleton and circular cross-section by simulation software. This paper proposes a theoretical derivation to calculate the mutual inductance between the conductors of any shape and Rogowski coils with skeletons of any shape. Based on the derivation, the influence of four skeleton shapes of Rogowski coils and four shapes of the primary conductors on the mutual inductance of Rogowski coils are studied by the comparison between the ideal cases and some non-ideal ones. The gap and gap compensation of the openable Rogowski coils are also considered. Experiments verify the numerical results according to the derivation. It is shown that to reduce the errors of the measurement the circular skeleton deformation should be avoided, the coil’s skeleton should be with curved angle, the primary conductor should be as straight as possible and should go through the center of the skeletons vertically. Furthermore, for the Rogowski coils of the rectangular skeleton, we propose a new skeleton structure to reduce the deviation influence of the primary conductors.

The Rogowski coils are usually used to measure high frequency currents due to their air core. The higher the turn number, the higher the mutual inductance between Rogowski coil and the primary conductor. Performance of Rogowski coil with a high turn number is very difficult to figure out by simulation software. This paper presents an improved lumped parameter model of Rogowski coils. According to the model, the influence of parasitic turn-turn capacitance, turns self-inductance, turn-turn inductance, turn-shielding capacitance, turns resistor, sample resistor and others on the linearity of Rogowski coil with a high turns number at several different frequency ranges are explored. The model is applicable to both double and single layer coils. Three high-turn-number coils of double and single layer are studied and compared in this paper. The improved lumped parameter model is verified by experiments. Five main influence factors on the linearity of the Rogowski coil at a certain frequency range are studied both theoretically and experimentally. The five influence factors include sampling resistors, turns density, shielding layer, positions of the primary conductor, and turns nonuniformity. The numerical and experimental results analyze the influence of the factors on the linearity of the output voltage of the coils at different frequencies quantitatively and qualitatively. The principle of the improved model can be used for multi-layer coils. The model can be a useful guidance in designing coils with suitable structure for special applications at different frequencies.

Printed circuit board (PCB)-based headers of Rogowski current sensors (HRCSs) have garnered growing interest from researchers in recent years, and they have been widely applied in many electrical scenarios due to their advanced features, such as nondeformable volume, flexible shape and size, and low cost. The determination of PCB HRCS parameters mainly depends on a time-consuming finite-element simulation, but these simulations cannot be used to calculate PCB HRCSs with a complex irregular structure or high-density thin winding wire. This article presents a method to calculate a PCB HRCS to address complex irregular structures and high-density thin winding wire. Furthermore, because it does not involve the time-consuming meshing process, this method is relatively fast. Taking two HRCSs with a given complex structure and thin winding wire as examples, we used the proposed method to determine its sensitivity, anti-interference characteristics, and the influence of the obtained internal parasitic capacitance and inductance parameters on the measured output voltage. Moreover, the optimal structural parameters of these PCB HRCSs are discussed when the installation space and the cross section of the measured conductor are given. The results of the high-density PCB HRCS experiments verify the proposed method.