Durative Monitoring of Sulfur Hexafluoride Characteristic Gases under Hydrogen Interference Using a Time2Vec-Encoded CNN-Transformer-LSTM Model Based on a Heterogeneous Gas Sensor Array.

The presence of free moving metallic particles and discharges deteriorates the insulation strength of a compressed SF/sub 6/ (sulphur-hexafluoride) gas insulated switchgear (GIS) and causes serious limitation in its practical application. Free moving metallic particles and other defects inside the GIS cause partial discharge (PD) which can degrade the insulating properties of the insulant gas SF/sub 6/ to such an extent that breakdown may occur in the GIS system. So, it is necessary to detect partial discharge in the GIS at an early stage before system failure and extensive damage to the equipment and the rest of the power system. Partial discharge emits acoustic signals which can be detected by applying an acoustic emission sensor (AE sensor) outside the GIS chamber. This paper initially describes some fundamental aspects related to PD detection. Finally, the development of an acoustic PD detection system and some experimental investigations are presented.

Accurate modeling of partial discharge (PD) within gas-insulated switchgear (GIS) has become of paramount significance to improve the utilization of ultrahigh frequency (UHF) sensors within such capital assets. Providing better insight on how electromagnetic (EM) waves propagate within GIS systems provides valuable information that can be used to determine the optimal allocation of UHF sensors. Thus far, most existing models are not considering the full complexity of the GIS systems due to the many associated challenges and the high computational demand of such system modeling. In this work, a GIS system is modeled using the Maxwell solver of COMSOL multiphysics. CIGRE sensitivity verification recommendations (CSVR) are used to inject UHF signals using an internal sensor. Then, two external sensors, placed on the outer belt of dielectric spacers, are used to capture the radiated EM waves. The electric field distribution at selected frequencies is presented and discussed under two operating conditions for disconnecting switches. The modeled GIS contains many barriers, including two bends, multiple sudden changes in the outer to inner diameters ratios, six dielectric spacers, and two disconnecting switches. Time-domain simulation results are also presented to provide an insight into the attenuation properties of EM waves due to the different barriers. Finally, a comparison between simulated and measured results has been carried out to verify the modeling accuracy. The results show that GIS systems form complex structures for the EM waves, and fully understanding the wave propagation can be tedious.

When the ultra high voltage (UHV) substation used the gas insulated switchgear (GIS) system to switch lines with capacitive current, the pre-breakdown and multiple reignition will occur between switching contacts. Because of discharge characteristics of SF6 and special structures of GIS system, the variations of the electromagnetic transient process and very fast transient overvoltage (VFTO) are very fast and extremely complex. The development process and characteristics of VFTO remain to be dealt with to seek effective suppression measures. The overvoltage of closing no-load long lines on GIS system is analyzed in detail. The simulation model of the very fast transient process of closing no-load long lines on UHV GIS system is established. The VFTO of closing no-load China UHV transmission demonstration project Jindongnan-Nanyang-Jingmen 1000 kV transmission lines is analyzed using electromagnetic transient simulation software ATP-EMTP. The VFTO of closing no-load long lines on UHV GIS system is simulated and calculated, when the system has no additional closing resistors and lightning arresters, additional single stage closing resistors, additional multistage closing resistors, lightning arresters in two terminals of lines, lightning arresters in two terminals and the middle of lines, and additional multistage closing resistors and lightning arresters in two terminals and the middle of lines. The simulation results show that the additional multistage closing resistors and lightning arresters in two terminals and the middle of lines can effectively restrain the VFTO of closing no-load long lines on UHV GIS system.

Gas-insulated switchgear (GIS) is a common electrical equipment, which uses sulfur hexafluoride (SF6) as insulating medium instead of traditional air. It has good reliability and flexibility. However, GIS may have internal defects and partial discharge (PD) is then induced. PD will cause great harm to GIS and power system. Therefore, it is of great importance to study the intrinsic characteristics and detection of PD for online monitoring. In this chapter, typical internal defects of GIS and the PD characteristics are discussed. Several detection methods are also presented in this chapter including electromagnetic method, chemical method, and optical method.

Gas-insulated switchgear (GIS) is widely used across multiple electric stages and different power grid levels. However, the threat from several inevitable faults in the GIS system surrounds us for the safety of electricity use. In order to improve the evaluation ability of GIS system safety, we propose an efficient strategy by using machine learning to conduct SF6 decomposed components analysis (DCA) for further diagnosing discharge fault types in GIS. Note that the empirical probability function of different faults fitted by the Arrhenius chemical reaction model has been investigated into the robust feature engineering for machine learning based GIS diagnosing model. Six machine learning algorithms were used to establish models for the severity of discharge fault and main insulation defects, where identification algorithms were trained by learning the collection dataset composing the concentration of the different gas types (SO2, SOF2, SO2F2, CF4, and CO2, etc.) in the system and their ratios. Notably, multiple discharge fault types coexisting in GIS can be effectively identified based on a probability model. This work would provide a great insight into the development of evaluation and optimization on solving discharge fault in GIS.

This paper aims to compare the performance of the classical machine learning (CML) model and the deep learning (DL) model, and to assess the effectiveness of utilizing fusion radiomics from both CML and DL in distinguishing encephalitis from glioma in atypical cases. We analysed the axial FLAIR images of preoperative MRI in 116 patients pathologically confirmed as gliomas and clinically diagnosed with encephalitis. The 3 CML models (logistic regression (LR), support vector machine (SVM) and multi-layer perceptron (MLP)), 3 DL models (DenseNet 121, ResNet 50 and ResNet 18) and a deep learning radiomic (DLR) model were established, respectively. The area under the receiver operating curve (AUC) and sensitivity, specificity, accuracy, negative predictive value (NPV) and positive predictive value (PPV) were calculated for the training and validation sets. In addition, a deep learning radiomic nomogram (DLRN) and a web calculator were designed as a tool to aid clinical decision-making. The best DL model (ResNet50) consistently outperformed the best CML model (LR). The DLR model had the best predictive performance, with AUC, sensitivity, specificity, accuracy, NPV and PPV of 0.879, 0.929, 0.800, 0.875, 0.867 and 0.889 in the validation sets, respectively. Calibration curve of DLR model shows good agreement between prediction and observation, and the decision curve analysis (DCA) indicated that the DLR model had higher overall net benefit than the other two models (ResNet50 and LR). Meanwhile, the DLRN and web calculator can provide dynamic assessments. Machine learning (ML) models have the potential to non-invasively differentiate between encephalitis and glioma in atypical cases. Furthermore, combining DL and CML techniques could enhance the performance of the ML models.

PD (partial discharge) is one of the most effective parameters to evaluate the insulation condition of power equipment. When PD happens, the optical signals will be emitted from the discharge sources. PD can be detected by measuring the optical signals. The PD optical method has electromagnetic immunity and especially suitable for the GIS (gas insulated switchgears) system which has a totally enclosure structure. At present, the related studies about development of optical sensors, optical detection of typical defects, and analysis of relationship between PD electrical and optical signals have been conducted in laboratory. However, the researches on propagation characteristics of PD optical signals are lack. In order to study the effects of propagation distance, GIS structure, receiving angle, and GIS component on the characteristics of optical signals induced by PD in GIS, a simulation model is built based on the size and structure of a physical GIS in this work. The photon amount and luminous flux detected by optical sensor under the condition of different distance, different GIS structure types, different positions of discharge source, and existing insulator with air hole are investigated. The results show that the photon amount and change rate of photon number will decrease with the increasing in distance between optical sensor and discharge source. the optical signals are the strongest when the PD source directly faces the detection sensor. The attenuation of optical signals is mainly caused by the hide of the GIS component when the detection angles change. With the increasing of voltage levels of GIS, the attenuation of optical signals is more obvious. The maximum change rate of photon number detected by the sensor in the L type GIS are about 61% of that in the line type GIS, which are 45% in the T type GIS. The related results can provide support for the installation of optical sensor and detection region of optical method.

Gas-insulated switchgear (GIS) has been widely used in power grid systems. The busbar is a crucial element within a GIS system, responsible for facilitating electrical connections between different GIS components. Any defects during manufacturing or improper installation procedures can lead to busbar misalignment, resulting in abnormal vibrations within the GIS. This study focuses on analyzing the mechanical vibration characteristics of GIS bus misalignment using a simulation model based on the finite element method for a 252kV GIS. When the busbar is misaligned and the conductor experiences an unbalanced force, it undergoes Coulomb forces at double the voltage frequency. Under normal or misaligned conditions, the conductor exhibits forced vibrations at the power frequency of 100Hz. However, when misalignment occurs, there is a coupling effect between electrodynamics and vibrations, causing the shell to display integer frequency vibration components.

Analysis of the transmission and leakage characteristics of partial discharge (PD) signal in Gas Insulated Switchgear (GIS) is of great importance for PD detection. Traditional connector of two GIS tanks is a disc insulator, so the leakage of PD signal is easily detected. Recently, to improve the performance of GIS, the disc insulator is often sealed by a metal belt and an aperture is made in the metal belt for PD detection. The leakage of PD is weak and the PD detection is delicate. This paper is devoted to the theoretical and experimental study of the transmission and leakage characteristics of the PD signals in such GIS. The electric field in the GIS and the characteristics of the leaked signals are simulated with software Ansoft HFSS. The PD source is assimilated by two simple models: “line feed” and “waveguide feed”. The simulation results show that the frequency band and amplitude of the leaked signals depend on the size of the aperture and the distance of the detector from the GIS. A novel ultra wideband planar antenna is developed and a full experiment setup is built to validate the simulation and to test the characteristics of the leaked PD signal. The transmission coefficient S21 has been measured with a vector network analyzer. The experimental results correspond better to the simulation data with “line feed” than “waveguide feed”. But the prediction of the frequency characteristics by the dipole model is very different from the experiment results in the high frequency domain. Furthermore, we show by the experiments that the amplitude and frequency profile of the transmission coefficient S21 is sensible to the position of the PD source in the GIS.

Gas-insulated switchgears have become essential parts of the electric power systems utilities because of their reliability, compactness, and effectiveness. Different techniques have been developed to protect these devices from partial discharge. The Ultra-High Frequency techniques, which detects electromagnetic waves emitted during the event of partial discharge, showed great performance on the localization of such phenomenon. In this paper, three different machine learning classification models have been built to localize partial discharge in a modeled gas-insulated switchgear using COMSOL multiphysics. Results for five different sensors are obtained and processed to establish concrete attributes for machine learning purposes. Various learning algorithms have been evaluated and compared, such as artificial neural networks, support vector machine (SVM), and a subspace discriminant ensemble model yielding a range of accuracies. An accuracy of 94.8% was obtained through the subspace discriminant ensemble model.

The recent activity on electrical insulation and its ageing aspect in gas-insulated switchgear (GIS) as well as future trends are described in this paper. Present status of research shows that although there are a lot of reports on computer based partial discharge measurement, there are few on the intrinsic phenomena of fundamental ageing characteristics. These data will be indispensable for extending new concepts of GIS insulation. As an example, it is pointed out that research on electric stress enhancement in solid materials is important, in particular, in relation to life-estimation and life-control techniques as well as with making more compact GIS. For this purpose, the introduction of consistent concepts, from the development/design stages to the testing/operation stages, of GIS is needed. This concept would include sensor integrated GIS, failure-prediction techniques and stress-controlled life-estimation systems. These proposals are described in this paper from the manufacturer's viewpoint. >

To ensure the safe operation of the gas insulated switchgear (GIS), more and more GIS will be tested under impulse voltage on-site. Oscillating impulse voltage is very suitable for the test due to the high efficiency. By detecting the partial discharge (PD) during the process of on-site impulse withstand voltage test, the insulation state of GIS can be overall known. Surface defect in GIS is typical and harmful, so it is very significant to research the PD characteristics of surface defect in GIS under oscillating impulse voltage. In this paper, a surface defect model was designed. The oscillating impulse voltage generator in this paper could generate oscillating lightning impulse voltage (OLI) and oscillating switching impulse voltage (OSI). The pulse current and UHF method were used to detect the PD signals. The PD sequences, PD phase distribution, PD amplitude and PD density were analyzed. The results show that the negative PDs happen at the rising edge of voltage, and the positive PDs happen at the rising edge of voltage. The PD magnitudes and the PD numbers increase linearly with the increase of voltage. Moreover, the PCA (phase circle amplitude) spectrum and OPRPD (oscillating phase resolved partial discharge) spectrum were introduced to describe the PD characteristics of surface defect. The results in this paper are very helpful for the impulse withstand voltage test in field.

Gas Insulated Switchgear (GIS) has high reliability. However, partial discharge (PD) may occur. PD may lead the insulation breakdown. It is needed to detect PD before breakdown occurs. PD emits electromagnetic (EM) wave. UHF detection method is the most effective method to detect PD induced EM wave in GIS. The characteristics of EM wave propagating through GIS are needed to optimize sensor position for partial discharge detection. This paper discusses the propagation of PD induced EM wave in single phase and three phase 150 kV GIS. The difference between single phase and three phase GIS cause different electromagnetic wave propagation characteristic in GIS. By knowing the difference, UHF sensor settings can be determine. The simulation results show that single phase GIS has higher transmission rate, voltage intensity, and dominant frequency spectra than three phase GIS.

SummarySpatiotemporal solar radiation forecasting is extremely challenging due to its dependence on metrological and environmental factors. Chaotic time‐varying and non‐linearity make the forecasting model more complex. To cater this crucial issue, the paper provides a comprehensive investigation of the deep learning framework for the prediction of the two components of solar irradiation, that is, Diffuse Horizontal Irradiance (DHI) and Direct Normal Irradiance (DNI). Through exploratory data analysis the three recent most prominent deep learning (DL) architecture have been developed and compared with the other classical machine learning (ML) models in terms of the statistical performance accuracy. In our study, DL architecture includes convolutional neural network (CNN) and recurrent neural network (RNN) whereas classical ML models include Random Forest (RF), Support Vector Regression (SVR), Multilayer Perceptron (MLP), Extreme Gradient Boosting (XGB), and K‐Nearest Neighbor (KNN). Additionally, three optimization techniques Grid Search (GS), Random Search (RS), and Bayesian Optimization (BO) have been incorporated for tuning the hyper parameters of the classical ML models to obtain the best results. Based on the rigorous comparative analysis it was found that the CNN model has outperformed all classical machine learning and DL models having lowest mean squared error and highest R‐Squared value with least computational time.

It has been revealed that imperfection such as microscopic protrusions on the interior of the Gas Insulated Switchgear systems (GIS) metal work, foreign particles, components that float in potential and deterioration of solid insulation reduces the insulation level of switchgear1.