As the unique power-discharging channel for high-speed trains, the 'train-rail' mobile coupling grounding system plays a vital role in sending the traction power back to the terrestrial substations, ...through the electrical contact between the rotating wheel and the steel rail. The whole system is composed of two kinds of grounding modes: working grounding is in charge of draining the traction power into the rail track; the protective grounding provides the reference grounding potential for the vehicle-mounted equipment. Different from the fixed terrestrial grounding technique normally implemented in power system, when the relative motion occurs between the train and the rail, both the 'catenary-to-train' and 'train-to-ground' impedances vary transiently, which may lead to the current flowing through each grounding wheel varying unpredictably. Herein, the 'catenary-train-rail' mobile coupling grounding model was built based on the equivalent circuit modeling technique for evaluating the transiently-varying tendency of the grounding impedances and currents flowing through each grounding wheel. To cope with different operational conditions, the adjustable grounding impedance methodology is firstly proposed for adapting the unpredictable variation of the grounding impedance, when the relative motion occurs between the train and the rail. Ultimately, the validity of the novel grounding system applying the adjustable grounding impedance technology is verified under different operational conditions, with the consideration of both TB current and voltage.
As a unique reflux path for trains, the 'train-rail' mobile grounding system has the double functions - the traction power feedback and the transmission of train's communication signal. With the ...constant increment of operational velocity and traction power for high-speed railway, the coupling relationship existing in the 'train-rail' mobile coupling grounding system becomes complicated, when the train's wheels roll over the insulated joint. It may result in the frequent occurrence of rolling arc and the ablation of rail ends and insulated joints, which might lead to the series connection of the rail circuit or even trigger the misjudgment of operational control system. As the impedances existing in the grounding system vary transiently when the relative distance between the train and the fixed terrestrial grounding installations changes, the dynamic coupling mechanism between the transiently-varying topology of reflux path and the distribution of reflux at each grounding wheel tends to be analyzed. Depending on the built 'catenary-train-rail' dynamic coupling grounding reflux model, the correlation between the 'train-rail' mobile grounding reflux and the evolutionary process of rolling arc is explored. In order to reduce the risk of 'wheel-rail' arcing phenomenon, a novel method of automatically controlling insulated joint is proposed herein, as its feasibility is verified via the comparison with the original type of the insulated joint.
This paper presents extensive verification, sensitivity, and application studies for an online substation grounding grid condition monitoring method proposed in a companion paper. The studies are ...conducted using a realistic substation grounding grid and associated feeder configurations commonly encountered in North America. The verification studies confirm that the proposed method is able to estimate the substation grounding impedances with acceptable accuracy. Main factors impacting measurement accuracy are identified and quantified through sensitivity studies. The findings lead to a guide for the implementation and operation of the proposed method. An application case is presented to demonstrate how the gradual deterioration of substation grounding grid can be monitored using the proposed method.
This paper presents an online method to monitor the conditions of the substation grounding grid. The proposed method utilizes phase-to-ground faults occurring in downstream feeders as test current ...sources and the neutral of a specially selected feeder as the electrode for GPR (ground potential rise) measurement. The measured fault currents and GPRs are then used to estimate substation grounding grid impedance. The method, once established, can monitor the grounding grid impedance continuously and automatically, and the trend of the impedance change can be observed. As a result, the proposed method can yield much more reliable information on the substation grounding conditions. The method also overcomes many disadvantages of the offline methods. The proposed idea, algorithm, and hardware setup are described in the paper. Factors contributing to measurement errors are investigated in detail.
In the commissioning of high-voltage substations, the determination of the grounding grid resistance stands out. In real-life procedures, usually the involved measurements are carried out considering ...the guard wires of the transmission lines connected to the substation grounding grid. This issue can lead to an overestimation of the ground current, which may indicate an erroneous unsafe grid condition, in terms of touch and step voltages exposed to human beings. Considering the practical difficulty or impossibility of measuring the portion of the total fault current flowing through the ground, this paper deals with the proposition of a formulation (equation that govern the phenomenon) to estimate the ground current. The proposed simplified approach is based on an optimization method subject to adequate constraints, in which the determination of the substation grid resistance and the equivalent impedance of the guard wires allow estimating the most important parameter: the portion of the total ground-fault current that produces the touch and step voltages. The methodology was compared with results available in the literature and with field tests performed in a real substation.
Grounding systems (GSs) must be tested periodically in order to maintain the touch voltage (TV) and step voltage (SV) below a safe value in all of the zones of the installation. Measurements of the ...ground resistance and of the TV and SV are typically done by the fall-of-potential (FoP) method, locating the auxiliary current electrode at remote distance to test the effective behavior of the GS. In urban areas, it could be very complicated or impossible to install the auxiliary current electrode as required, not having area around with sufficient accessibility. At this aim, this paper describes a methodology of using multiple current electrodes at short distances, modifying the classic FoP practice, so that the measurements of TV and SV are always conservative. The adequacy of a GS is verified if the values of the TV and SV, tested inside and in the vicinity of the GS, are below the permissible limits, regardless if they are true or conservatively increased. Thus, the measured TV and SV by the suggested method, always conservative, allow verifying the adequacy of GSs, in the cases where it is impossible to locate the remote auxiliary electrode.
The scheme of configuring the Petersen coil after actively measuring the capacitive current of a network has been largely adopted, but the feedback of the compensation state is not usually used. This ...paper proposes a new method of adjusting the Petersen coil by using the information obtained after arc extinction. Using the equivalent circuit of a single-phase grounding fault in the resonant grounding system, the electrical character are analyzed. It is found that the transient process after arc extinction is the underdamped resonance between the Petersen coil and the capacitance to ground, and the resonant frequency is different from the power frequency. Therefore, the resonant angular frequency and damping ratio of the process after arc extinction can be used to calculate detuning, and the three-phase undervoltage can be used to preliminarily judge whether the system is undercompensated or overcompensated. A method of adjusting Petersen coil to achieve reasonable detuning is given. The effectiveness and accuracy of the theory and the method are supported by simulation and field fault data. The method does not need to add any special equipment or adjust the running state of the existing equipment, and it has high safety.
Faults in a grounding network pose a serious threat to power system equipment protection and worker safety. A large number of vertical electrodes are used in a grounding network and break-points in ...these due to soil induced corrosion is one of the prime reasons for faults. We present a rod insertion time domain reflectometry (TDR) technique for vertical grounding electrode break-point detection. A secondary bare rod is inserted adjacent to the grounding electrode while a fast risetime pulse is continuously applied to excite a transverse electromagnetic (TEM) wave that propagates along the rods and reflects back from the break-point. The reflected signals for different depths of the secondary rod are analyzed to identify the location and severity of the break-point. We show that fault detection is possible without prior knowledge of soil electrical parameters. To show the feasibility of the proposed technique, a full wave simulation approach is used to evaluate the wideband input impedance of the grounding and inserted rod system and then FFT is applied to obtain TDR responses. The results show the method is capable of detecting faults for a wide range of soil conductivity for a system bandwidth of 300MHz.
Image-language matching tasks have recently attracted a lot of attention in the computer vision field. These tasks include image-sentence matching, i.e., given an image query, retrieving relevant ...sentences and vice versa, and region-phrase matching or visual grounding, i.e., matching a phrase to relevant regions. This paper investigates two-branch neural networks for learning the similarity between these two data modalities. We propose two network structures that produce different output representations. The first one, referred to as an embedding network , learns an explicit shared latent embedding space with a maximum-margin ranking loss and novel neighborhood constraints. Compared to standard triplet sampling, we perform improved neighborhood sampling that takes neighborhood information into consideration while constructing mini-batches. The second network structure, referred to as a similarity network , fuses the two branches via element-wise product and is trained with regression loss to directly predict a similarity score. Extensive experiments show that our networks achieve high accuracies for phrase localization on the Flickr30K Entities dataset and for bi-directional image-sentence retrieval on Flickr30K and MSCOCO datasets.
Visual grounding (VG) aims to locate a specific target in an image based on a given language query. The discriminative information from context is important for distinguishing the target from other ...objects, particularly for the targets that have the same category as others. However, most previous methods underestimate such information. Moreover, they are usually designed for the standard scene (without any novel object), which limits their generalization to the open-vocabulary scene. In this paper, we propose a novel framework with context disentangling and prototype inheriting for robust visual grounding to handle both scenes. Specifically, the context disentangling disentangles the referent and context features, which achieves better discrimination between them. The prototype inheriting inherits the prototypes discovered from the disentangled visual features by a prototype bank to fully utilize the seen data, especially for the open-vocabulary scene. The fused features, obtained by leveraging Hadamard product on disentangled linguistic and visual features of prototypes to avoid sharp adjusting the importance between the two types of features, are then attached with a special token and feed to a vision Transformer encoder for bounding box regression. Extensive experiments are conducted on both standard and open-vocabulary scenes. The performance comparisons indicate that our method outperforms the state-of-the-art methods in both scenarios.