Accurate remote estimation of smart meter errors is crucial for ensuring the reliability of metering data, thereby enabling fair electricity trading and supporting advanced applications such as load forecasting. Existing remote smart meter error estimation methods face key issues, including loss calculation, measurement synchronization, and solution stability, which lead to low estimation reliability. To address these issues, an improved remote error estimation method for smart meters is proposed in this article. A functionally equivalent calculation of line losses is achieved by describing the two-stage power profile within each metering period. To mitigate estimation deviations caused by measurements asynchrony, the metering period is expanded, and the optimal expansion length is derived from the analysis of energy measurement imbalances within the metering periods. Furthermore, the precision-driven adaptive regulations strategy for the coefficient matrix is employed to suppress model ill-conditioning, enabling accurate and stable remote estimation of metering errors. The effectiveness and superiority of the proposed method are validated through verification experiments.

With the development of intelligent and interconnected industrial equipment, the demand for equipment diagnosis and prediction promotes the optimization of smart factory maintenance strategies. As the core transmission component of rotating machinery, the bearing directly affects the production efficiency and operation safety, so that the efficient performance evaluation of the bearing is very important for the predictive maintenance and intelligent management of the equipment. The traditional signal acquisition method is affected by the challenging industrial field conditions. In recent years, noncontact methods have been gradually applied to industrial measurement. However, causes such as difficulties in calibration and image recognition, weak vibration, and so on, visual condition monitoring of complex systems, such as bearing faults has not been widely and effectively explored in existing research. Therefore, this article incorporates visual displacement tracking into the field of bearing fault detection, and uses Lucas-Kanade (LK) optical flow to extract subpixel vibration from equipment operation video, the optimal filter size is searched by Fibonacci, and the pulse of each fault period is enhanced by multipoint maximum kurtosis-harmonic to noise ratio deconvolution (MMKHD), which directly expresses the accurate fault characteristic frequency through the spectrum. The signal extraction and analysis of bearings at different fault positions under various rotational speeds are carried out, respectively, and the contact acceleration sensor is used to collect the vibration signal synchronously for comparison and verification. Furthermore, the experiment is carried out by using the main light source with decreasing power, and the robustness of the vibration extraction method under different illumination environments is verified.

Abstract With the advancement of the electric locomotive industry, pantograph-catenary (P-C) contact conditions face stringent requirements during high-speed operation. Real-time detection of arcs between the P-C structure and timely countermeasures are essential for stable train operation. This paper proposes a method for detecting P-C arcing based on the asymmetric current component (ACC) and discusses the relationship between the asymmetry degree (AD) and arcing during P-C disconnection. A composite model encompassing the train traction circuit, the local electric field of the P-C air gap, and arc thermodynamics is established to replicate the waveform and dynamic variations of electrical parameters during arcing. Subsequently, a random forest (RF) classifier is developed based on the signal segments generated from this model. The detection method avoids strong interference from power-frequency, harmonic, and high-frequency backgrounds, and incorporates multiple factors not considered in conventional arc models. Experimental result indicates the precision of the method ≥ 96% (conservatively), demonstrating its effectiveness and feasibility.

To address critical technical bottlenecks in high-impedance fault (HIF) detection—namely, delayed detection and insufficient localization accuracy—in distribution networks, this paper proposes a novel fault-location methodology based on the Differential Power Spectrum Singularity (DPSS). First, a fault-impedance identification technique leveraging low-frequency active-power differences is introduced. Second, a physics–electricity coupled model of HIF is formulated to unveil the mapping mechanism between the power-spectrum characteristics at the fault point and its spatial location, along with an adaptive upper-frequency sweep criterion under dynamic line-scale conditions. Subsequently, an offline validation platform for a 10 kV single-phase cable is developed by integrating PSCAD-based electromagnetic transient simulations with Matlab signal-processing algorithms to systematically evaluate the robustness of the DPSS indicator under diverse operating scenarios. Experimental results demonstrate that the proposed method consistently suppresses HIF localization error within 0.2

Due to the complexity and randomness of a series fault arc in a low-voltage distribution system, it would be a great challenge to accurately describe the full-wave arcing process through a concise mathematical form. To address this issue, a circuit characteristic arc model of based on the asymmetric arcing (AA-CCAM) of low-voltage series arc fault is proposed. Firstly, the asymmetric arcing process is exactly depicted using field characteristic description, and then the AA-CCAM is constructed based on reasonable assumptions and field-path evolution. Afterwards, a time-domain difference method is adopted to implement the partitioned iterative computation of the model. More specially, the problem of solving model parameters is transformed into a single objective optimization problem, which applies the competitive particle swarm algorithm. Finally, a low-voltage series arc fault experimental platform is established for model validation. Based on the training set, reference parameter values are determined through stable distribution fitting. Test results demonstrate that the simulated fault waveforms achieve an overall fitting accuracy of 97.26

As one of the important electrical parameters to characterize the cable state, the accurate measurement of voltage is especially vital for the reliable operation of cables. To address the problems of complicated calculation of electric field in the vicinity of three-phase cables, too much reliance on historical experience, and the difficulty of three-phase voltage inverse calculation, this paper proposes a non-contact measurement method of three-phase cable voltage based on electric field inverse calculation. An optimal analytical calculation model of three-phase cable electric field distribution is established by combining the actual parameters of three-phase cable. An array of electric field sensors is made to obtain the electric field information of the cable neighborhood. A genetic algorithm is used to solve the electric field inverse calculation problem, which guarantees the global optimality of the solution results. The effectiveness of the proposed method is tested on a three-phase cable voltage measurement test platform, and the experimental results show that the three-phase measurement accuracy of the method proposed in this paper is less than 2% regardless of the three-phase voltage is balanced or unbalanced.

Current measurements of multiconductor systems are essential for smart grids, but conventional intrusive measurement approaches require power outages before deployment and maintenance and thus are expensive and may reduce system stability. Though many contactless current measurement approaches have been proposed in recent years, the challenge of easing the impact of conductor positions and decoupling current information remains. This paper proposes a novel approach that combines a cost-effective annular magnetic field (MF) sensor array with an inverse calculation technique for precise, contactless current measurements in multiconductor systems. Specifically, the MF sensor array captures the MF distribution surrounding the multiconductor systems for characterizing the coupled current information. Then, the hardware-oriented measurement task is reformulated into a computational optimization problem, where the currents and conductor positions are mathematically related to the MF distribution for decoupling. Meanwhile, an effective algorithm is tailored, where knowledge about the current and MF distribution is used to generate promising initial solutions for efficiency and effectiveness improvement. Experimental results demonstrate that the proposed approach achieves high accuracy in current measurement considering the impact of conductor positions, with error rates maintained below 1% and 2% in balanced and unbalanced cases, respectively. Additionally, an abnormal MF sensing data correction method is developed to further ensure measurement accuracy, showing resilience to sensor anomalies and maintaining a relative measurement error below 2% after correction.

The modal analysis of the label-free visual analysis structure proposed in this paper is a new measurement method that slices the continuous high-speed camera sampling video, and compares the pixels of the region of interest one by one with the digital image correlation algorithm to obtain the vibration response information of the measurement points. While ensuring the accuracy and robustness of the measurement and analysis results, the label-free measurement improves the data processing efficiency. Secondly, the digital image correlation algorithm combined with the inverse Gaussian-Newton method was used to obtain the full-field dynamic displacement. Finally, the random subspace method based on covariance is used to identify the modal frequency of the structure. To verify the effectiveness of the proposed method, the vibration response of the Charpy structure is measured by using a high-speed camera and an accelerometer sensor, and the results show that the relative error between the proposed visual measurement method and the eddy current sensor for the first six modal frequencies of the Charpy structure is not more than 1.17

Due to the increase in power electronified and unknown scenarios within low-voltage distribution systems, the detection based on current features often confuses series arc faults (SAFs) with complex loads. To address this issue, an SAF detection method is proposed based on the asymmetry of the arc current signal. First, through comprehensive analyses on a priori knowledge about the asymmetry of the arc current signal, a signal preprocessing via half-cycle decomposition is presented to further emphasize the asymmetry. Then, asymmetry feature quantification and subsequent feature selection would be employed to determine the feature set. Afterwards, a classifier based on random forest algorithm is established to draw a conclusion as either "Normal" or "SAF". Finally, an experimental platform with a non-contact current sensor is constructed, and experiments are made to verify the proposed method’s validity facing power electronic loads and unknown loads.

Single-phase grounding faults (SPGFs) in distribution networks are often accompanied by electric arcs, making detection challenging due to irregular zero-sequence voltage and current. This article proposes a new method for detecting arc grounding faults (AGFs) in distribution networks using Toeplitz inverse covariance-based clustering (TICC) and dynamic time warping (DTW) to recognize chaotic direct current (CDC) waveform features. Initially, the article examines the asymmetry in current during AGF occurrences and concludes that asymmetric components persist throughout the fault duration. The concept of CDC is then proposed and obtained through specially designed sensors. Subsequently, the TICC algorithm is utilized to segment real-time CDC waveforms, extract the abnormal CDC waveform, and identify faulty lines. Finally, real-time AGF diagnosis of abnormal CDC waveforms is conducted using the DTW algorithm. By collecting data from a real substation in Jiangsu, China, and taking experiment verification in China ultrahigh voltage (UHV) Test Base, the results demonstrate that the classification accuracies for the training and test sets reached 96% and 100%, respectively, with the errors in diagnosing AGFs not exceeding 0.3 s. As a result, it enhances the safety and stability of power grid operations.

The distribution network is highly susceptible to environmental impacts, often leading to short circuits or ground faults. Traditional fault line selection devices rely on a rigid threshold of neutral point zero-sequence voltage, which can mistakenly classify "virtual grounding"(mainly harmonic and power frequency resonance) as faults, leading to false outages, equipment damage, economic loss, and reduced system reliability. To address this, we propose a method for identifying virtual grounding in the distribution network based on harmonic distortion energy and a voltage-current phase grid diagram (VCPD). Using the peak distribution of the zero- sequence voltage harmonic spectrum, we developed a harmonic distortion energy algorithm to detect virtual grounding dominated by harmonic resonance. Considering that power frequency resonance has low harmonic content but distinct spatiotemporal characteristics in zero-sequence current and three-phase voltages compared to single-phase ground faults (SPGF), we introduced a CNN-based VCPD algorithm to identify virtual grounding dominated by power frequency resonance. This approach distinguishes between virtual grounding and SPGF. Finally, simulations and actual data confirm the method's reliability and accuracy.

Reliable pantograph-catenary (P-C) contact is fundamental to locomotive operational stability. P-C arcing (P-C arc) discharges are induced by deteriorated contact conditions; thus, real-time arc detection becomes imperative for proactive maintenance interventions. Conventional methodologies are designed based on optical/acoustic signals induced by the P-C structure or normal frequency bands of electrical parameters accompanying arcs. However, some limitations exist: environmental interference would largely affect the out-carriage sensing modalities. For approaches based on conventional electrical parameters, the accuracy is compromised, attributable to two principal factors: the selection of electrical parameters is constrained, which might be inadequate to delineate arc features. The frequency band utilized is relatively low, failing to effectively pinpoint arc features. This article presents a novel transient analysis framework, targeting electrical parameters during arc current-zero periods. Multiple electrical parameters are analyzed to identify the optimal detection frequency bands, exhibiting maximal discriminative power between arcing/normal contact states. The features are calculated by windowed fast Fourier transformation (FFT) according to the analysis and then processed and temporally synchronized to form datasets. Finally, a random forest (RF) classifier constructed based on the datasets enables real-time detection. Experimental validation confirms superior detection efficacy.

The electromagnetic compatibility of electronic transformers faces new challenges due to the strong transient electromagnetic environment caused by the primary and secondary fusion structures. This paper thoroughly examines the scenarios where the fusion structure is applied to electronic transformers. Five disturbance sources are chosen: surges, damping oscillations, fast transient pulses, ground potential rise, and operation overvoltage. An evaluation index system is then constructed, encompassing multiple coupling ports, various disturbance types, and multiple time-frequency characteristics. The Attribute Hierarchical Model (AHM), in combination with the entropy-weight method, is utilized to thoroughly weigh the indices, while the fuzzy comprehensive evaluation method is applied to determine the degrees of index membership. This approach ultimately assesses and classifies risks related to electromagnetic compatibility failures in electronic instrument transformers. Additionally, the method proposed in this paper is implemented on two instrument transformer samples. The results indicate that the suggested hierarchy attribute model can proficiently assess the electromagnetic compatibility of electronic instrument transformers and offer assistance for their monitoring and maintenance of the risk of failures.

As the testing device for smart electricity meters (SEMs), a calibration device (CD) should be ensured to have qualified metering performance. The conventional inspection could only be conducted periodically, usually once a quarter, making the deterioration between two inspections neglected. Thus, studies have been made on the online inspection. Most of them are developed only for functional failure but not performance deterioration, while several theoretically feasible methods rely on too large data volume and variety to be practically applied. In this article, an accuracy evaluation method is proposed based on the idea that the difference in two CDs' metering performance could be reflected through the correlation of time-series data formed by their testing results. Two-stage evaluation strategies are proposed to evaluate the CD's metering performance under multiple abnormal devices. Also, the advantage in the data volume and variety is demonstrated, and the effectiveness and superiority of the proposed method are further verified with an application.

With the rising popularity of computationally expensive multiobjective optimization problems (EMOPs) in real-world applications, many surrogate-assisted evolutionary algorithms (SAEAs) have been proposed in the recent decade. Nevertheless, high-dimensional EMOPs remain challenging for existing SAEAs attributed to their requirement in massive fitness evaluations and complex models. We propose an SAEA with a supervised reconstruction strategy, namely SR-SAEA, for solving high-dimensional EMOPs. In SR-SAEA, we first select several well-converged reference solutions to form a set of reference vectors in the decision space. Then each candidate solution is projected onto these reference vectors, reflecting the closeness between the candidate solution and those reference solutions. Each candidate solution is then projected onto these reference vectors, generating a projection vector that reflects its proximity to the reference solutions. This allows the optimization of the high-dimensional decision vector to be approximated by optimizing the low-dimensional projection vector. Subsequently, a supervised autoencoder is employed to reconstruct the optimized low-dimensional projection vector back to the original decision space. Notably, the latency vector of the autoencoder is replaced with the projection vector for supervised reconstruction. An ablation study confirms the effectiveness of the proposed supervised reconstruction strategy. The superiority of SR-SAEA, compared with six state-of-the-art SAEAs, is validated on benchmark problems with up to 200 decision variables.

As voltage measuring devices are widely used in the high voltage power system with a primary-secondary-fusion structure, electronic voltage transformers are directly connected to the primary conductor and would face much more serious electromagnetic environments than general secondary equipment. As a result, they would still experience failures even when general protection measures are adopted to suppress overvoltages' amplitudes. To address this issue, an innovative method is proposed after analyses are conducted on the resonance between a transformer and an overvoltage. Within this method, an air-cored coil is installed in a transformer to artificially shift its natural frequency and to dodge the high-frequency dominant component of a transient overvoltage, avoiding the aforementioned resonance and suppressing the secondary overvoltage. In addition, simulations and laboratory tests are conducted to prove this method's validity.

During a train running at speed, the pantograph-catenary structure atop the train would be influenced by many factors, resulting in disconnections. A timely detection on arcs across this structure helps to guide works on real-time operating and later maintenance, to eliminate latent factors of disconnection. A theory on the generation of arcs is researched, simulated, and verified with practical data. Two features of entry current, in frequency bands at kHz degree and at harmonic-Hz degree, are picked for the detection. Based on the discrepancy on these bands while arcing and normally running, a model of support vector machine (SVM) is constructed. Existing signals on train are processed and rearranged to be datasets for training SVM. After applying it to a practical railway, the results illustrate: the accuracy on arc detections is up to 99.96

When transfer learning is applied to fault diagnosis, firstly, it will bring serious negative transfer problem due to the distribution difference between source and target domain data, secondly, the excessive reliance on source data will bring certain privacy and security problems. A clustering guided source-free domain transfer diagnosis method is proposed in this paper. First, to address the problem that the samples in the source and target domains of mechanical devices do not conform to a normal distribution, a Gaussian mixture model (GMM) is introduced to divide the data and generate sample weight scores as a data-level judgment. Secondly, the predictive probability of the model is regarded as the model score as the model level judgment, and a hybrid weighting strategy is used to combine the sample weights with the model weights. The overall effective fusion of source domain classification prediction and sample Gaussian mixture model distribution prediction. In addition, classification target diversity loss was introduced to improve model accuracy. Finally, the effectiveness of the proposed method is verified using a cross-device migration diagnostic test. The experimental results show that the proposed method can fully exploit the fault feature information and improve the diagnostic accuracy under passive and unsupervised cross-domain conditions, which has higher diagnostic accuracy than the current popular methods.

Due to the great diversity of loads in low-voltage systems, the detection based on characteristic parameters of the current often confuses series arc faults (SAFs) with complex loads. To address this issue, an SAF detection method is proposed based on the inevitable dc component. First, comprehensive analyses, as well as observations, are made on the electrode-arcing-current asymmetry (EACA) to demonstrate that an inevitable dc component is inevitably induced during an SAF. Then, a dc-related dominated index and several asymmetry-related supplemental indices are gathered to form a feature set with strong generality. Afterward, a specific scheme is developed based on the uni-period state evaluation and the multiperiod fault judgment to reduce the false detection, where the eXtreme gradient boosting (XGBoost) algorithm is employed as a classifier. After that, experiments are made to verify the proposed method's validity. Finally, with monitored samples used to construct an ultrageneral testing set, simulations are conducted to prove its superiority in generality.

Non-contact three-phase instantaneous voltage measurement is an emerging and challenging topic in modern smart grids. Existing measurement methods can hardly obtain accurate or instantaneous results attributed to the coupled three-phase information. Even though some advanced optimization algorithms have been developed, their performance should be further promoted. In this study, we first transform the measurement task into a single-objective optimization problem to address the deficiencies of existing methods. Then six problems with scalable numbers of decision variables and complexity of objectives are gathered to form a test suite for global optimization. Moreover, a knowledge-based cooperative co-evolutionary algorithm is proposed for solving the formulated problem. The main idea is to incorporate the physical properties and rules of the system into the design of an effective and efficient algorithm. By proposing the knowledge-based grouping and local search strategies, the proposed algorithm follows an iterated manner for balancing diversity maintenance and convergence enhancement during the cooperative co-evolution. Numerical studies comparing the proposed algorithm with 14 popular optimization algorithms demonstrate its effectiveness and efficiency. The practicability of the proposed modelling and optimization approach is validated on a hardware platform.