High-impedance arc faults (HIAFs) pose significant detection challenges in microgrids, particularly as the integration of distributed energy resources (DERs) increases, where conventional overcurrent-based protection schemes often fail. This study presents a novel, cost-effective approach for enhanced HIAF detection based solely on residual voltage measurements. The proposed method applies Feature Mode Decomposition (FMD) to extract multiple signal modes from residual-voltage measurement data, then selects a mode to compute differential energy and formulate a fault detector index (FDIN). The fault detection framework is rigorously tested under diverse operating conditions, including unbalanced loading, noise, variable wind speed, low-impedance faults, and common switching events (e.g., capacitor banks, loads, generators, and transformers), across various grounding configurations (solid, impedance, and resonant grounding). Following fault detection, a residual-feature-driven bagged tree ensemble classifier (RFBTEC) is employed to distinguish HIAFs from normal system conditions and accurately classify the faulty feeder. The effectiveness and practicality of the proposed scheme are validated through real-time simulations on an OPAL-RT (OP4510) platform and further corroborated using field data from Delhi Transco Limited. The results confirm that the proposed voltage-only method achieves fast, accurate, robust, and secure HIAF detection, thereby supporting the reliable and resilient operation of the microgrid system.

This paper presents a data-driven protection scheme for high impedance fault (HIF) classification in microgrids. Conventional overcurrent protection often lacks the sensitivity to detect HIFs because of their low current magnitude and nonlinear characteristics. To address these limitations, we investigate an approach based on multiple machine learning (ML) classifier models for distribution line fault diagnosis. The proposed technique trains ML classifiers directly on raw three-phase differential current signals, allowing the models to autonomously learn patterns without manual feature extraction. The classifiers are optimized through a grid-search procedure and evaluated using a PSCAD/EMTDC model of a 5-bus AC microgrid test system with distributed generations (DGs) operating in both grid-connected and islanded modes. The generated dataset includes healthy conditions, internal faults (with/without HIF), and external faults with current transformer (CT) saturation. Nine ML models are assessed using accuracy, precision, recall, F1-score, receiver operating characteristic (ROC) analysis, computational cost, and noise sensitivity tests. The results indicate that the feedforward neural network (FFNN), random forest (RF), and stochastic gradient descent (SGD) classifiers exhibit robust and consistent classification performance on the test set. Significantly, even when using raw input data, the proposed methodology effectively discriminates subtle HIF signatures from other transient events and CT saturation conditions. These findings highlight the potential of optimized ML models, particularly FFNN and RF classifiers, to provide accurate and robust support for real-time microgrid protection.

Conventional relays encounter difficulties in protecting transmission lines (TLs) connected to converter-based energy sources (CBESs) due to the influence of power electronics on fault characteristics. This article proposes a single-ended intelligent protection method for the TL segment between the grid and a Photovoltaic (PV) plant. The approach utilizes a Recurrence Matrix and an InceptionTime-based system to identify faults by using the mean change in quantiles of 3-phase currents. It determines the fault position and identifies the faulty phase. ReliefF feature selection is applied to extract the optimal quantile features. The scheme's performance is assessed under abnormal conditions, including faults and capacitor and load-switching events, simulated in Power Systems Computer Aided Design / Electromagnetic Transients Program (PSCAD/EMTDC) on the Western System Coordinating Council (WSCC) 9-bus system, with various fault and switching parameters. The scheme is also validated on the New England IEEE 39-bus system and in presence of partially rated converters. Additionally, the validation of the proposed strategy takes into account various conditions, including double-circuit line configuration, noise, series compensation, high-impedance faults, current transformer (CT) saturation, evolving and cross-country faults, remote and local faults, as well as variations in PV capacity, sampling frequency, and data window size. To address label scarcity and improve generalization, semi-supervised learning paradigms including label spreading, label propagation, and self-training are integrated with the InceptionTime framework, enabling near-supervised performance with limited annotated fault data. The results demonstrate that the approach is effective in handling different system configurations and conditions, ensuring the protection of TLs connected to large PV plants.

This article suggests a new and innovative method of embedding dimensionality reduction using variance-based selection methods to achieve maximum storage with optimal accuracy. By leveraging the central limit theorem, we identify specific dimensions with significantly low variance when the embedding engine processes a limited range of semantic meanings. Using diverse datasets, we demonstrate that a small trade-off in accuracy can yield substantial memory savings by mapping and omitting dimensions with minimal impact. This technique is particularly beneficial when embeddings are stored in data repositories. Experiments are conducted across various embedding models to validate the robustness of the method. Additionally, the approach is compared with OpenAI's dimensionality reduction feature in their embedding-large-3 model, highlighting the respective advantages and limitations of each method.

This paper introduces a Simplified Perturb and Observe (SP&O) based control aimed at managing the power transfer between a three-phase/single-phase AC load and a regulated DC load, powered by a three-phase Isolated Induction Generator (IIG) connected to an uncontrolled hydro turbine in remote mountainous areas. The controller’s effective operation depends on precise feedback from sensors, which act as input signals. Specifically, two sensors — one measuring voltage and the other measuring current — are used to supply input to the Real-Time Implementation (RTI) platform. The AD202KY voltage sensor is placed at the Point of Common Coupling (PCC) of the IIG, while the LA55-P current sensor measures the AC load current. To improve signal quality and control feedback voltages, additional Signal Conditioning Circuits (SCCs) using op-amp based TL084 and LF353 are connected to the voltage and current sensors, respectively. These SCCs eliminate noise signals and ensure optimal performance. Additionally, we suggest an external excitation scheme for the three-phase IIG, which includes a single capacitor (1CS scheme) to enable single-phase household supply. The proposed control methodology for power transfer has been successfully tested on two different three-phase, 415 V machines rated at 2.2 kW and 1.5 kW respectively using the dSPACE 1104 as the RTI platform, with a sampling time of 1ms. Importantly, the proposed method shows a reduction in power fluctuations and terminal voltage oscillations in domestic loads during sudden main load applications, underscoring its effectiveness in ensuring smooth power transitions.

In this article, we propose a novel approach for determining image similarity, leveraging advancements in generative artificial intelligence. At the heart of our method is the use of OpenAI’s GPT-4 large language model for generating image captions, combined with the Ada v2 word embedding model for semantic analysis. This technique involves creating textual descriptions of images via GPT-4 and subsequently computing cosine similarity of these descriptions using Ada v2 word embeddings. We compare this innovative approach with traditional image similarity methods, with a particular focus on the VGG16 neural network approach, employing the DISC21 dataset for our analysis. Preliminary results demonstrate the promising potential of this method in the field of image similarity assessment. The paper delves into both the advantages and current limitations of our approach, including constraints like rate limits in experimentation and the rapidly evolving capabilities of language models in vision tasks. Our findings indicate a trajectory towards improved outcomes as these models continue to advance, underscoring the growing intersection of language and vision models in artificial intelligence for applications like image similarity evaluation.

The electricity generated from the present-day large capacity doubly fed induction generator (DFIG) installed wind farm is generally transmitted to utility grid via medium or high voltage transmission line (TL). Due to the restriction of building new TLs, series compensated TLs are some cases preferred for such applications. But, the nonlinear output power versus wind speed relation, control strategies of power electronic interfaced DFIG-wind turbine generators and the nonlinear operation of the thyristor-controlled series capacitor (TCSC) during fault impose adverse impact on the performance of the conventionally used distance relaying-based TL protection schemes. In this article, an improved fault detection and classification technique is proposed to assist distance relay in ensuring fast and reliable protection to TCSC compensated TL linked to DFIG-installed wind farm. In this method, a feature called transient monitoring indexed (TMI) is derived from the measured three-phase currents at the relay location for fault detection and TMI-assisted support vector machine is employed further for fault classification. Performance of the proposed scheme is validated on various fault and nonfault transients simulated on a test power system through MATLAB/Simulink. This protective scheme is farther validated throughout real-time assembled dSPACE DS 1104 control prototype hardware. The superiority of the proposed method is also demonstrated through comparative assessment results with few existing techniques. The overall results justify the merits of the proposed method for fast and accurate detection and classification of faults in such crucial TLs.

Conventional relays face challenges for transmission lines connected to inverter-based resources (IBRs). In this article, a single-ended intelligent protection of the transmission line in the zone between the grid and the PV farm is suggested. The method employs a fuzzy logic and random forest (RF)-based hybrid system to detect faults based on combined linear trend attributes of the 3-phase currents. The fault location is determined and the faulty phase is detected. RF feature selection is used to obtain the optimal linear trend feature. The performance of the methodology is examined for abnormal events such as faults, capacitor and load-switching operations simulated in PSCAD/EMTDC on IEEE 9-bus system obtained by varying various fault and switching parameters. Additionally, when validating the suggested strategy, consideration is given to the effects of conditions such as the presence of double circuit lines, PV capacity, sampling rate, data window length, noise, high impedance faults, CT saturation, compensation devices, evolving and cross-country faults, and far-end and near-end faults. The findings indicate that the suggested strategy can be used to deal with a variety of system configurations and situations while still safeguarding such complex power transmission networks.

This paper presents a systematic approach to detecting High Impedance Faults (HIFs) in medium voltage distribution networks using recurrence plots and machine learning. We first simulate 1150 internal faults, including 300 HIFs, 1000 external faults, and 40 normal conditions using the PSCAD/EMTDC software. Key features are extracted from the 3-phase differential currents using wavelet coefficients, which are then converted into recurrence matrices. A multi-stage classification framework is employed, where the first classification stage identifies internal faults, and the second stage distinguishes HIFs from other internal faults. The framework is evaluated using accuracy, precision, recall, and F1 score. Tree-based classifiers, particularly Random Forest and Decision Tree, achieve superior performance, with 99.24% accuracy in the first stage and 98.26% in the second. The results demonstrate the effectiveness of integrating recurrence analysis with machine learning for fault detection in power distribution networks.

Protective relays can mal-operate for transmission lines connected to doubly fed induction generator (DFIG) based large capacity wind farms (WFs). The performance of distance relays protecting such lines is investigated and a statistical model based intelligent protection of the area between the grid and the WF is proposed in this article. The suggested method employs an adaptive fuzzy inference system to detect faults based on autoregressive (AR) coefficients of the 3-phase currents selected using minimum redundancy maximum relevance algorithm. Deep learning networks are used to supervise the detection of faults, their subsequent localization, and classification. The effectiveness of the scheme is evaluated on IEEE 9-bus and IEEE 39-bus systems with varying fault resistances, fault inception times, locations, fault types, wind speeds, and transformer connections. Further, the impact of factors like the presence of type-4 WFs, double circuit lines, WF capacity, grid strength, FACTs devices, reclosing on permanent faults, power swings, fault during power swings, voltage instability, load encroachment, high impedance faults, evolving and cross-country faults, close-in and remote-end faults, CT saturation, sampling rate, data window size, synchronization error, noise, and semi-supervised learning are considered while validating the proposed scheme. The results show the efficacy of the suggested method in dealing with various system conditions and configurations while protecting the transmission lines that are connected to WFs.

Cascading failure studies help assess and enhance the robustness of power systems against severe power outages. Onset time is a critical parameter in the analysis and management of power system stability and reliability, representing the timeframe within which initial disturbances may lead to subsequent cascading failures. In this paper, different traditional machine learning algorithms are used to predict the onset time of cascading failures. The prediction task is articulated as a multi-class classification problem, employing machine learning algorithms. The results on the UIUC 150-Bus power system data available publicly show high classification accuracy with Random Forest. The hyperparameters of the Random Forest classifier are tuned using Bayesian Optimization. This study highlights the potential of machine learning models in predicting cascading failures, providing a foundation for the development of more resilient power systems.

In this article, the data set of diabetic patients that was used to test the Random Interaction Forest (RIF) algorithm is examined. An Artificial Neural Network (ANN) is designed which matches the accuracy obtained by the RIF algorithm earlier. The proposed ANN model has several regularization methods which are added to prevent over-fitting. The benefits of combining dropout layers with batch normalization layers are justified using the Central Limit Theorem. The model is tuned with TensorFlow’s Bayesian Optimization Tuner to explore a more extensive range of hyper-parameters at a faster pace.

This dissertation highlights the growing interest in and adoption of machine learning (ML) approaches for fault detection in modern power grids. Once a fault has occurred, it must be identified quickly and preventative steps must be taken to remove or insulate it. As a result, detecting, locating, and classifying faults early and accurately can improve safety and dependability while reducing downtime and hardware damage. ML-based solutions and tools to carry out effective data processing and analysis to aid power system operations and decision-making are becoming preeminent with better system condition awareness and data availability. Power transformers, Phase Shift Transformers or Phase Angle Regulators, and transmission lines are critical components in power systems, and ensuring their safety is a primary issue. Differential relays are commonly employed to protect transformers, whereas distance relays are utilized to protect transmission lines. Magnetizing inrush, overexcitation, and current transformer saturation make transformer protection a challenge. Furthermore, non-standard phase shift, series core saturation, low turn-to-turn, and turn-to-ground fault currents are non-traditional problems associated with Phase Angle Regulators. Faults during symmetrical power swings and unstable power swings may cause mal-operation of distance relays and unintentional and uncontrolled islanding. The distance relays also mal-operate for transmission lines connected to type-3 wind farms. The conventional protection techniques would no longer be adequate to address the above challenges due to limitations in handling and analyzing massive amounts of data, limited generalizability, incapability to model non-linear systems, etc. These limitations of differential and distance protection methods bring forward the motivation of using ML in addressing various protection challenges.

Energy efficiency is an existing research challenge in wireless sensor networks (WSNs); to make WSNs energy efficient, many researchers moved to low-duty cycle WSNs with different data forwarding schemes. Low-duty cycle is considered to be a promising approach to design routing protocols for resource-constrained WSNs. Many applications have real-time constraints, which requires an event must be reported to the sink before deadline. Furthermore, wireless links between low-power radios are highly unreliable, and end-to-end latency requirements are challenging due to multiple transmissions for a single packet delivery. By considering probabilistic delay (i.e., delay bounded data delivery with reliability constraints) and energy efficiency issues, little research has been done using different data forwarding schemes, but still, the problem exists. In this work, two strategies are proposed for data forwarding: an energy-optimal path and a delay-optimal path. The early-arrived packet follows an optimal energy route and a diverse path to achieve a delay guarantee with minimum transmission cost. The proposed scheme improves the performance of low-duty cycle WSNs in terms of delay, reliability, stability, and transmission cost. It is observed that under different low-duty cycles, the proposed scheme achieves up to 25% energy conservation compared with both MinEEC and MinEED. At last, undirected spanning trees are considered to balance communication costs. The simulation results of the proposed scheme are compared with the existing methods.

The outages and power shortages are common occurrences in today's world and they have a significant economic impact. These failures can be minimized by making the power grid topologically robust. Therefore, the vulnerability assessment in power systems has become a major concern. This paper considers both pure and extended topological method to analyse the vulnerability of the power system to single line failures. The lines are ranked based on four spectral graph metrics: spectral radius, algebraic connectivity, natural connectivity, and effective graph resistance. A correlation is established between all the four metrics. The impact of load uncertainty on the component ranking has been investigated. The vulnerability assessment has been done on IEEE 9-bus system. It is observed that load variation has minor impact on the ranking.

As renewable energy sources become more widely available, distributed generation (DG) is expected to grow in distribution networks. This research presents a methodology for analysing the reliability of such distribution networks, which can be used in early planning studies. The IEEE 33-bus radial distribution system's reliability is tested first without DG and subsequently with DG connected at various load sites. The reliability indices are calculated using an analytical method. The influence of DG location, number, and availability on reliability indices is examined. The effect of random load variation on system reliability is studied. Because different DG locations provide different reliability indices, the location with the lowest value is picked.

This article presents differential protection of the distribution line connecting a wind farm in a microgrid. Machine Learning (ML) based models are built using differential features extracted from currents at both ends of the line to assist in relaying decisions. Wavelet coefficients obtained after feature selection from an extensive list of features are used to train the classifiers. Internal faults are distinguished from external faults with CT saturation. The internal faults include the high impedance faults (HIFs) which have very low currents and test the dependability of the conventional relays. The faults are simulated in a 5-bus system in PSCAD/EMTDC. The results show that ML-based models can effectively distinguish faults and other transients and help maintain security and dependability of the microgrid operation.

Distance relays mal-operate for transmission lines connected to type-3 Wind Farms (WFs). This paper proposes a waveshape property based protection of the intertie zone between wind farm and grid during 3-phase faults. It mitigates the challenges faced by the normally used distance relays and ensures the protection systems' security and dependability. The proposed scheme uses the auto-regressive coefficients of the 3-phase currents obtained from the Current Transformer at one end to distinguish the faults fed by the type-3 WFs and the primary grid and determines the fault zone accurately. PSCAD/EMTDC is used to verify the validity of the technique on three test systems. The results obtained with different wind speeds, crowbar resistance, fault resistance, inception time, and fault locations are encouraging and suggest the possible utilization of feature-based algorithms to improve the power system distance relaying system. In addition, the protection scheme can be utilized for lines compensated with series capacitors and phase shifting transformers.

Several conventional and non-conventional transient conditions cause differential relays associated with Phase Angle Regulators to malfunction. For Two-core Symmetric Phase Angle Regulators, this article investigates the suitability of time and time-frequency feature-based estimators to differentiate internal faults from other transient conditions such as overexcitation, external faults with current transformer (CT) saturation, and magnetizing inrush. Subsequently, the faulty core unit (series or exciting) is located, and the transients are identified. Six well-known classifiers are trained on features extracted from one-cycle of post transient 3-phase differential currents filtered by an event detector. Maximum Relevance Minimum Redundancy, Random Forest, and exhaustive search with Decision Trees are used to select the relevant wavelet energy, time-domain, and wavelet coefficient features respectively. The fault detection scheme trained on XGBoost classifier with hyperparameters obtained from Bayesian Optimization gives an accuracy of 99.8%. The reliability of the proposed scheme is verified with varying tap positions, noise levels, and transformer ratings; and under different conditions like CT saturation, fault during magnetizing inrush, series core saturation, low current faults, and integration of wind energy. As a potential application, the methodology can be deployed to supervise microprocessor-based differential relays to improve the security and dependability of the protection system.

This article solves the problem of accurate detection of internal faults and classification of transients in a five-bus interconnected system for phase angle regulators (PARs) and power transformers (PTs). The analysis prevents mal-operation of differential relays in case of transients other than faults, which include magnetizing inrush, sympathetic inrush, external faults with current transformer (CT) saturation, capacitor switching, nonlinear load switching, and ferroresonance. A gradient boosting classifier (GBC) is used to distinguish the internal faults from the transient disturbances based on 1.5 cycles of three-phase differential currents registered by a change detector. After the detection of an internal fault, GBCs are used to locate the faulty unit (PT, PAR series, or exciting unit) and identify the type of fault. In case, a transient disturbance is detected, another GBC classifies them into the six disturbances. Five most relevant frequency- and time-domain features obtained using information gain are used to train and test the classifiers. The proposed algorithm distinguishes the internal faults from the other transients with a balanced accuracy ($\bar{\eta }$) of 99.95%. The faulty transformer unit is located with $\bar{\eta }$ of 99.5% and the different transient disturbances are identified with $\bar{\eta }$ of 99.3%. Moreover, the reliability of the scheme is verified for different ratings and connections of the transformers involved, CT saturation, and noise levels in the signals. These GBC classifiers can work together with a conventional differential relay and offer a supervisory control over its operation. PSCAD/EMTDC software is used for simulation of the transients and to develop the two- and three-winding transformer models for creating the internal faults including interturn and interwinding faults.