Search

Quadratic cost function such as Mean Squared Error (MSE) has been a widely used objective function for training deep neural networks to develop energy forecasting models in Smart Grids. In this work, Penalizing Underestimation Logarithmic Squared Error (PULSE), a novel objective function is proposed with the aim of reducing the tendency of deep learning models to underestimate the target variable. Stacked Long Short-Term Memory (LSTM) networks are adopted on the time series load demand data to investigate the performance of the proposed cost function against the widely used MSE cost function. The evaluation is performed using open-source real-world electricity load diagrams dataset covering a period of three years. The performance of the proposed scheme is examined with deep learning models through several experiments. The results demonstrate that the proposed scheme is able to eliminate the tendency to underestimate and provides competitively accurate load demand forecasting results. The results are additionally compared against the state-of-the-art machine learning models developed in the literature. The proposed cost function maintains the RMSE around 4*10-2 kWh which is also the RMSE for deep learning models with MSE cost function and delivers 25% improvement in MAPE while also eliminating the underestimation of load demand.

This paper proposes a method to analyze aging response variance in terms of statistical parameters and geometry specifications of medium voltage power cable. The electric stress distribution in the single-core medium voltage power cable model is logged quantitatively using finite element modeling. The dependable parameters such as electrical conductivity, thermal conductivity, ambient temperature, relative permittivity, defect position and size, act as variables in the model to analyze simulated electric stress in the power cable. The variation in electrical stress distribution under the influence of different factors impact distinctly on cable degradation over time. Moreover, the correlation between these factors is quantified using derived statistical parameters to form a correlation matrix. In this work, both the variation coefficients and mean stress intensity are utilized as response measurement variables. A multivariable linear regression is applied to derive the relationship between the model parameters and the response variables. This study shows that the stress variation is strongly correlated with electrical conductivity and thermal conductivity, while the maximum stress intensity is most impact by defect sizes and position. The implication and consequence of the rest of the model parameters on electric stress distribution are also analyzed and discussed.

Effective short-term Locational Marginal Price Forecasting (LMPF) is difficult in view of the high sensibility of the electricity price in deregulated markets. This paper proposes an accurate forecasting algorithm for LMPF using the latest breakthroughs in deep learning. Specifically, the proposed strategy is composed of a Hybrid Feature Selector (HFS), hyperparameter tuning using Simulated Annealing (SA)-based multi-objective optimization algorithm, and self-Attention-based Long Short-Term Memory (ALSTM). The proposed HFS includes Extreme Gradient Boosting, Elastic Net, and random forest models to rank the features based on their relevance. The experimental results are compared with multiple benchmark algorithms to demonstrate the robustness and efficiency of the proposed framework. The main contributions of this paper include 1) An efficient model perfectly tailored for LMPF is introduced; 2) The effectiveness superiority of the proposed SA-ALSTM is verified on publicly available electricity market data. Extensive experimental results validate the competitive performance of the proposed SA-ALSTM in terms of score measures.

Rectifier circuits have a pivotal role in many home appliances and industrial devices. On the other hand, the non-linear nature of these devices results in power quality degradation in the power system. To minimize the negative impact of rectifiers on the power system, various active rectifier topologies are proposed. Although the active rectifiers show better performance in terms of power quality, the control of such rectifiers needs to be given special attention because it is widely used in different applications. This study proposes a model predictive control (MPC) for the full-bridge boost rectifier with constant switching frequency. This study also presents the design steps of the boost-type rectifier. Furthermore, a traditional MPC controller is used for comparison purposes. The system performances are assessed through Matlab/Simulink.

This paper builds upon an autonomous indexing-based model predictive control (MPC) framework for smart photovoltaic (PV) inverters in a connected community with dynamic topology and changing configurations. The proposed controller offers black start capability for connected communities with a high penetration of PV sources after a failure or power outage event. A higher margin of robustness and flexibility is achieved by enabling autonomous PV inverters operating modes while remaining fully synchronized. The autonomous indexing scheme assigns operational modes of distributed PV inverters, i.e. voltage or current control, in a sequential manner and enhances the resiliency of islanded connected communities against the contingency of grid-forming PV inverter interruption. The proposed controller is demonstrated to optimize the black start performance of an islanded and fully de-energized system. Simulated case studies explore the controller’s functionality and validate its black start restoration performance.

Power electronics-dominated grid (PEDG) has low inertia owing to the fast switching and the reduced penetration of rotating masses, which makes such grids vulnerable to disturbances. This paper proposes a predictive control scheme that emulates virtual inertia to achieve resilient and stable operation of the PEDG in response to load and generation disturbances. The conventional primary control layer for power electronics-based generation commonly utilizes cascaded PI controllers to regulate active/reactive power. This paper eliminates the need for cascaded linear control and employs a model predictive control (MPC) that enables virtual inertia emulation. Leveraging the inherent bandwidth of MPC enables tracking the desired rate of change of frequency (ROCOF) for various disturbances superior than the PI-based controllers. The MPC's penalty function replaces the cascaded linear control while eliminating significant tuning efforts. The investigated virtual inertia topologies for gird-following (GFL) and grid-forming (GFM) inverters are capable of regulating the ROCOF under frequency events. Several case studies are presented to demonstrate the inertia emulation effect considering the interaction between GFL and GFM inverters in PEDG.

This paper proposes a smart inverter that is resilient against cyber-physical intrusion on the sensors. The proposed system is intended to detect intentional electromagnetic interference (EMI) injection in the current sensor data and to provide a healing solution for system restoration. A solution based on super-twisting sliding mode observer (STSMO) is designed to estimate the grid current. The proposed solution is aimed to detect the jeopardized current sensor due to the cyber-physical attack by taking into consideration the error between the observed and measured currents. In addition, this work adopts the flatness-based control (FBC) technique to control the inverter and feed the current to the grid with low harmonic content even under EMI attack. The adopted flatness controller relies on the reference feedback instead of disturbance feedback which helps in reducing the external noises. Therefore, the proposed solution is suitable to establish a safe and secure connection of the inverters to the grid. Real-time simulation results are carried out to prove the concept of the proposed solution.

This paper presents a advanced field-oriented control (FOC) for an open-end stator windings six phase induction motor (SPIM-OESW) with fuzzy logic (FL) speed controller. The proposed motor topology is fed by four, three-phase voltage inverters with two isolated DC sources. The FOC technique requires a precise knowledge of some machine's variables such as rotor position and rotor flux which are difficult to measure. Usually, the rotor position and motor speed are measured by a shaft-mounted encoder, but that makes the maintenance activities more difficult, and reduces system's reliability. Motor flux measurement is more complicated and requires machine's built in sensors. Therefore, a rotor flux-based model reference adaptive system (RF-MRAS) is proposed in this paper for the estimation of both rotor flux and rotor speed. In addition, the classic integral proportional (PI) controller is replaced by a FL controller to achieve better speed response at different operating points. The proposed control scheme is built within the Simulink environment combined with the Real-Time platform. Real time results are obtained to illustrate the effectiveness of the proposed control under different operating conditions.

This paper proposes an effective Sliding Mode PWM (SMPWM) control strategy based on the average model of the 5-level Packed-U-Cell (PUC5) inverter. The proposed controller is designed from a stability point-of-view based on the Lyapunov control theory, to balance the PUC5 capacitor voltage and feed a sinusoidal current to the load with low harmonics distortion. The proposed controller is offering the following advantages over the conventional controllers: 1) Reduced switching stress and low harmonics content of the load current owing to the fixed switching frequency; 2) High robustness against disturbances and parameters mismatch. Simulation and experimental results are presented to confirm the effectiveness of the proposed controller.

Power utilities are struggling to reduce power failure incidents in substations and their components to operate more reliably and economically [1]. Many power failures are produced directly or indirectly because of the insulation system of utility components [2], [3]. The selection of the insulation should ensure power plant operational continuity along with completely resolving or significantly limiting the actual power system's failures [4]. Gas insulated substations (GIS) have the best insulation performance which ensures achieving minimum failure incidents, although at high installation cost. The most common insulating gas used in GIS is Sulfur hexafluoride (SF 6 ) gas, which is widely used as an effective electrical insulation as well as an arc-quenching medium [5]. Basic GIS and gas insulated transmission lines (GITL or GIL) consist of a conductor supported by solid insulators inside an enclosure filled with SF 6 gas or its mixture [6].

There are two main types of solar power systems: thermal and photovoltaic (PV). The direct thermal use of solar energy for water heating is very old, mature and cheap technology. The world capacity of such solar thermal energy is over 400 GW‐thermal and is expected to increase sharply. For electricity generation, concentrated solar power is a promising technology for places with the right environmental conditions for generating electricity with higher efficiency than PV technology. This technology is only practical for high electrical power generation and storage, because of the complicated technology with relatively high initial and running costs. The use of solar energy is very attractive in many applications such as water heating, solar cooling (heat‐powered air conditioning), and even water desalination. Solar thermal technologies have various levels of temperature and pressure, with various heat‐conducting and storage materials.

Smart grid (SG) is the solution to solve existing problems of energy security from generation to utilization. Examples of such problems are disruptions in the electric grid and disturbances in the transmission. SG is a premium source of Big Data. The data should be processed to reveal hidden patterns and secret correlations to extrapolate the needed values. Such useful information obtained by the so-called data analytics is an essential element for energy management and control decision towards improving energy security, efficiency, and decreasing costs of energy use. For that reason, different techniques have been developed to process Big Data. This paper presents an overview of these techniques and discusses their advantages and challenges. The contribution of this paper is building a recommender system using different techniques to overcome the most obstacles encountering the Big Data processes in SG. The proposed system achieves the goals of the future SG by (i) analyzing data and executing values as accurately as possible, (ii) helping in decision-making to improve the efficiency of the grid, (iii) reducing cost and time, (iv) managing operating parameters, (v) allowing predicting and preventing equipment failures, and (vi) increasing customer satisfaction. Big Data process enables benefits that were never achieved for the SG application.

This paper presents a decoupled active and reactive power control scheme for grid-tied quasi-impedance source cascaded multilevel inverter (qZS-CMI). For photovoltaic (PV) applications, the proposed control scheme is based on an enhanced finite-set model predictive control (MPC) to harvest the desired active power from the PV modules with the ability to provide the ancillary services for the grid. The proposed control scheme has two modes of operation: normal grid mode and low voltage ride through (LVRT) mode. In normal grid mode, the controller commands the qZS-CMI to operate at the global maximum power point (MPP). The proposed technique regulates the impedance network's current and voltage according to the MPP of PV strings and grid current/voltage requirements. In LVRT mode, the controller commands the qZS-CMI to provide the required reactive power to the grid during voltage sags as an ancillary service from the inverter as imposed by the grid codes. The main features of the proposed system include the global MPP operation during normal grid condition, LVRT capability during a grid voltage sag, mitigation of the PV modules mismatch effect on overall energy harvesting, seamless transition between a normal grid and LVRT modes of operation, and an efficient predictive controller that exploits the model redundancies in the control objectives. Several real-time experiments are conducted to verify the system performance with transients in both the solar irradiance and the grid voltage.

The rapid growth of gas insulated switchgears as a compact, efficient, and reliable device has recently been given great attention. Albeit gas insulated switchgears can seldom suffer from failure due to the high resiliency and robustness, some severe damages have been experienced by such devices particularly in the event of partial discharge. Thus, monitoring such accidents has become a vital part of power systems reliability. The ultra-high frequency techniques have recently shown superior performance in the detection and classification of electromagnetic waves produced by partial discharge. This is mainly due to the great immunity to the noise of the ultra-high frequency detection techniques compared with the very-high-frequency counterparts. This review paper highlights the mathematical aspects of the electromagnetic waves generated by partial discharge. It also delivers an overview of the electromagnetic wave behavior in the complex structure of gas insulated switchgears, and outlines the important characteristics of the internal and external partial discharge detection using ultra-high frequency methods.

In this paper, a deadbeat predictive control strategy, which accounts for saturation effects, is developed for the control of electric drives with neutral-point-clamped inverter. Although predictive control offers essential advantages, its performance strongly relies on the model accuracy and can be compromised when encountering complex magnetic phenomena. Therefore, a methodology based on finite-element methods is suggested in this paper for accurately extracting the system parameters and determining the dynamic motor trajectories as functions of the core saturation. The incorporation of the direct- and cross-saturation effects into the deadbeat control routine allows the developed control scheme to reduce the current distortion and operate efficiently in both constant torque and power regions. The suggested controller, which is accompanied by a space vector-based modulator for operating with constant switching frequency, is compared with a linear control strategy by considering several performance indices. Experimental and simulation results are presented for assessing the effectiveness of the complete control scheme under steady-state and transient conditions.

The power grid stability is highly impacted by the fluctuating nature of renewable energy sources. This paper proposes a deep learning method-based bidirectional gated recurrent unit for smart grid stability prediction. For automatic tuning, this study employs Simulated Annealing algorithm to optimize the selected hyperparameters and enhance the model forecastability. The proposed forecasting model's performance is evaluated using electrical grid stability simulated data set. The proposed method provides an accurate point and interval grid stability prediction. Simulation results are conducted to prove the high performance of the proposed method. Furthermore, comparative analysis is performed to demonstrate the superiority of the proposed strategy over some state-of-the-art available solutions.

This article proposes an accurate Stacking Ensemble Classifier (SEC) for decentral Smart Grid control Stability Prediction. The proposed SEC consists of stacking two base classifiers; specifically, eXtreme Gradient Boosting machine (XGBoost) and Categorical boosting (Catboost), and one meta-classier, Light Gradient Boosting Machine (LGBM). The proposed technique shows an excellent ability to classify the grid instabilities using a supervised learning approach accurately. Extensive experiments have been conducted, demonstrating the superiority of the proposed SEC model over multiple benchmarks. In summary, this paper's main contributions consist of 1) proposing a new model-based ensemble learning 2) tailoring an efficient data-driven technique for grid stability detection and classification. Numerical results are to validate the proposed model's high effectiveness.

This paper deals with an interactive three-phase shunt active power filter (SAPF) connected to a Photovoltaic (PV) source through a Z-source inverter (ZSI). The filter is controlled by synchronous reference frame (SRF) controller based on Multi-variable filter (MVF). The proposed system is designed to reduce the total harmonic distortion (THD) of the source current. In addition, the filter utilizes the PV source to inject active power into the grid as an additional support. The proposed system has been validated through extensive real-time simulation carried out using OPAL-RT-5600 under several scenarios. The conducted investigations are aimed to demonstrate the good performance of the filter in terms of harmonic elimination, operating under weak grid conditions, and injecting active power to the grid.

One of the major factors that lead to energy losses for utility distribution systems is electricity or energy theft. Energy theft is tampering with smart meter reading to reduce customer energy usage and reduce electricity bills. A thief customer tends to consume more energy and hence, the theft negatively affects the power supply quality in the form of transformer overload, voltage unbalance, and voltage drop on system buses. Meanwhile, it also causes great economic losses for the business of electric utility. In order to enable efficient energy theft detection, data-driven approaches including utilizing trained deep neural networks are proposed in this paper. The machine learning approaches can detect energy theft involving stealthy connections or meter tampering at the level of smart meters or aggregated levels. In this work, the detection effectiveness of different approaches is evaluated on real case study data at the end consumer level. The challenges of class imbalance and the missing values (around 25% of the whole fields) are addressed in the LSTM-based methodology. In this paper, results are obtained on real energy consumption data to show the higher performance of the proposed solutions compared to previously presented work.

The energy paradigm is making the modern power grid more difficult to study, design, and control. Precisely speaking, the pace of the new energy paradigm involves the high penetration of power electronics systems in the power grid, which becomes a challenge from stability, vulnerability, and power-quality points of views. Therefore, various literature have focused on defining the modern power grid as the power electronics-dominated grid (PEDG).

This paper proposes an effective deep learning framework for Short-Term Load Forecasting (STLF) of multivariate time series. The proposed model consists of a hybrid Convolutional neural network-Bidirectional Long Short-Term Memory (CBiLSTM) based on the Evolution Strategy (ES) method and the Savitzky–Golay (SG) filter (SG-CBiLSTM). The adopted methodology incorporates the virtue of different prepossessing blocks to enhance the performance of the CBiLSTM model. In particular, a data-augmentation strategy is employed to synthetically improve the feature representation of the CBiLSTM model. The augmented data is forwarded to the Partial Least Square (PLS) method to select the most informative features above the predefined threshold. Next, the SG algorithm is computed for smoothing the load to enhance the learning capabilities of the underlying system. The structure of the SG-CBiLSTM for the ISO New England dataset is optimized using the ES technique. Finally, the CBiLSTM model generates output forecasts. The proposed approach demonstrates a remarkable improvement in the performance of the original CBiLSTM model. Furthermore, the experimental results strongly confirm the high effectiveness of the proposed SG-CBiLSTM model compared to the state-of-the-art techniques.

Accurate Short-Term Load Forecasting (STLF) has been considered a topic of extreme importance for efficient energy management, reliable energy transactions, and economic operation dispatch in smart grids. However, the continuous instability of the load demand essentially due to the high volatility of weather conditions and customers’ demand behavior dramatically affects the STLF accuracy. In order to overcome this problem, five effective Deep Learning (DL) techniques are proposed for multivariate time series STLF based on Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), and stacked Auto-Encoder (AE). These DL based techniques are consolidated to build stacked Bidirectional GRU (BiGRU), Convolutional LSTM (ConvLSTM), stacked Bidirectional LSTM-AE (BiLSTM-AE), hybrid CNN-LSTM-AE (CNN-LSTM), and LSTM-AE (LSTM-AE) techniques. Simulation studies are conducted to demonstrate the performance superiority of BiLSTM-AE compared to the other DL models. The main contributions of this paper include 1) integrating a variety of deep neural networks for STLF; 2) employing time series as a benchmark to compare between heterogeneous DL architectures; 3) conducting the analyses on real data set.

This work provides cybersecurity analytics for a high photovoltaic (PV) penetrated distribution network, representing future power electronics dominated grid (PEDG). Firstly, the impact of active and reactive power (PQ) set-points manipulation on the network is studied. Then, an intrusion detection system (IDS) is developed for identifying the potentially compromised PV inverters in the network. The proposed IDS is based on defining the normal, safe, and abnormal operation regions of the PV inverters from point of view of the steady state voltage stability of the network. These three-operation regions are identified by utilizing active power, reactive power, and voltage (PQV) limits of each specific grid-following inverter in the network. Each grid-following inverter’s PQV contour includes the information of network topology, inverter ratings, and inverter controller. The developed PQ limits are integrated into the centralized secondary control layer for realization of the proposed IDS. Furthermore, the proposed secondary control layer is capable of providing remedial actions during an anomaly event to enhance the grid resiliency. The theoretical analyses are verified by several attack scenarios for a network of grid-following inverters.

This paper presents a proactive intrusion detection system (IDS) for smart distribution power systems. The considered attack scenario is manipulation of the advanced measuring infrastructures (AMIs) readings and/or smart inverters data. These manipulated data from the grid edge devices mislead the grid operator for making proper operational planning decisions. In a stealthy attack model, where the attacker compromises significant number of these smart devices, serious demand-supply unbalance can occur that may result in major blackouts. The proposed IDS is based on a condition monitoring vector (CMV) equipped with a learned ultra-short-term demand forecasting (USTDF) mechanism. This cybersecurity approach is able to verify smart devices readings. In the proposed method, the instantaneous difference of collected AMIs and other smart devices data with the ultra-short term forecasted demand is defined as the CMV. This vector probes a pre-defined error band for identifying the compromised smart devices. The learned USTDF mechanism is based on the distribution grid historical load profile and the temperature data for the goal area. An accurate multi-dimensional regression model is developed and learned for forecasting the load behavior in this area. Finally, the suspicious areas are flagged or become separated from the main grid by the network operator based on the proposed CMV outcomes and the output of decision-making module. The proposed IDS aims to enhance the cybersecurity of the smart devices at the grid-edge that plays major role in ensuring the resiliency of the grid. The theoretical analyses are verified by several case studies.

High voltage (HV) testing of Gas-insulated switchgears (GIS) has been predominantly employed using the ultra-high frequency (UHF) detection techniques. Initiated partial discharge (PD) currents are coupled with electromagnetic (EM) waves, which are then captured by the UHF antennas. To determine the minimum detectable signal received by a UHF sensor, CIGRE sensitivity verification recommendations are utilized. In this paper, a 3-phase GIS model, which utilizes CIGRE recommendations to analyze the EM waves behavior, is built based on 3-D full-Maxwell finite element (FE) solver. This approach permits initiating a PD source using a transmitting sensor, which can then be acquired using other receiving sensors. The modeling approach introduced in this paper can be used to understand the behavior of signals propagating inside GIS systems for better utilization of UHF sensors in detecting PD inside such capital assets. The proposed three-phase GIS model investigates the EM wave propagation in time-domain and frequency-domain using step 1 of the CIGRE sensitivity verification recommendations. The impact of the relative position of the PD source and sensors, and the dielectric spacers on EM wave propagation is investigated and analyzed. The obtained results show that three-phase GIS structures act as cavity resonators when PD is created. Moreover, both low-frequency and high-frequency modes of propagation exist in such cavities.

This paper presents a control scheme for single-phase grid-following inverters that does not rely on the phase-locked-loop (PLL) algorithm for synchronizing the injected current with the grid. The absence of the PLL avoids unstable low damping modes that are possibly originated from the PLL’s nonlinear nature. Thereby, assuring the control scheme robustness, especially, in weak grid conditions. Furthermore, the proposed PLL-less control achieves tracking the commanded active and reactive power (PQ) set-point references in a single loop control scheme. Hence, reducing controller design complexity compared to conventional cascaded dual loop PQ control scheme. This work also illustrates the necessary conditions to assure that the control is asymptotically stable with weak resistive and inductive grids. Finally, several case studies are implemented to validate the theory developed.

This paper proposes a hybrid constant power generation - power ramp rate control (CPG-PRRC) scheme to regulate the ramp-rate of the power injected into the grid by a cluster of photovoltaic (PV) sources. The ramp-rate regulation of the injected PV active power is achieved dynamically in real-time under highly fluctuating PV ambient conditions and without considering PV forecasting data. The proposed hybrid CPG-PRRC control scheme integrates the concept of constant power generation with power ramp rate control to ensure mitigation of the potential voltage/ power fluctuations seen by the grid and protects it from source side instabilities. Moreover, the proposed scheme is capable of regulating the PV power ramp-rate within the set limit during power ramp-up as well as ramp-down scenarios and does so without the use of energy storage devices. The ramp-rate is regulated by maintaining the PV operation set– point on the left side of the maximum power point (MPP) and dynamically increasing or decreasing the power reference at a rate which is within the ramp limit. Furthermore, this paper addresses potential stability concerns for the PV architecture in hand with the CPG-PRRC. Several test scenarios are provided to verify the theoretical expectations.

This paper presents an observer based predictive control scheme for grid-interactive inverters with intrusion detection capability. The proposed framework is robust to potential cyber-physical attacks due to intentional electromagnetic interference (EMI) on the vulnerable Hall-effect current sensors of the inverter. The robust operation of smart inverters at the grid-edge is highly dependent on the reliable and accurate current measurement in their feedback control scheme. The proposed security framework highlights two layers; the first layer detects an intrusion by EMI attack on the current sensor on-board of the inverter, while the second layer supports the first layer by establishing a secure line of communication between the inverter and the supervisory controller by using a time-sensitive passcode key. The observer uses the voltage measurements, which are not susceptible to EMI attacks targeting current sensors. The smart inverter flags the compromised current sensors at the grid-edge in a proactive manner by evaluating the current sensor measurements and the observer output in a constraint penalty function – leveraging the model predictive control phenomena. The intrusion detection and controller performance has been verified by multiple case studies.

In this study, a new multilevel inverter (MLI) configuration is proposed to generate higher number of levels with minimal control switches. The proposed inverter's nomenclature is E K θ MLI, termed from its stems of shape. Conventional single-phase H-bridge module is modified to design the structure of θ MLI called `θ cell' (where, K = 0) by incorporating three symmetrical DC-sources and replacing one leg of H-bridge by bidirectional switches. The θ cell of the proposed E K θ MLI generates five-level AC waveform at the output by using only four controlled switches (two unidirectional and two bidirectional). Therefore, the size, cost and switching logic of the MLI is reduced. The basic structure of E K θ MLI called `Eθ MLI' (where K = 1) is designed which is capable of generating seven-level. For further extension, double-switch-E cell (DS-E cell) is added in the middle and each DS-E cell increases the four levels by utilising two bidirectional switches and two DC supplies. The generalised MLI configuration is explained in niceties to generate a higher number of levels (more than 5). The modes of operation, voltage across the switches, power losses of E K θ are discussed. The proposed MLI is experimentally investigated using level shifting pulse width modulation.

In this paper, Finite Element (FE) solver of COMSOL Multiphysics is used to model Partial Discharge (PD) defects inside an L-structured Gas-Insulated Switchgear (GIS). The defects are modeled by short dipoles, which were randomly placed inside the GIS. Gaussian Pulses were used as inputs to the short dipoles. Consequently, Electromagnetic (EM) waves start propagating inside the GIS enclosure. Five Ultra-High Frequency (UHF) sensors were used to acquire the propagating waves. A large number of utilized sensors is attributed to the limited number of simulation scenarios that were used. The captured signals were fed into a Gradient Boosting Machine (GBM) algorithm to localize the defects inside the GIS. The dataset fed into the GBM algorithm were power, voltage and cumulative energy received by each sensor. The biggest challenge associated with applying the GBM algorithm is the limited amount of obtained data. This paper provides a method of modeling PD defects inside medium and high voltage devices which can then be used for PD localization. The obtained results showed high accuracy with RMSE values of 31.46mm, 71.85mm, and 58.13mm for the X, Y, and Z axes, respectively.

This paper evaluates the performance of different feature extraction or dimensionality reduction techniques for the applications of short-term power forecasting using smart meters' data. The number and data type of input features are crucial to the performance of power forecasting models. The performance of the machine learning models decreases with the increase in the number of input features. That is, the machine learning models tend to overfit, and the forecasting accuracy is reduced. The performance of the feature extraction or dimensionality reduction techniques has been evaluated in the context of the forecasting applications with models involving Artificial Neural Networks (ANN), Long Short-Term Memory (LSTM), and Linear Regression (LR). The application is day-ahead forecasting on a real and open dataset of energy utilization by households in England. The obtained results depict the importance of dimensionality reduction techniques for higher accuracy and faster training times. While linear Principal Component Analysis (PCA) is a preferred dimensionality reduction technique for faster training times, kernel PCA, Non-negative Matrix Factorization (NMF), Independent Component Analysis (ICA) and Uniform Manifold Approximation and Projection (UMAP) yield better accuracies.

Rotating machine is one of the critical elements of the power systems whose reliability is improved with proper condition assessment and fault detection techniques. Partial discharge (PD) identification is essential for the assessment of insulations particularly for medium and high power units operating at higher voltage levels. This paper proposes a finite element based model that illustrates the PD behavior in the electric machine stator winding. It also computes the PD features from the defects similar to the characteristics obtained by commercial detection systems. The proposed finite element analysis based computational model utilizes multiphysics which includes electrodynamics and thermal boundary conditions. This model describes complete variation in properties of defects and insulation of the stator winding under various operating stresses. The simulated PD features that include charge magnitude and PD occurrence are comparable with commercially available solutions. This model illustrates the level of deterioration in the winding insulation, whose characterization is used for defects classification.

The forecasting of the high intermittency of Photovoltaic (PV) energy in smart grid is a persisting challenge. The proposed paper takes this challenge by presenting accurate forecasting techniques. PV power forecasting contributes to energy sector stability, controllability, and utilization through systematic monitoring for proper energy operation and optimization of grid-load balance. This paper addresses a novel paradigm that effectively copes with unpredictable extreme meteorological conditions. The proposed technique combines the Bayesian Ridge Regression (BRR) model, Continuous Wavelet Transform (CWT), and Gradient boosting with categorical features (Catboost). The architecture of the proposed model is based on the acquisition of features inputs, which sorts those features according to their importance. This ranking deploys a Bayesian Ridge Regression model to select the most relevant features. Then, the CWT decomposition technique converts the features chosen into a time-frequency domain. Catboost model generates the forecast output for one day ahead. The final results are deduced using inverse CWT. The Australian weather data have been used to evaluate the performance of the proposed technique on short short-term power forecasting for large-scale PV plants. The evaluation has been conducted using score metrics, visualization curves, and to-fold cross-validation. Simulation results are conducted to confirm the performance of the proposed technique.

The tuning of weighting factor has been considered as the most challenging task in the implementation of multi-objective model predictive control (MPC) techniques. Thus, this paper proposes an artificial intelligence (AI)-based weighting factor autotuning in the design of a finite control set MPC (FCS-MPC) applied to a grid-tied seven-level packed U-cell (PUC7) multilevel inverter (MLI). The studied topology is capable of producing a seven-level output voltage waveform and inject sinusoidal current to the grid with high power quality while using a reduced number of components. The proposed cost function optimization algorithm ensures auto-adjustment of the weighting factor to guarantee low injected grid current total harmonic distortion (THD) at different power ratings while balancing the capacitor voltage. The optimal weighting factor value is selected at each sampling time to guarantee a stable operation of the PUC inverter with high power quality. The weighting factor selection is performed using an artificial neural network (ANN) based on the measured injected grid current. Simulation and experimental results are presented to show the high performance of the proposed strategy in handling multi-objective control problems.

This paper proposes customization for a hybrid balancing scheme to reduce capacitor voltage ripple in a two-leg five-level neutral point clamped inverter. This customization introduces modulation changes in two regions of the hybrid balancing scheme. Following these changes, the active balancing control target options are re-evaluated to better suit the voltage ripple reduction objective. The paper outlines the backgrounds of changes and alternatives regarding the improvisation being adapted. Lastly, the applicability of the improvisations is demonstrated using a comparative Matlab-Simulink simulation of a two-leg five-level transistor clamped inverter under transient and steady-state conditions.

This paper deals with the selective harmonic elimination pulse width modulation (SHE-PWM) technique. This technique is used for the elimination of selected dominant low order harmonics in the multi-level inverter output voltage. The presence of these harmonics is the essential drawback of such kind of inverters; especially when it is used for the control of different AC drivers. The SHE-PWM is based on the minimization of a constrained nonlinear objective function whose variables are the switching angles used for multi-level inverter control. The solution of this optimization problem can be achieved using different metaheuristic optimization algorithms. An approach of SHE-PWM based on Particle Swarm Optimization (PSO) algorithm is proposed in this paper. Different patterns of optimal switching angles shown in a previous work, which are based on groebner bases and symmetric polynomials theory (GBSP), are improved in this paper using PSO. The improved earlier patterns are compared with the proposed approach. The obtained experimental and simulation results are aimed to verify the efficiency and the capability of the proposed approach in improving the Total Harmonic Distortion (THD) while eliminating the desired low frequency harmonics.

This paper provides a review of the control schemes implemented for microgrids (MGs) or grid clusters to enable seamless transition between grid-connected (GC) and islanded (IS) modes of operation. Grid of MGs is a potential solution towards a resilient power grid with high penetration of renewable energy resources (RES). The deployed MGs must be capable of operating in both GC as well as IS mode of operation in the occurrence of anomalies or natural disasters. Such required ability of the MG will ensure uninterrupted energy services to critical loads and infrastructures. This mandates the implementation of control strategies to enable smooth transition of the MG between GC and IS modes and avoid deviations in voltage/current due to misalignment in phase and frequency during the transition process. The main purpose of this paper is to provide an overview of the challenges and existing techniques in literature to mitigate voltage and frequency $(\pmb{V}/\pmb{f})$ fluctuations at the MG's point of common coupling (PCC) and the power grid during the transition process; and to motivate the development of advanced control schemes in this area of MG-related researches.

With the advent of Internet-of-Things (IoT) devices, including smart meters and sensors in the smart grid, there has been immense research interest in big data management, analytics, and parallel processing of data. However, complex hardware and software parameters configurations and in-depth understanding of the data processing design are essential for efficient utilization of big data analytics platforms. In this work, we analyze the parallelization of load prediction by utilizing spark regression python library to assess the performance with workloads of up to 8 nodes. The results of different configurations have been studied and analyzed against the performance of Apache Spark. It was found that a trade-off between the number of nodes and cores is necessary to perform efficient parallel computing. Multiple sets of combinations of nodes and cores are considered in this paper to evaluate the performance. The work also signifies the importance of high-performance computing capability for smart meters big data management. The obtained results indicate that the computational time is not only dependent on the data size but also on the number of compute nodes and the number of cores assigned to run the job.

Fault Detection and Isolation (FDI) in Heating, Ventilation, and Air Conditioning (HVAC) systems is an important approach to guarantee the human safety of these systems. Therefore, the implementation of a FDI framework is required to reduce the energy needs for buildings and improving indoor environment quality. The main goal of this paper is to merge the benefits of multiscale representation, Principal Component Analysis (PCA), and Machine Learning (ML) classifiers to improve the efficiency of the detection and isolation of Air Conditioning (AC) systems. First, the multivariate statistical features extraction and selection is achieved using the PCA method. Then, the multiscale representation is applied to separate feature from noise and approximately decorrelate autocorrelation between available measurements. Third, the extracted and selected features are introduced to several machine learning classifiers for fault classification purposes. The effectiveness and higher classification accuracy of the developed Multiscale PCA (MSPCA)-based ML technique is demonstrated using two examples: synthetic data and simulated data extracted from Air Conditioning systems.

Load forecasting in smart grid is the process of predicting the amount of electrical power to meet the short, medium and long term demands. Accurate load forecasting helps electrical utilities to manage their energy production, operations, control and management. Most of the state-of-the-art forecasting methodologies utilize classical machine learning algorithms to predict the electrical load. There is a need that big data platforms and parallel distributed computing are utilized to their potential in the available solutions. In this paper, the Apache Spark and Apache Hadoop are utilized as big data platforms for distributed computing in order to predict the load using available big data. In this paper, MLib, Spark library for machine learning algorithms, is utilized for distributed computing. Using MLib allows testing the classic regression algorithms such as linear regression, generalized linear regression, decision tree, random forest and gradient-boosted trees in addition to survival regression and isotonic regression. The obtained results show that Spark produces high accuracy while parallelizing the process of load forecasting in highly competent training and test times. Actual big data are used in the load forecasting process.

Quasi-Z source indirect matrix converter (QZS-IMC) has been proved to have abilities of voltage boost, current filtering, variable voltage, and variable frequency. The voltage gain of QZS-IMC depends on QZS network shoot-through duty ratio D , rectifier modulation ratio $m_{i}$ , and inverter modulation ratio $m_{o}$ . Their multiple combinations are able to achieve the same voltage gain, but determining which is the optimal has not been addressed. In addition, the size of D is associated with the duration of shoot-through events of the QZS network, voltage and current stresses of power devices, and the system power losse. This paper proposes to find the optimal operation curve of D based on the constrained optimization theory. Simulation and experimental results validate theoretical analysis, the proposed optimal control, and the power loss reduction of the QZS-IMC.

In this paper, a model-based current control (MBCC) approach with a compensating of dc-side inductor current ripple, active damping, and virtual time constant is proposed for single-phase grid-tied quasi-Z-source inverters with an LCL filter. The idea behind the ripple compensation is based on the inherent relationship between the ripple components of the dc-side inductor and capacitor voltages. It is shown that dc-side inductor current ripple can be compensated if the conventional simple boost control involving proportional-integral (PI) controllers is modified by subtracting the measured dc-side inductor voltage from the error signal of the first PI controller. Also, it is shown that the proposed MBCC causes the ac-side inverter current to track its reference in all circumstances. In addition, a virtual time constant is added to the control variable so that the dynamics of the ac-side inverter current can be adjusted as desired. Finally, in order to damp the LCL resonance, an active damping method is employed in the closed-loop system by modifying the ac-side reference inverter current. Experimental results are presented to show the validity and performance of the proposed control approach.

This paper proposes two second-order continuous-time algorithms to solve the economic power dispatch problem in smart grids. The collective aim is to minimize a sum of generation cost function subject to the power demand and individual generator constraints. First, in the framework of nonsmooth analysis and algebraic graph theory, one distributed second-order algorithm is developed and guaranteed to find an optimal solution. As a result, the power demand constraints can be kept all the time under appropriate initial condition. The second algorithm is under a centralized framework, and the optimal solution is robust in the sense that different initial power conditions do not change the convergence of the optimal solution. Finally, simulation results based on five-unit system, IEEE 30-bus system, and IEEE 300-bus system show the effectiveness and performance of the proposed continuous-time algorithms. The examples also show that the convergence rate of second-order algorithm is faster than that of first-order distributed algorithm.

Fault Detection and Classification (FDC) in Heating, Ventilation, and Air Conditioning (HVAC) systems is an important approach to guarantee the human safety of these systems. Therefore, the implementation of a FDC framework is the focus in this paper. The developed approach aims at reducing the energy needs for buildings and improving indoor environment quality. It merges the benefits of multiscale representation, Principal Component Analysis (PCA), and Machine Learning (ML) classifiers in order to improve the efficiency of FDC in heating systems. Firstly, a multiscale decomposition is used to extract the dynamics of the systems at different scales. The multiscale representation gives several advantages for monitoring heating systems generally driven by events in different time and frequency responses. Secondly, the multiscaled data-sets are then introduced into the PCA model to extract more efficient characteristics. Thirdly, the ML algorithms are applied to the extracted and selected characteristics to deal with the problem of fault diagnosis. The FDC efficiency of the developed technique is evaluated using a simulated data extracted from heating systems.