Search

This paper aims to estimate the three-phase voltage sensitivity matrix and the network topology of a low-voltage distribution network from smart meter data, which measures voltage magnitude, current magnitude, and power factor with a lead/lag indicator. The targeted networks are three-phase networks with single-phase loads. Understanding network sensitivity and topology is crucial for fault detection, unmetered load identification, and addressing Dynamic Operating Envelope problems. The problem is formulated as a constrained optimization problem to estimate both the three-phase voltage sensitivity and the low-voltage transformer voltage. The estimated voltage sensitivity is used to further identify the network topology, by implementing an enhanced Recursive-Grouping and Backtracking algorithm, as well as a candidate topology selection technique. The proposed method is tested on the 55-node European feeder and several synthetic networks. Compared to the state-of-the-art, the results show a four-fold improvement in the accuracy of voltage sensitivity estimation and substantially fewer mistakes in the topology estimates. The result underscores the efficacy of a three-phase network model and voltage angle approximation in enhancing estimation accuracy.

Demand response (DR) is the change in electric consumption by prompting consumers to change their normal consumption patterns in response to financial incentives. In various forms of DR, the contract-based DR is a framework to achieve a certain reduction by the contract with individual participants for their reduction. However, in practice, it is inevitable to encounter a default, implying that some of the participants fail contracted reduction owing to an unexpected event. This study develops a method to detect defaulting participants in contract-based DR. The method iterates two steps: estimating the most suspicious participant and inspecting its smart meter. By utilizing block-sparse structures in the detection problem, it can exactly determine the defaulting participants with a small number of inspections. The performance is evaluated by simulation, which indicates that our method has much higher performance than the conventional method in the sense that the number of inspections approximately 96.8% fewer than that of the conventional one. • An algorithm is developed to detect defaulting participants in demand response. • Four special properties of contract-based demand response are extracted. • Four properties are captured as block sparsity of defaults and utilized in detection. • Our method improves detection performance by 96.8% over the conventional method. [ABSTRACT FROM AUTHOR]

• This research proposes a multifactor index system for evaluating the operating state of smart meters. • An improved combinatorial weighting method is used to estimate the weight of factors. • The research concludes that the proposed method can contribute to improve the efficiency of smart meter detection. As an important terminal of smart grid, smart meter has the functions of equipment health monitoring and power metering. However, it is difficult to quantify the impact of factors such as ex-factory parameters, historical failure rates, electromagnetic interference when conducting an operational condition assessment. Therefore, in view of the large number of influencing factors and the difficulty in quantifying the degree of influence, an improved multifactor evaluation model for the operating status of smart meters is constructed. Firstly, a multifactor index system for evaluating the operating state of smart meters is constructed by analyzing the working mechanism and operating characteristics of smart meters. Then, an improved combinatorial weighting method is proposed by combining the subjective weight and the objective weight. The combinatorial weighting method is used to estimate the weight of factors in the health status evaluation of smart meters, and the improved multifactor evaluation method is presented to evaluate the health status of smart meters. Finally, an example is given to verify and analyze the proposed method. Compared with the other three methods, this method can provide effective evaluation results, and help guide targeted maintenance or replacement to improve the efficiency of smart meter detection. [ABSTRACT FROM AUTHOR]

• A system-level distribution power system integrating energy source, EV consumers and smart meters is modeled for the EVs charging and discharging scenario, leading a new research perspective for studying EVs charging and discharging with consideration of the vehicle owners' information exchange and interactions in the power system framework. • A two-stage optimization problem with priority for EVs charging and discharging is solved with a hierarchical optimal method, which decompose the problem as high-priority load shifting problem and low-priority cost minimization problem. • A penalty-based consensus algorithm is proposed to solve the cost minimization problem, in which two equational constraints are split and two similar optimization sub-problems which can be solved by a consensus distributed algorithm, and the convergence analysis of the algorithm is given mathematically. In addition, this paper combines the hierarchical algorithm with receding horizon method to deal with a more practical situation, where the EVs arrive and leave randomly. In preparing for the widespread adoption of electric vehicles (EVs), this paper investigates a two-stage optimization problem with priority for EV charging and discharging scheduling, while taking into account power load stability, economic objectives, and technical constraints of the distribution network. To this end, a distribution power system integrating energy source and EV owners is modeled. Considering the disturbance or even loss of the communication link, a periodical connected topology is adopted for information exchange among EV owners with the deployed smart meters, which is substantially different from the existing works. The main challenges of the problem are that the information flow among EV owners is time-varying and asymmetric, and each EV owner's demand should be satisfied throughout the whole scheduling horizon. A novelty distributed hierarchical optimal algorithm, combined with iterative water-filling-based method and penalty-based consensus method, is proposed for the two-stage optimization problem with priority. The convergence and optimality of the algorithm are presented mathematically. In addition, the proposed method combined with receding horizon optimization is performed for EVs random arrival and departure scenarios in the system. Finally, we demonstrate advantages of the proposed algorithm in both theory and simulation. [ABSTRACT FROM AUTHOR]

• For different types of electricity theft has obvious characteristics. • The proposed method can detect specific types of electricity theft. • Real-time electricity theft detection without historical data. Electricity theft detection is very important for the economic benefits of power companies and the effectiveness of safe operation of power systems. At present, the traditional power theft detection method can only identify whether the user has power theft, but cannot perform rapid and accurate inspections for various types of power theft users. Aiming at the characteristics of medium-voltage users with large power consumption and regular power consumption, this paper proposed a method for detecting the type of power theft in medium-voltage distribution lines based on robust regression and convolutional neural network. Firstly, considering the existence of abnormal data due to factors such as communication delay interruption, a robust regression algorithm is used to reduce its impact and improve the accuracy of regression analysis. Secondly, the correction coefficient and error term of each user obtained by regression are taken as the characteristics of user stealing electricity, and input into the convolutional neural network model for training to complete the identification of stealing electricity type. Finally, the method is verified by simulation and measured data. The results show that under different disturbance conditions, the proposed method can accurately identify different types of power stealing behaviors, which can better assist on-site investigation, narrow the investigation scope and improve the verification rate. [ABSTRACT FROM AUTHOR]

This paper aims to estimate the three-phase voltage sensitivity matrix and the network topology of a low-voltage distribution network from smart meter data, which measures voltage magnitude, current magnitude, and power factor with a lead/lag indicator. The targeted networks are three-phase networks with single-phase loads. Understanding network sensitivity and topology is crucial for fault detection, unmetered load identification, and addressing Dynamic Operating Envelope problems. The problem is formulated as a constrained optimization problem to estimate both the three-phase voltage sensitivity and the low-voltage transformer voltage. The estimated voltage sensitivity is used to further identify the network topology, by implementing an enhanced Recursive-Grouping and Backtracking algorithm, as well as a candidate topology selection technique. The proposed method is tested on the 55-node European feeder and several synthetic networks. Compared to the state-of-the-art, the results show a four-fold improvement in the accuracy of voltage sensitivity estimation and substantially fewer mistakes in the topology estimates. The result underscores the efficacy of a three-phase network model and voltage angle approximation in enhancing estimation accuracy. • Estimating voltage sensitivity to current of low-voltage distribution network. • Estimation uses smart meter measurements only. • Transformer voltage is estimated simultaneously. • Approximating unknown phase angle and considering coupling effect on neutral line. • A fast Recursive-Grouping and Backtracking algorithm to find topology. [ABSTRACT FROM AUTHOR]

• The algorithm acknowledges the current cyber-physical infrastructure limitations. • Using voltage readings, the proposed method alleviates data privacy concerns. • Voltage interval estimations are possible from single grid measurement. • Algorithm performance is verified through Monte Carlo simulations. The increased complexity of electric distribution networks, entails higher requirements on the interoperability between the information and communication technology (ICT) infrastructure and the power system. In current literature, distribution network monitoring methods require reliable operation of the ICT infrastructure, and depend on load estimation accuracy in cases of unavailable measurements. In the low voltage (LV) distribution network, ICT infrastructure is sensible to noise and congestion, and load estimation is challenged by irrational consumer behaviour. Such limitation impede the implementation of currently proposed methods. The conditions are complicated further when considering the impact of emerging data privacy concerns that can prevent utilization of household consumption data. Therefore, this work proposes a novel method for voltage magnitude interval estimation by utilizing the information from existing data acquisition systems in LV distribution networks. The proposed method is based on knowledge of load connection, and estimates the entire radial feeder voltage magnitude conditions from measurements of voltage magnitudes from minimum one node. The proposed method can therefore be implemented on existing ICT infrastructure and effectively keeps the privacy of consumers consumption profiles. The ability of the proposed method to estimate the feeder voltage interval is illustrated by an example and evaluated using Monte Carlo simulations for different meter distribution and accuracy scenarios. [ABSTRACT FROM AUTHOR]

• The present study aimed to develop a new model of a smart PV inverter with novel control schemes. • The proposed LVRT control strategy improved the voltage drop caused by the line-to-line-to-ground fault. • The central control system changed the switching mode of the inverter in the islanded mode. This article proposes a central control system that communicates with both grid-tied and off-grid control systems to offer various control strategies for operating a smart photovoltaic (PV) inverter. The target is to connect two sets of PV panels and one set of battery storage unit to either a 440 V/60 Hz utility grid or to feed local loads at 380 V/50 Hz using a smart inverter. When the smart PV inverter is connected to the grid, on the one hand, it injects fixed and programmed active power into the grid under all operating conditions, both normal and critical conditions, and on the other hand, by reciprocal exchanging reactive power with the grid, it addresses balanced and unbalanced fluctuations of the grid voltage. Also, when the smart PV inverter is disconnected from the grid, it can operate as an islanded mode to feed time-varying three-phase loads. Nonlinear simulations of the time domain in MATLAB-Simulink software show that the set of control strategies offered for the smart PV inverter not only by controlling the active power leads to the sale of electricity to the grid, but also by controlling the reactive power problems on the grid side such as voltage drop as a result of single-line-to-ground (SLG) or line-to-line-to-ground (2LG) faults, balanced grid voltage fluctuations such as voltage sag-swell and unbalanced voltage conditions will be resolved. Also, variable three-phase loads are fed well in off-grid mode. [ABSTRACT FROM AUTHOR]

The increasing penetration of distributed energy sources demands higher observability of the distribution networks for better scheduling, operation, dispatching and control. However, the huge amounts of nodes in the distribution networks implies that any additional measuring instrument would be of huge costs burden for DSO. In this paper, we proposes a full-scale framework for line parameters, topology, phase label estimation and angle recovery utilizing smart meter measurements only for the non- μ PMU distribution networks. Firstly, we linearize the three-phase power flow model and induce the relationship of nodal voltage magnitudes and power injections. Then, a joint estimation method based on linear regression and topology characteristics is proposed to find line parameters, topology, and phase label. Finally, the nodal angle information is recovered by non-linear regression based on the estimation result. Simulations on both single-phase and three-phase networks are implemented. The results suggest that the proposed framework can provide an accurate estimation result with limited measurements. • Linearized power flow-based method to estimate line parameters without nodal angle. • It enables topology identification, phase labeling and nodal angle recovery. • The required data can be obtained from smart meters under current standards. [ABSTRACT FROM AUTHOR]

Short-term residential load forecasting is of great significance for the demand-side energy management of the grid and the home energy management of resident customers. The massive penetration of distributed renewable energy, especially small-scale solar photovoltaic (PV), in the residential sector urgently requires us to move from traditional load forecasting to net load forecasting. In recent years, deep learning techniques that can improve model forecasting performance have developed rapidly in the field of residential load forecasting. However, the opaque nature of deep learning makes its practical application very difficult. This paper proposes a Transformer-based probabilistic residential net load forecasting method that utilizes quantile regression to quantify uncertainty in future load demand. Meanwhile, to improve the interpretability of deep learning model, local variable selection network is developed to automatically select relevant features and provide feature-level explanations. Additionally, interpretable sparse self-attention mechanism is proposed to extract long-term temporal dependencies. Numerical experiments are carried out with data from real household smart meters. The results show that the proposed model outperforms other state-of-the-art forecasting models. In terms of point forecasting, compared with the most common deep time series forecasting model LSTM, the proposed model decreases by 21.3%, 27.3% and 20.9% On three point forecasting performance metrics. Compared with Vanilla Transformer, the proposed InterFormer decreases by 9.9%, 10.5% and 9.0% On three point forecasting performance metrics. In terms of probabilistic forecasting, compared with LSTM and Vanilla Transformer, the average pinball loss of the proposed model decreases by 26.2% and 17.0%, respectively. Additionally, and most importantly, our model provides users with explanations in terms of feature importance and temporal patterns. • A novel local variable selection network to integrate input features. • A new interpretable sparse self-attention mechanism. • A generative Transformer prediction module abandons the Encoder–Decoder architecture. [ABSTRACT FROM AUTHOR]

• Digital substation. • Centralized protection and control (CPC). • Protection functions implemented in a generic hardware platform. • Multicriteria Faulted Phase Selection algorithm. • Improved protection algorithm operation under renewables and conventional scenarios. The integration of devices based on renewable energies and power electronics makes it necessary to rethink some of the methods used for protection and control in distribution and transmission grids. The difference between short circuit current contributions from renewable and conventional generators may affect protection system operation. Fault phase selection is one of the elements to be analyzed under renewable generation scenarios due to its importance in protection functions such as distance protection. The digitization of substations presents new opportunities in control and protection systems, such as functionality increase, reliability improvement, cost reduction of the installation and capacity for remote maintenance. This paper proposes a multi-criteria fault phase selection algorithm developed to be implemented in digital substations. The proposed faulted phase algorithm combines the results provided by four independent protection algorithms and has been designed to be applied in transmission power systems. Depending on the electrical variables measured in each event, quality criteria are defined to consider the strengths and weaknesses of each independent algorithm. The proposed algorithm has been implemented and tested on the centralized protection and control platform called EPICS, obtaining very satisfactory results. [ABSTRACT FROM AUTHOR]

Highlights • Investigated the effect of data collection frequency on the early detection of voltage anomalies. • An anomaly detection algorithm is developed in MATLAB. • The data collection frequency has little effect on post-event analysis under normal operation. • At abnormal conditions, the data collection frequency has significant effects. • The shorter-duration anomalies can be detected effectively with more frequent data. Abstract Smart grids are required to respond quickly and efficiently to any types of electrical incidents. This is possible only if advanced monitoring and decision support tools are available for the operators to collect and analyze the real-time data from the entire power system. Sensors and smart meters become necessary for an effective smart grid operation, as they can play a significant role in measuring system parameters such as the temperature of the transmission line, power outages and power usage. These sensors allow communication between the generation side and user side to ensure full network observability. Hence, the volume of data which needs to be analyzed for more reliable electrical services has been increased. Upon the analysis of all the data collected, useful insights can be gained to bring benefits to the utilities, consumers and the related organizations. In this project, the effect of smart meter data collection frequencies on the early detection of short-duration voltage anomalies has been investigated. An anomaly detection algorithm that analyzes the voltages collected from smart meters is developed. The proposed anomaly detector checks the normality of the voltage data collected from the smart meters installed at the residential loads. To test the effectiveness of the frequent-and-large smart meter data, a smart grid model was simulated in MATLAB/Simulink and the model was tested under three different operating conditions. The efficacy of the anomaly detector with the smart meter data collected at different data collection intervals was compared under the three operating conditions. The investigation of the smart meter data clearly shows that the shorter-duration anomalies can be detected effectively with more frequent data and hence it reveals the potential of smart meters in the early detection of short-duration anomalies in smart grids. [ABSTRACT FROM AUTHOR]

Highlights • We design an attribute identity-based signcryption algorithm. • We improve the remote anonymous attestation protocol based on the designed algorithm. • A privacy protection scheme that can verify the meter's trustworthiness is proposed. • The specific implementation process of the scheme is given. Abstract In order to monitor and predict the status of the power grid effectively in real time, smart meters need to send the power information of users to the control center frequently. However, power consumption data may contain users' privacy, e.g., living habits and hobbies. Further, as the smart meter terminal's working environment is complex and diverse, and smart meters suffer from endless physical attacks and cyber attacks, so the trustworthiness of smart meters is under threat. Aiming at addressing the problems above, a privacy protection scheme for smart meters that can verify the terminal's trustworthiness based on trusted computing is proposed in this paper. We design an attribute identity-based signcryption algorithm, and improve the remote anonymous attestation protocol based on the algorithm, which on the one hand realizes privacy protection of smart meters with high efficiency, and on the other hand completes real time verification of the smart meters' trustworthiness. The specific implementation process of the scheme is given in this paper. The anonymity, privacy protection of the platform configuration, and unforgeability of the scheme are proved in theory. Corresponding experiments show that this scheme has higher computational efficiency than others. [ABSTRACT FROM AUTHOR]

For distribution network operators to make effective decisions about real-time applications, they should have complete knowledge of all system variables. However, measuring all variables is infeasible due to the large number of system buses and the consequently high cost of measurement devices. Network operators are thus in serious need of methods that can estimate system voltages and currents with only a few measuring devices. This paper presents a novel voltage, current, and power loss estimation technique for distribution networks characterized by a high level of distributed generation (DG) penetration. The proposed method is based on online measurements from smart meters (SMs) placed at a few selected locations in addition to the measurements from DGs production meters; the estimation is derived without any pseudo measurements. The ingenuity of the proposed technique is that the SM locations are dependent on the network topology only, which means that their locations remain unchanged regardless of penetration levels and/or DG injection points. The proposed technique also includes consideration of variations in X/R ratios and laterals. The developed algorithm was implemented and tested on three radial distribution feeders to show the capability of the proposed technique for estimation for balanced as well as unbalanced distribution networks. The results of a comparison with the actual load flow demonstrate the accuracy and effectiveness of the new technique. [ABSTRACT FROM AUTHOR]

Voltage regulation has been one of the major challenges for distribution network operators (DNOs) due to the integration of distributed energy resources in the recent years. Control approaches mitigating over-voltage (OV) generally require full network model, which for low-voltage distribution networks (LVDNs), can be inaccessible to DNOs. This paper presents a novel data-driven model-free control framework for improving voltage regulation in 4-wire 3-phase LVDNs. In the proposed approach, first, using smart meter data, the aggregated resistance and reactance matrices of the network are estimated. Then, these matrices serve as an input for the proposed optimal power flow (OPF) method to control the DERs' active and reactive power outputs. This method also lets the DNO adjust the fairness in prosumers' generation curtailment. The performance of the proposed approach is evaluated in various case studies conducted on a real-world low-voltage test feeder. The simulation results exhibit significant effectiveness in solving the OV problem associated with DERs. For example, in the studied network, the proposed approach limits the maximum voltage of the network to 1.1 pu, while without having control over the DERs output power, it can even overpass 1.17 pu during the peak generation period. In addition, utilizing DERs' reactive control capability reduces the amount of active power curtailment, and prosumers are able to collectively inject 5.66% more active power. • Model-free management of DERs. • Enabling DNO to find a trade-off between total generation and fairness. • Robustness against measurement noises. [ABSTRACT FROM AUTHOR]

Load forecasting is essential for the operation and planning of a utility company. Recent large-scale smart meter deployments enabled the collection of fine-grained load data and created opportunities for sensor-based load forecasting. Machine learning has achieved great successes in load forecasting; however, conventional machine learning techniques require data transfer to the cloud or another centralized location for model training. This not only exposes data to privacy and security risks but also, with a large number of smart meters, increases network traffic. Federated Learning (FL) has a potential to alleviate mentioned concerns by training a single ML model in a distributed manner without requiring participants to share their data. Consequently, this paper proposes FedNorm, a novel asynchronous FL approach for load forecasting with smart meter data. While most FL strategies are synchronous and require all clients to complete local training in each round of aggregation, FedNorm is asynchronous and aggregates updates without waiting for lagging clients. To achieve this, FedNorm measures the clients contributions considering similarities of local and global models as well as the loss function magnitudes. The experiments demonstrate that FedNorm achieves higher accuracy than seven state-of-the-art FL techniques. Moreover, experiments show that FedNorm converges in fewer communication rounds compared to other FL models. • Training traditional machine learning models requires sharing data with the server. • Sharing smart meter data exposes data to privacy and security risks. • Our federated learning approach, FedNorm trains ML models without sharing data. • In contrast to other federated learning (FL) approaches, FedNorm is asynchronous. • FedNorm achieves higher accuracy than seven state-of-the-art FL techniques. [ABSTRACT FROM AUTHOR]

• Using random forest to select typical user behavior characteristics and reduces the complexity of user behavior data. • Combine ensemble learning and imbalance data classification model. • Proposed weighted support vector description to solve electricity theft detection with very unbalance dataset. • Diverse experimental results show that the proposed algorithm has better feature characterization capability and better classification performance on high-dimensional and unbalanced datasets. Improving the detection rate of electricity theft users in smart grids is crucial to the safe operation of the power system and the economic efficiency of the grid. The traditional electricity theft detection relies on manual household inspection, which is costly and inefficient. With the application of smart meters, machine learning-based methods are gradually applied to detect electricity theft users. However, the training of the model is limited by the unbalanced electricity consumption dataset, and the performance of the model is limited in the process of processing high-dimensional data. Therefore, this paper adopts ensemble learning and imbalance data classification model to analyze the electricity theft detection problem in complex grid environment, and proposes an electricity theft detection model based on random forest and weighted support vector description. Firstly, the feature selection algorithm for mixed load data with statistical features based on random forest (FS-MLDRF) is used to extract the features that can best represent the customers' electricity consumption behavior, so as to reduce the complexity of user data. Finally, an electricity theft detection based on hybrid random forest and weighted support vector data description is constructed (HFR-WSVDD), and the SVDD classifier is trained with the features weighted by feature importance to improve the performance of power theft detection in the case of unbalanced data. The simulation experiment results on the real electricity dataset show that compared with existing feature selection algorithms and anomaly detection algorithms, the algorithms proposed in this paper have excellent feature representation capability and better classification performance. [ABSTRACT FROM AUTHOR]

To address the challenges that the decarbonization of the energy sector is bringing about, advanced distribution network management and operation strategies are being developed. Many of these strategies require accurate network models to work effectively. However, distribution network data are known to contain errors, and attention has been given to techniques that allow to derive improved network information. This paper presents a novel method to derive line impedance values from smart meter measurement time series, with realistic assumptions in terms of meter accuracy, resolution and penetration. The method is based on unbalanced state estimation and is cast as a non-convex quadratically constrained optimization problem. Both line lengths and impedance matrix models can be estimated based on an exact nonlinear formulation of the steady-state three-phase network physics. The method is evaluated on both the IEEE European Low Voltage Feeder (906 buses) and a real Belgian LV feeder, showing promising results. • Impedance values and operating conditions (state estimation) are jointly derived. • Accurate unbalanced impedance models are derived from smart meter measurements. • Impedance estimation is cast as a quadratic constrained optimization problem. • Constraints are added to impose structure of multi-conductor impedance matrices. • Steady-state network physics is exploited: impedance estimation is model-aware. [ABSTRACT FROM AUTHOR]

In recent years, technological progress in power electronics and in battery storage systems, along with the introduction of smart meters, have made possible the application of improvements in the control of power peaks for residential customers. The demand charge is an important component in the residential customer’s electricity invoice in Spain, which has increased by nearly 100% in the last few years, prompting the consumers to reduce their peak usage. This paper proposes a model which allows residential consumers to save a significant part of their invoice by lowering contracted capacity through the use of a battery energy storage system. Recent academic literature has already studied in-depth the benefits of load-shifting energy throughout the day in the context of time of use (ToU) tariffs, but, regarding power optimization, there is a long way to go; not much research has been conducted in this field. In this article, a model to calculate optimum battery size which reduces contracted power is proposed. The model starts with a given load curve from a typical consumer in Spain in a context of constant energy prices. This system could be financed by the end user directly or by an independent company which participates in the savings. [ABSTRACT FROM AUTHOR]

Knowledge of the currents is a key foundation for smart grid applications. However, knowledge of low voltage currents is generally poor. The new information streams from advanced metering infrastructure (AMI)/smart meters and the monitoring of distribution substations offer the opportunity of rectifying this. Unfortunately, often not all the smart meter readings will be available in real-time. For example, this situation will arise when older (non-compliant) smart meters do not have real-time reporting capabilities. This paper investigates how knowledge of the substation currents can be combined with the available real-time AMI/smart meter readings and the historical readings from the non-real-time meters, to estimate these missing values. It is found that the k-nearest neighbor weighted average approach performs best but that the gains over using simpler methods are relatively modest. [ABSTRACT FROM AUTHOR]

In distribution networks, the lack of measurement data is usually thought to be an inevitable bottleneck of conventional grid operation and planning. Recently, the availability of smart meters in the distribution network has provided an opportunity to improve the network observability. In medium-voltage (MV) distribution networks, there is an increasing demand to use aggregated smart meter data for the state estimation, instead of adopting pseudo-measurements with a low level of accuracy. However, the performance of an estimator requires good knowledge of the available measurements, in terms of both expected values and associated uncertainties. Therefore, this paper intends to firstly pave a new way of utilizing smart meter data gathered from the low-voltage (LV) feeders in a concrete and reliable manner. For the purpose of state estimation in MV distribution networks, smart meter data is to be processed through three steps: phase identification, data aggregation and uncertainty evaluation. The feasibility of the proposed method is verified on the IEEE European LV Test Feeder with a set of real-world smart meter data. Afterwards, the influence of the aggregated smart meter data on the three-phase state estimation are investigated on the modified IEEE 13-node test system and IEEE 34-node test system. Simulation results show that the effect of aggregated smart meter data on the accuracy of state estimators is dependent on both the accuracy level of the aggregated data and the measurement configuration in the network. Furthermore, the use of aggregated smart meter data is shown to be able to provide improved state estimation. [ABSTRACT FROM AUTHOR]

• A CNN is trained to deal with data imbalances, considering the fraud patterns can be predicted by the metered consumption. • A CNN is trained to deal with data imbalances, considering the fraud patterns can be predicted by the metered consumption. • Consumption patterns of normal and fraudulent cus-tomers are clustered to achieve higher DR and lower FPR. • A DNN classifier is developed to judge normal and suspicious customers, considering different fraud detection scenarios. Electricity fraud as a noteworthy portion of non-technical losses has continuously been a worldwide concern. On the other hand, once a fraud is detected, on-site inspection is necessary for final verification. Since inspecting all metering data is laborious and expensive, utilities constantly try to narrow the inspection to instances with a higher odds of fraud. The rise of smart meters has introduced AI-based methods to reduce the inspection's scope by leveraging consumption patterns. However, their performances are limited due to the insufficiency of malicious samples, known as data imbalance. In the following investigation, a two-stage deep-learning-based model is provided for detecting energy fraud using metering data. first, a deep architecture is developed to model suspicious behaviour of fraudulent customers; which is used to handle the problem of data imbalance by predicting possible theft scenarios for normal consumers. then, A DNN network is developed to distinguishes normal and suspicious consumers. Consideration use of clustering and deep learning, as well as concurrent use of consumption and livelihood data, has enabled the algorithm to understand customer's behaviour and not fall against non-malicious changes in consumption patterns. Evaluation of the following method on a repository with 6000 real-world metering data demonstrates its high performance. [ABSTRACT FROM AUTHOR]

• A new intermittent electricity theft attack is presented by repeating normal and fraudulent behaviors alternately. • Electricity feature engineering is conducted on the automatically divided time intervals. • A machine learning-based detection framework is proposed to detecting the intermittent electricity theft attack. • VBGMM is used to overcome the difficulty of parameter selection of detection threshold. The widespread installation of advanced metering infrastructure (AMI) brings convenience to applications including but not limited to load management and demand response. However, AMI is also at risk of electricity theft and non-technical loss. Using the smart meter data provided by AMI to dig out user electricity consumption behavior is an effective way to construct electricity theft detectors. In this paper, a new intermittent electricity theft attack behavior is presented which switches between committing electricity theft and honestly consuming electricity alternately to skillfully evade the existing detectors. Based on the assumption that the labels of intermittent adversaries are unavailable, a new machine learning-based detection framework is proposed to detect this attack. Initially electricity features are constructed based on time intervals divided by a numerical iteration method. Then light gradient boosting method (LightGBM) is used to classify the normal users and adversaries. Further, the disperse degree of users is designed for capturing the differences between the intermittent adversaries and others. A new 2D label set is then constructed by combining the predicted labels of LightGBM and the disperse degree. Finally, a variational Bayesian Gaussian mixture model is employed based on the 2D label set to sort the users into the normal users and adversaries visually. Results of the case studies show that the presented attack can evade state-of-the-art detectors but still gain high profits. In addition, the proposed machine learning-based detector outperforms state-of-the-art detectors on both persistent attacks and intermittent attacks. [ABSTRACT FROM AUTHOR]

• A new deep learning-based clustering method is proposed to deal with the large number of daily load curves. • A novel customer selection framework is proposed to identify the potential good candidates for demand response programs. • The proposed clustering method outperforms the classical K-means and two state-of-the-art deep clustering methods. • The customers selected by the proposed framework contribute to 59 % of total peak reduction with only accounting for 23 % of population. Demand response (DR) is regarded as a promising solution to the problem of renewable energy integration, while it remains one of the key barriers for DR to target the right customers, i.e. the customers with high potential in peak reduction. Exploring the electricity consumption behaviors of customers can help DR program practitioners evaluate customers' peak reduction potential. Clustering has been the most common approach to describe customers' consumption behaviors in the literature. However, the existing clustering methods suffer from either unscalability or poor performance when handling a large number of daily load curves. To overcome these issues, a new deep learning-based clustering method is proposed in this paper, which integrates both intra-cluster compactness and inter-cluster separation into the objective function of clustering for consideration. Based on the clustering results, a new customer selection framework is developed to identify the potential good candidates for DR programs, which takes the stability of consumption behaviors and the characteristics at peak hours into account. Case studies on a London dataset demonstrate the superiority of the proposed clustering method over classical K-means and two state-of-the-art deep clustering methods in terms of clustering validity indexes. Furthermore, the simulations performed on an actual price-based pilot show that the customers selected by the proposed customer selection framework contribute to 59 % of total peak reduction with only accounting for 23 % of population. [ABSTRACT FROM AUTHOR]

• False data injection attacks pretend to mislead power scheduling and routing strategies in internet of energy-based smart grids besides monetary purposes. • The conventional cyber defense systems cannot detect well-developed false data injection attacks since intruders alter and enhance attack methods considering the dynamic nature of the smart grids. • A Deep Q-Learning-based attack generation algorithm is able to create a wide range of possible false data injection attacks, including dynamic attacks. • A two-layer defense framework is needed to detect passive and active attacks simultaneously. The Internet of Energy envisions the next generation of smart grids as a highly interconnected network, including advanced metering infrastructures, distributed energy resources, and bidirectional communication systems. The open architecture of IoE-based smart grid results in manifold security concerns, especially the risk of False Data Injection Attacks. The attack may target the technical aspects of a system since fabricating the network's data misleads power scheduling and routing strategies and interrupts the healthy operation of the power system. Additionally, monetary motivation for the intruder sometimes is the main motivation. The conventional cyber defense strategies are unable to detect well-developed False Data Injection Attacks, particularly once the intruder takes advantage of a Deep Reinforcement Learning-based attack development framework that analyzes the dynamic nature of the smart grids. This paper primarily outlines various possible passive attacks using statistical methods. Then, a reinforcement learning-based intruder as an active attack generator is developed, initialized by modeled passive attacks. The attack generator algorithm can simulate the network environment and subsequently creates unclassified attacks. After creating a dynamic attacker, a multilayer defense framework is developed using Snapshot Ensemble Deep Neural Network and an adoptable Deep Auto Encoder network to detect known and unknown threats. Performance evaluations and a real-world simulation prove that the proposed framework can successfully detect both passive and active, where the accuracy and false positive detection rate of the developed framework are 98.82% and 97.42%, respectively. [ABSTRACT FROM AUTHOR]

• A novel data-driven approach to discover phase topology of low-voltage customers is proposed. • An up-to-date certainty about phase connection of each customer can be given thanks to Bayesian inference. • It obtains accuracies higher than 97 % in almost any case regardless of the Smart Meter penetration. • Prior knowledge of the grid improves the performance even if it contains some errors. Knowledge of customer phase connection in low-voltage distribution networks is important for Distribution System Operators (DSOs). This paper presents a novel data-driven phase identification method based on Bayesian inference, which uses load consumption profiles as inputs. This method uses a non-linear function to establish the probability of a customer being connected to a given phase, based on variations in the customer's consumption and those in the phase feeders. Owing to the Bayesian inference, the proposed method can provide up-to-date certainty about the phase connection of each customer. To improve the detection of those customers that are more difficult to identify, after obtaining the up-to-date certainty for all users, the consumption of those who have an up-to-date certainty above a certain percentile compared with the rest of the substation (those that are more likely to be correctly classified) is subtracted from the phase in which they are classified. The performance of the proposed method was evaluated using a real (non-synthetic) low-voltage distribution network. Favourable results (with accuracies higher than 97 %) were obtained in almost all cases, regardless of the percentage of Smart Meter penetration and the size of the substation. A comparison with other state-of-the-art methods showed that the proposed method outperforms (or equals) them. The proposed method does not necessarily require previously labelled data; however, it can handle them even if they contain errors. Having previous information (partial or complete) increases the performance of phase identification, making it possible to correct erroneous previous labelling. [ABSTRACT FROM AUTHOR]

This paper addresses the customer-phase identification problem in three-phase distribution grids including three-phase customers characterized by aggregated energy measurements. The proposed technique first solves a relaxed problem, in which the binary nature of the variables is ignored, which leads to a constrained, least-squares estimation, using as inputs the active and reactive energy readings provided by the smart meters, along with the energy delivered by each phase at the head of the feeder. With the estimated values of the decision variables, and their corresponding variances, a confidence-based selection technique is then applied for the sequential assignment of the customer with the highest joint probability of being connected to one of the three phases but not to the other two. The performance of the proposed procedure is assessed with five different scenarios in terms of accuracy for increasing number of loads and measurement errors. The robustness of the algorithm is additionally tested in the presence of model errors, and its performance is compared to that of existing methods. • A novel approach is proposed for customer-phase identification in distribution grids. • The proposed method iteratively solves an equality-constraint least squares problem. • This technique deals with three-phase clients with only aggregated energy readings. • Confidence levels of single-phase consumptions being associated to a phase are used. • The results obtained show higher success rates compared to other published methods. [ABSTRACT FROM AUTHOR]

• A novel dynamic time warping sampling method is proposed to overcome the class imbalance problem. • A Bayesian convolutional neural network is developed for the household characteristic estimation which can capture uncertainties and provide the confidence level. The knowledge of household characteristics can help energy providers carry out more personalized demand-side management programs. Obtaining such information through surveys is costly and time-consuming in practice. This paper proposes a novel estimation method for household characteristics with uncertainties using the residential electricity consumption data. To alleviate the class imbalance problem in the dataset, a dynamic time warping sampling (DTWS) method is proposed to generate synthetic data for the minority class. To overcome the problem that the existing methods for identifying household characteristics cannot provide the confidence level of the results, a Bayesian convolutional neural network (BCNN) model is developed for feature extraction and characteristic identification with uncertainties. These quantified uncertainties can be regarded as a measure of confidence and can be used to target customers more effectively for energy efficiency and demand response programs. The effectiveness of the proposed model is validated by experiments on ground truth data. [ABSTRACT FROM AUTHOR]

In Low Voltage (LV) distribution networks, the high volatility of distributed photovoltaic (PV) generation has a severe impact on the variation of operation indices, in steady state conditions. During periods with high PV injection and low demand, LV feeders are more and more subject to overvoltage events and temporary PV units’ cut-offs. As a result, the delivered power quality is affected and network operational expenses increase. Moreover, the income of the PV owner is decreased due to the loss of generated energy. For efficiently addressing such operational issues, long term observability analytics of the LV network are required. Distribution System Operators (DSOs) currently deploy such studies in a deterministic manner, focusing on “worst-case” hypothesis, without considering the uncertainty of nodal power injection and consumption. This approach can lead to over restrictive decisions and costly technical solutions. For refining DSO strategies to the variability of network states, probabilistic methods are highly recommended. In this context, this paper presents a Monte-Carlo (MC) framework that simulates the steady operation of the LV network by elaborating user-specific smart metering (SM) measurements. The presented framework integrates a complete three-phase power flow algorithm that can analyse most possible LV network configurations, balanced and unbalanced, considering nodal power injections and consumptions as random variables of each network state. Such unbalanced power flow algorithms had not up to now been linked with probabilistic analysis using network-specific SM readings. For demonstrating the interest of the proposed framework, the latter is used to simulate several configurations in an existing LV feeder with high PV integration and SM deployment. [ABSTRACT FROM AUTHOR]

The ongoing deployment of Distributed Energy Resources, while bringing benefits, introduces significant challenges to the electric utility industry, especially in the distribution grid. These challenges call for closer monitoring through state estimation, where real-time topology recovery is the basis for accurate modeling. Previous methods either ignore geographical information, which is important in connectivity identification or are based on an ideal assumption of an isolated sub-network for topology recovery, e.g., within one transformer. This requires field engineers to identify the association, which is costly and may contain errors. To solve these problems, we propose a density-based topology clustering method that leverages both voltage domain data and the geographical space information to segment datasets from a large utility customer pool, after which other topology reconstruction methods can carry over. Specifically, we show how to use voltage and GPS information to infer associations within one transformer area, i.e., to identify the meter-transformer connectivity. To give a guarantee, we show a theoretic bound for our clustering method, providing the ability to explain the performance of the machine learning method. The proposed algorithm has been validated by IEEE test systems and Duquesne Light Company in Pittsburgh, showing outstanding performance. A utility implementation is also demonstrated. • DERs introduce challenges to the distribution grid. • Such challenges call for meter-transformer mapping. • Using voltage and GPS information to identify the meter-transformer connectivity. • Theoretical guarantee for the robustness of the algorithm. • Outstanding performance on IEEE test systems and a utility in Pittsburgh. [ABSTRACT FROM AUTHOR]

• Microservice-based edge computing apparatus in smart distribution transformer area. • Computing resource demand and supply model. • Sequential logic-based evolution algorithm. • Increase of vertical elastic rate improves computing resource utilization and delay. • Effectiveness in both stochastic and concurrent simulation scenarios. With the development of the Internet of Things technology in the smart grids, terminal devices such as advanced sensors and smart meters present a trend of massive access in distribution network, which poses great challenges to the information perception, analysis and processing ability of distribution automation system. The microservice-based edge computing apparatus is a novel technology in the smart distribution transformer area to improve the adaptability of the distribution automation system. Research on modeling methods and computing resource capacity planning methods for the microservice-based edge computing apparatus has become an urgent and significant research topic in the modern electrical power system. Therefore, this paper proposes a configuration method of computing resources for the microservice-based edge computing apparatus in the smart distribution transformer area. The suitability of the microservice technology and edge computing apparatus is analyzed, and a model of microservice-based edge computing apparatus is proposed. Inspired by power demand and supply in the power system, a computing resource demand and supply model is proposed to represent the computing resource demand of power Internet of Things services and the computing resource supply of the microservice-based edge computing apparatus. Furthermore, a sequential logic-based evolution algorithm is designed to determine the configured capacities of computing resources for the microservice-based edge computing apparatus considering multi-aggregated service flows. Finally, simulation results validate the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

This paper analyzes the impacts of the first wave of COVID-19 (March 2020 - September 2020) on the electricity demand of different types of consumers in Chile, including residential, commercial, and industrial demand. We leverage data from 230 thousand smart meters of residential and commercial consumers in 32 communes of Santiago (the capital city of Chile), which allows us to investigate the evolution of their demands with an hourly temporal resolution. Additionally, we use demand data of large industrial consumers provided by the Chilean system operator to study the impact of the pandemic on different economic sectors. This paper demonstrates that the COVID-19 pandemic, and the associated containment measures, have featured a drastically different impact on the various types of consumers in Chile. In particular, we show that the demand of residential consumers has increased throughout the first wave, even when we isolate the effects of the pandemic from those related to weather. Furthermore, we study how these effects change in different communes of Santiago, contrasting our findings with the socio-economic levels of the population. In effect, we find different demand response patterns depending on the socio-economic background of consumers. We also show that commercial demand has significantly declined due to the containment measures implemented and that the hospitality and construction economic sectors have been the most affected in the country. • The impacts of the COVID-19 on the electricity demand in Chile are quantified. • Demands from residential, commercial, and industrial sectors are studied. • The impacts significantly depend on the type of consumers. • The impacts also depend on the socio-economic level of consumers. • Hospitality and construction are the most affected economic sectors. [ABSTRACT FROM AUTHOR]

• According to the authors' knowledge, this article is one of the first to report Nontechnical Loss detection using Contrastive Predictive Coding. The machine learning algorithm introduced within this article has the following advantages: • It is found that the ND-CP algorithm pre-trained by unlabeled data is at least 14.6% higher on AUC and F1 indexes compared to other algorithms without pre-training. • The results suggested that in comparison with current machine learning methods in NTL detection, the ND-CP algorithm search achieved a more encouraging results (measured with accuracy, ROC-AUC and F1 scores indexes). • ND-CP algorithm achieved the same performance as the Wide & Deep CNN algorithm with only one third of the training samples. This can address the scarcity of labeled data for many applications. Non-technical losses cause substantial commercial concerns to distribution network operators (DNOs). 80% of NTLs are related to electricity theft, which contains various high-techs and is increasingly difficult to detect. Advanced metering infrastructure (AMI) has enabled supervised machine learning (ML) to detect the NTLs, which significantly improved the detection rates. A further advance in ML type of methods requires sufficient labeled datasets, which is usually not available. To address this, this article proposes a self-supervised detection method that extracts long-term consumption patterns to detect fraud in low-voltage networks, known as NTL detection contrastive predictive coding (ND-CP). Smart meter data sequences are fed into a one-dimensional convolutional neural network (1D-CNN) first. The gated recursive unit (GRU) data is then used to extract global information. After that, the output of the prediction from the GRU model is used to construct positive and negative sample pairs for contrastive learning. Eventually, a single-layer neural network classifier for detection is trained using the long-term features extracted by ND-CP. Experiments are conducted with real electricity consumption data to verify the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

Smart meter (SM) deployment in the residential context provides a vast amount of data that allows diagnose the behavior of household inhabitants. However, the conventional methods to analyze household-consumption load profiles based on time-series techniques, such as Fourier and Wavelet transform, have problems with nonlinear and nonstationary processes. This paper presents a methodology to perform a comprehensive analysis of consumption load profile features based on the detection of oscillation modes in the time–frequency domain for off-line systems. The methodology is based on the Hilbert–Huang transform (HHT) that evaluates the instantaneous frequency (IF) using the empirical mode decomposition (EMD) and some of the variants of this standard technique such as ensemble empirical mode decomposition (EEMD) and complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN). The study presented in this paper compares the accuracy of different techniques by conducting a case study of two households in Spain. Nonetheless, there is a trade-off between the computational burden and accuracy. Furthermore, the methodology reveals the importance of the data sampling frequency (temporal data granularity) for accurate characterization of the load profile. Some household loads produced oscillation modes that became increasingly ill-suited for resolutions of 1 min or higher. [ABSTRACT FROM AUTHOR]

Load forecasting is essential for energy management, infrastructure planning, grid operation, and budgeting. Large scale smart meter deployments have resulted in ability to collect massive energy data and have created opportunities in sensor-based forecasting. Machine learning (ML) has demonstrated great successes in sensor-based load forecasting; however, when prediction is needed on a smart meter level, typically a single model is trained for each smart meter. With a large number of meters, this becomes computationally expensive or even infeasible. On the other hand, with conventional ML, training a single model for several smart meters requires participants to share their data with the central server. Consequently, this paper proposes federated learning for load forecasting with smart meter data: this strategy enables training a single model with all participating smart meters without the need to share local data. Two alternative federated learning strategies are examined: FedSGD, which performs one step of gradient descent on client before merging updates on the server, and FedAVG, which carries out several steps before the merging. Specifically, residential consumers are diverse what makes training a single model challenging as load profiles vary across consumers. The results show that the FedAVG achieves better accuracy than FedSGD while also requiring fewer communication rounds. Comparing to individual models for each meter and a single central models for all meters, FedAVG achieves comparable or better accuracy. • Traditional machine learning requires sharing data with the server for training. • With federated learning, we train the forecasting model without sharing local data. • FedAVG enables one load forecasting model for diverse residential consumers. • FedAVG achieves better accuracy than FedSGD and conventional ML models. • FedAVG achieves high accuracy even for clients that did not participate in training. [ABSTRACT FROM AUTHOR]

Load balancing is one of the most widely used techniques to reduce losses and improve service quality in low-voltage networks. Many methods have been proposed, including the customer phase reassignment, which has multiple advantages. However, its application represents a real challenge if the data availability is limited (only a few variables from some nodes are measured). This paper presents a new method for load balancing on a three-phase feeder by switching the phase connection of single-phase customers. It is based on Deterministic Crowding, a variant of Genetic Algorithms with few and easily adjustable parameters. In contrast to related works, we establish an imbalance target value and minimize the number of customers required to switch. The proposed method requires the phase current profiles from only a percentage of the customers, reducing the number of smart meters necessary and the related investment. The method was validated using real data, analyzing different scenarios with different percentages of measured customers compared on the basis of imbalance index, neutral current, and transformer lifespan reduction. Results show a significant improvement in load balancing with few changes, even when only 30% of customers are measured. • Phase reallocation methods require less maintenance and provide long-term solutions. • The proposed method can operate even if only a low percentage of customers are measured. • The proposed method maintains the balance for a long time without new manual phase changes. • The number of phase changes is an important aspect when applying a load balancing method. • A balance improvement is achieved even if only 30% of customers are measured. [ABSTRACT FROM AUTHOR]

• The proposed decentralized robust state estimation (DRSE) technically integrates multi-source data with different timescales for hybrid AC/DC distribution system state estimation. • A slow timescale smart meters aided DNN model is developed to extract hidden system statistical information and allow deriving nodal power injections that guarantee the network observability. • The proposed DRSE is executed in a decentralized manner with the detailed converter model for achieving efficient SE and accommodating flexible operating modes. • A linear WLAV-based formulation for AC and DC regions is derived that achieves higher computational efficiency and offers robustness of suppressing bad data automatically. Hybrid AC/DC distribution systems are becoming a popular means to accommodate the increasing penetration of distributed energy resources and flexible loads. This paper proposes a decentralized robust state estimation (DRSE) method for hybrid AC/DC distribution systems using multiple sources of data. In the proposed decentralized implementation framework, a unified robust linear state estimation model is derived for each AC and DC regions, where the regions are connected via AC/DC converters and only limited information exchange is needed. In this context, estimation accuracy may be suffering due to linearization. To enhance the estimation accuracy, a deep neural network (DNN) based on the smart meter data is used to extract hidden system statistical information and allow deriving nodal power injections that keep up with the real-time measurement update rate. This provides the way of integrating smart meter data, SCADA measurements and zero injections together for state estimation. Simulations on two hybrid AC/DC distribution systems show that the proposed DRSE has only slight accuracy loss by the linearization formulation but offers robustness of suppressing bad data automatically, as well as benefits of improving computational efficiency. [ABSTRACT FROM AUTHOR]

• Harmonic distortion can be forecasted with no specialized metering device. • Demand response meters can be used for power quality monitoring. • Artificial intelligence helps monitoring power quality at low voltage networks. • Utilities will increasingly have to cope with harmonic distortion in a near future. This paper introduces a methodology to forecast voltage total harmonic distortion (THD) at low voltage busbars of residential distribution feeders based on the data provided by a limited number of smart meters. The methodology provides relevant power quality indices to system operators using only the existing monitoring infrastructure required for demand response operation. Different algorithms for voltage THD forecasting are implemented, including artificial neural networks, and their performance is tested and compared. The necessary coverage of smart meters for the acceptable accuracy of the estimated THD is also established. The estimation algorithms are validated considering probabilistic demand load model developed based on typical harmonic injections of household devices obtained from measurements and using a typical European low voltage test-feeder with 471 residential consumers. [ABSTRACT FROM AUTHOR]

• A novel smart metering infrastructure has been improved with multiple measurement. • The proposed system has been comprised with multi-channel communication. • A high performance and reliable microcontroller has been used to operate entire system. • The wireless metering infrastructure has been improved at 2.4 GHz transceiver. • Graphical user interface and database management have been coded with C# The distributed generation (DG) and the integration of renewable energy sources (RESs) have leveraged the improvement of the smart grid for a few decades. The smart home energy management systems (HEMSs) have also attracted attention in increasing the use of information and communication technologies in the context of smart grid applications. This paper proposes a smart metering infrastructure with DC and AC analog front ends (AFEs), wired and wireless communication interfaces, and remote monitoring software coded in Visual C#. The implemented metering device has been comprised of isolated measurement transducers for detecting the current and voltage rates of sources and various types of loads either in a microgrid or smart home application. The signal conditioning tier has been comprised of highly accurate and sensitive operational amplifiers, and signal processing has been accomplished by a PIC microcontroller which is capable to interface USB and ZigBee based wireless communication protocols. The processed measurement data have been transmitted with wireless communication, and a graphical user interface (GUI) software has been coded at the receiving end host computer. The measurement case studies have been experimentally performed in microgrid and HEMS infrastructures to validate the accuracy of the proposed smart metering device and software operated at the host computer. [ABSTRACT FROM AUTHOR]

• A new price adjustment strategy is constructed by the quadratic function. • The strategy is proposed to monitor the users' power consumption loads difference. • We adjust the price when the predicted load difference value exceeds the boundaries. • The new strategy is superior to linear function price adjustment in tests. Based on the data collected from smart meters, electricity pricing models can be developed to balance power supply and demand in time slot and obtain the optimal consumption loads and prices. However, in real life, users' reserved consumption requirement loads sometimes deviate significantly from the optimal consumption loads obtained from models, which results in overloaded power systems or even power cuts. To address this issue, an engineering process control strategy has been proposed in this paper to minimize the difference between the optimal and the users' reserved consumption requirement loads. We proposed an exponential weighted moving average model to predict the load difference in future time slots, and also developed a novel quadratic function based demand response mechanism to adjust the power price for power providers. The demand response mechanism can be used to adjust the price in the future time slots when the predicted demand exceeds the upper or lower boundary. Simulation results indicate that the quadratic function adjustment strategy has excellent performance in a practical power market in Singapore. Compared with the linear function based adjustment method, the proposed quadratic function based adjustment method decreases the adjustment times and standard errors of residuals, and increases the social welfare and power suppliers' profits under the same boundary conditions. In addition, the performance of the proposed strategy demonstrated its competency in peak-cutting and valley-filling and balancing energy provision with demands. [ABSTRACT FROM AUTHOR]

• An optimal scheduling scheme for smart home electricity is proposed. • The smart home load models are established under two kinds of demand response. • The user's privacy is protected by electric vehicle to blur power consumption. The implementation of demand response can not only ensure the stability of power system, but also reduce the cost of user's power consumption. However, the real-time, fine-grained bi-directional communication between smart meters of smart homes and the power control center may lead to the privacy leakage of users. Aiming at the problems above, an optimal scheduling scheme for smart home electricity considering demand response and privacy protection is proposed. First of all, the smart home load models are established under two kinds of demand response, and then the BAS algorithm is used to get the lowest cost of electricity consumption. In addition, the user's privacy are protected with the help of energy storage of electric vehicle to blur user's electricity consumption data, and a privacy quantification method with adjustable weight is given. The simulation results show that the scheme can significantly reduce the user's electricity cost by arranging household electricity reasonably in response to power grid DR, the users can also adjust the value of privacy protection weight α to meet their privacy and cost requirements. [ABSTRACT FROM AUTHOR]

• The proposed methodology allocates aggregator units in low voltage distribution networks. • The units are allocated considering the minimization of their installation costs. • The proposal can allocate the aggregators where they can better characterize the net load curves. • The proposal improves the characterization of the power losses in the aggregator coverage area. The aggregator units are essential in implementing smart grids since they are responsible for connecting smart meters installed on the consumers and ensuring exchanging information with the utility control centers. This paper presents a methodology to allocate these aggregators in low voltage distribution systems with solar photovoltaic generation. The proposed method is composed of two stages. In the first one, the aggregators are allocated considering the minimization of the distance to the remote terminal unit and their installation costs using a method based on Particle Swarm Optimization. In the second stage, the smart meters are connected to each allocated aggregator using a graph method. These connections are performed, taking into account a function, which depends on the distance between the meters and the load density within the aggregator coverage area. The proposed methodology was tested using a 400 V low voltage distribution system with 45 residential loads and 12 photovoltaic generators. The results showed that the proposal allocates aggregators at points where the consumer groups' load curves into the aggregator coverage area are better characterized. The values of the average voltage profile and the system's power losses in each aggregator coverage area presented an excellent match with values obtained with the test system without installing aggregators. From this characterization, it is possible to get useful information for operational planning and expansion planning of low voltage networks with photovoltaic generators by aggregation areas. [ABSTRACT FROM AUTHOR]

• A hybrid demand modeling framework for an electricity retailer is investigated. • A learning method based on deep learning is proposed into the bidding strategy problem. • The proposed model is able to forecast short-term load on the grid. • Optimal dynamic pricing for a retailer is obtained via proposed technique. • The efficiency of the proposed model is verified by several simulation cases. The main purpose of this study is to support a retail electric provider (REP) to make the best day-ahead dynamic pricing decisions in a realistic scenario. These decisions are made with the aim of maximizing the profit achieved by the REP under the assumption that mixed types of customers with different behaviors in the electricity market are considered. While some of the customers have installed smart meters with an embedded home energy management system (HEMS) in their home, others do not participate in the demand response (DR) programs. For this purpose, a bi-level hybrid demand modeling framework is proposed. It firstly uses an optimal energy management algorithm with bill minimization in order to model the behavior of customers with smart meters. Then, using a customers' behavior learning machine (CBLM), the behavior of other groups without smart meters is modeled. Therefore, the proposed hybrid model cannot only schedule usage of home appliances to the interests of customers with smart meters but can also be used to understand electricity usage behavior of customers without smart meters. The proposed model includes a stacked auto-encoder (SAE), one of the deep learning (DL) methods suitable for real-valued inputs, and adaptive neuro-fuzzy inference system (ANFIS). Based on the established hybrid demand model for all customers, a profit maximization algorithm is developed in order to achieve optimal prices for the REP under relevant market constraints. The results of the case studies confirm the applicability and effectiveness of the proposed model. [ABSTRACT FROM AUTHOR]

• A deterministic optimization problem to solve the voltage sag estimation problem. • Utilization of the smart metering capabilities to solve power quality problems. • Estimation of residual voltages by faults, even at points along a power line. • Use of sparse nature of voltage sag measurements to improve the estimation. This paper presents a new formulation for voltage sag state estimation based on compressive sensing (CS) theory. In a modern context, voltage sags can deteriorate the reliability, stability, and security of current networks and future smart grids. Several studies have focused on voltage sag estimation (VSE), but prefault currents, nonzero fault impedances, and unbalances are issues that can lead to inaccurate results. To overcome these limitations, we proposed a novel ℓ1-based voltage sag state estimator (ℓ1-VSSE). Only voltage measurements and a limited number of power quality meters with synchronization capabilities are considered. Sensing matrices of residual voltages per unit and a solution of a ℓ1-regularized least square problem using convex optimization are the innovative characteristics of the proposed estimator. ℓ1-VSSE efficiency has been validated by using three unbalanced distribution test systems: IEEE 34-bus, IEEE 123-bus, and EPRI-Ckt7. The results confirm the estimator's robustness to the different faults, voltage regulation, and loading profiles. [ABSTRACT FROM AUTHOR]

• A blind selection method for a threshold-based event identification. • Evaluation of unsupervised learning using GMM for energy disaggregation. • Identification of individual power levels in the data and constructing appliance models without any user intervention. • Comparison with other state of the art methods. • Evaluation of GMM based technique against a mean-shift clustering approach. • The results of the disaggregation are verified on a private dataset as well as public dataset. Non-intrusive loading monitoring (NILM) provides a smart solution to the problem of electrical energy monitoring of households at the appliance level. In blind disaggregation, the power level of each appliance is not known a priori. In this paper, we propose an event-based blind disaggregation algorithm that uses Gaussian mixture models (GMM) for clustering to automatically detect two-state appliances from the aggregate data. The benefit of using Gaussian mixture models over other clustering methods is that they can automatically learn the statistical distributions present in the data. This is beneficial, especially when the appliances have similar power consumptions. Since Gaussian mixture models do not determine the number of clusters automatically, we use Bayesian information criteria (BIC) to determine the number of clusters. The blind disaggregation method is tested with data from a real house collected by a smart meter which samples the aggregate consumption at 3.4 kHz and also from Reference Energy Disaggregation Dataset (REDD) public data, sampled at a frequency of 1 Hz. We compared the performance of our algorithm with other unsupervised methods and found comparable performance. We also compared the Gaussian mixture model with mean shift clustering in a blind disaggregation. We saw an improved performance by using Gaussian mixture models instead of mean-shift clustering in various accuracy measures. [ABSTRACT FROM AUTHOR]

• The multi-dimensional correlation characteristics of consumers in LVDN are deduced. • Multi-dimensional calibration based on voltage characteristics in LVDN are developed in CPI. • Multiply CPI processes are designed for different smart meter incomplete ratios. As results of poor communication quality, human error, electricity theft and other reasons, the obtained power consumption data of consumers from smart meters is incomplete and cannot fully reflect the power consumption of low-voltage distribution network (LVDN), which affects the recognition performance of the existing power-based and current-based consumer phase identification (CPI) methods. To tackle the limitation under incomplete data condition, this paper develops multi-dimensional calibration in CPI based on voltage characteristics in LVDN. Firstly, multi-dimensional correlation characteristics of consumers in LVDN are deduced to provide theoretical foundation, including the correlation characteristics among consumers and that between consumers and low-voltage buses of distribution transformer. Then, multiply CPI processes are designed for different smart meter incomplete ratios to increase the algorithm's robustness to incomplete data. In which, a consumer clustering method based on correlation characteristics among consumers is developed to decrease the variables in the identification process and a calibration method combined location index and the correlation characteristics among consumers is proposed to correct the phase connectivity of users far away low-voltage buses. Finally, the performance of the proposed algorithm is verified on a real-world LVDN in Guangdong. The comparison analysis between the proposed method and other published methods and the impact of the threshold coefficients on the identification accuracy are also investigated. Further, the application results of the proposed method in pilot LVDNs are presented. The results indicate that the proposed method achieves higher CPI accuracy than other published methods when the obtained power consumption data of consumers is incomplete and has good applicability in practice. [ABSTRACT FROM AUTHOR]