Deep neural networks have revolutionized many machine learning tasks in power systems, ranging from pattern recognition to signal processing. The data in these tasks is typically represented in Euclidean domains. Nevertheless, there is an increasing number of applications in power systems, where data are collected from non-Euclidean domains and represented as graph-structured data with high dimensional features and interdependency among nodes. The complexity of graph-structured data has brought significant challenges to the existing deep neural networks defined in Euclidean domains. Recently, many publications generalizing deep neural networks for graph-structured data in power systems have emerged. In this paper, a comprehensive overview of graph neural networks (GNNs) in power systems is proposed. Specifically, several classical paradigms of GNNs structures (e.g., graph convolutional networks) are summarized, and key applications in power systems, such as fault scenario application, time series prediction, power flow calculation, and data generation are reviewed in detail. Furthermore, main issues and some research trends about the applications of GNNs in power systems are discussed.

A Review of Graph Neural Networks and Their Applications in Power Systems

Load forecasting is a vital part of smart grids for predicting the required electrical power using artificial intelligence (AI). Deep learning is broadly used for load forecasting in the smart grid using the artificial neural network (ANN). Generally, computing the deep learning in the smart grid requires massive data aggregation or centralization and significant computational time. This paper presents a survey of deep learning-based load forecasting techniques from 2015 to 2020. This survey discusses the studies based on their deep learning techniques, Distributed Deep Learning (DDL) techniques, Back Propagation (BP) based works, and non-BP based works in the load forecasting process. Consequent to the survey, it was determined that data aggregation dependency would be beneficial for reducing computational time in load forecasting. Therefore, a conceptual model of DDL for smart grids has been presented, where the HSIC (Hilbert-Schmidt Independence Criterion) Bottleneck technique has been incorporated to provide higher accuracy.

/pdf/hsic-bottleneck-based-distributed-deep-learning-model-for-2dc43gdryl.pdf

HSIC Bottleneck Based Distributed Deep Learning Model for Load Forecasting in Smart Grid With a Comprehensive Survey

The current electric power system witnesses a significant transition into Smart Grids (SG) as a promising landscape for high grid reliability and efficient energy management. This ongoing transition undergoes rapid changes, requiring a plethora of advanced methodologies to process the big data generated by various units. In this context, SG stands tied very closely to Deep Learning (DL) as an emerging technology for creating a more decentralized and intelligent energy paradigm while integrating high intelligence in supervisory and operational decision-making. Motivated by the outstanding success of DL-based prediction methods, this article attempts to provide a thorough review from a broad perspective on the state-of-the-art advances of DL in SG systems. Firstly, a bibliometric analysis has been conducted to categorize this review’s methodology. Further, we taxonomically delve into the mechanism behind some of the trending DL algorithms. We then showcase the DL enabling technologies in SG, such as federated learning, edge intelligence, and distributed computing. Finally, challenges and research frontiers are provided to serve as guidelines for future work in the futuristic power grid domain. This study’s core objective is to foster the synergy between these two fields for decision-makers and researchers to accelerate DL’s practical deployment for SG systems.

/pdf/deep-learning-in-smart-grid-technology-a-review-of-recent-hmqi4c11lh.pdf

Deep Learning in Smart Grid Technology: A Review of Recent Advancements and Future Prospects

Nowadays, decentralized microgrids (DC-MGs) have become a popular topic due to the effectiveness and the less complexity. In fact, DC-MGs resist to share their internal information with the distribution system operator (DSO) to protect their privacy and compete in the electricity market. Further, lack of information sharing among MGs in normal operation conditions leads to form a competitive market. However, in emergency operation conditions, it results numerous challenges in managing network outages. Therefore, this paper presents a hierarchical model consisting of three stages to enhance the resilience of DC-MGs. In all stages, the network outage management is performed considering the reported data of MGs. In the first stage, proactive actions are performed with the aim of increasing the network readiness against the upcoming windstorm. In the second stage, generation scheduling, allocation of mobile units and distribution feeder reconfiguration (DFR) are operated by DSO to minimize operating costs. In the final stage, the repair crew is allocated to minimize the energy not served (ENS). Uncertainties of load demand, wind speed and solar radiation are considered, and the effectiveness of the proposed model is investigated by integrating to the 118-bus distribution network. Finally, the results of the simulation indicate that DFR and proactive actions decrease the ENS by 19,124 kWh and 4101 kWh, respectively. Further, the sharing of information among MGs leads to a 48.16% growth in the supply service level to critical loads, and consequently a 3.47% increase in the resilience index. • Presenting a hierarchical framework for distribution system resilience enhancement. • Increasing network readiness through performing proactive actions. • Sending mobile units to sensitive nodes to reduce ENS and accelerate system recovery. • Reducing ENS in critical consumers through data sharing among DC-MGs. • Reducing computational burden by dividing the problem to the three-stage. 

A hierarchical scheduling framework for resilience enhancement of decentralized renewable-based microgrids considering proactive actions and mobile units

Increasing waste generation has become a significant issue over the globe due to the rapid increase in urbanization and industrialization. In the literature, many issues that have a direct impact on the increase of waste and the improper disposal of waste have been investigated. Most of the existing work in the literature has focused on providing a cost-efficient solution for the monitoring of garbage collection system using the Internet of Things (IoT). Though an IoT-based solution provides the real-time monitoring of a garbage collection system, it is limited to control the spreading of overspill and bad odor blowout gasses. The poor and inadequate disposal of waste produces toxic gases, and radiation in the environment has adverse effects on human health, the greenhouse system, and global warming. While considering the importance of air pollutants, it is imperative to monitor and forecast the concentration of air pollutants in addition to the management of the waste. In this paper, we present and IoT-based smart bin using a machine and deep learning model to manage the disposal of garbage and to forecast the air pollutant present in the surrounding bin environment. The smart bin is connected to an IoT-based server, the Google Cloud Server (GCP), which performs the computation necessary for predicting the status of the bin and for forecasting air quality based on real-time data. We experimented with a traditional model (k-nearest neighbors algorithm (k-NN) and logistic reg) and a non-traditional (long short term memory (LSTM) network-based deep learning) algorithm for the creation of alert messages regarding bin status and forecasting the amount of air pollutant carbon monoxide (CO) present in the air at a specific instance. The recalls of logistic regression and k-NN algorithm is 79% and 83%, respectively, in a real-time testing environment for predicting the status of the bin. The accuracy of modified LSTM and simple LSTM models is 90% and 88%, respectively, to predict the future concentration of gases present in the air. The system resulted in a delay of 4 s in the creation and transmission of the alert message to a sanitary worker. The system provided the real-time monitoring of garbage levels along with notifications from the alert mechanism. The proposed works provide improved accuracy by utilizing machine learning as compared to existing solutions based on simple approaches.

Waste Management and Prediction of Air Pollutants Using IoT and Machine Learning Approach

Load forecasting is a pivotal part of the power utility companies. To provide load-shedding free and uninterrupted power to the consumer, decision-makers in the utility sector must forecast the future demand for electricity with a minimum error percentage. Load prediction with less percentage of error can save millions of dollars to the utility companies. There are numerous Machine Learning (ML) techniques to amicably forecast electricity demand, among which the hybrid models show the best result. Two or more than two predictive models are amalgamated to design a hybrid model, each of which provides improved performances by the merit of individual algorithms. This paper reviews the current state-of-the-art of electric load forecasting technologies and presents recent works pertaining to the combination of different ML algorithms into two or more methods for the construction of hybrid models. A comprehensive study of each single and multiple load forecasting model is performed with an in-depth analysis of their advantages, disadvantages, and functions. A comparison between their performance in terms of Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE) values are developed with pertinent literature of several models to aid the researchers with the selection of suitable models for load prediction.

/pdf/a-comprehensive-review-of-the-load-forecasting-techniques-3angmownpo.pdf

A Comprehensive Review of the Load Forecasting Techniques Using Single and Hybrid Predictive Models

Metrics and enhancement strategies for grid resilience and reliability during natural disasters

In recent times, waste management problem has become a crucial challenge for Bangladesh, which is having a detrimental impact on the environment. This paper presents the proposition of designing a smart dustbin similar to an Automated Teller Machine (ATM) along with an intelligent embedded system, which has been dubbed as Automated Teller Dustbin (ATD). An efficient convolutional neural network (CNN) based image classifier is developed, which is able to detect and recognize any object regarded as garbage by analyzing training features. Additionally, it can also count the number of labeled objects and assign a price value to each object. The waste brought by any individual to the ATD will readily be recognized by the image classifier and the recycle value, which has been assigned for that object can be withdrawn by that individual. Therefore, a direct exchange of waste and its equivalent price is possible, which will incentivize people to use our proposed smart dustbin. After the installation cost, the operation and maintenance cost can be gained by recycling the garbage in it. A pre-trained CNN-based model ALexNet has been utilized to train and test the model with a dataset of 20 images for each of the 10 categorized objects collected from different waste management shops in Dhaka, Bangladesh. The model that has been trained for object recognition has attained an accuracy of 96%, which bears testimony to the feasibility of our proposal.

Design of a Convolutional Neural Network Based Smart Waste Disposal System

In this paper, an automatic speech recognition system has been proposed for isolated Bangla word using Support vector machine with Dynamic time warping (DTW). For training purposes, we have collected data from 40 speakers for five different Bangla words. All the data was collected in a highly acoustic and noise-proof environment. Mel frequency cepstrum coefficients (MFCC’s) are considered as static features from the speech signal. For dynamic features, first and second derivatives of MFCC are utilized. After determining feature vectors, a modified DTW method is proposed for feature matching. Finally, for classification Support Vector Machine (SVM) with Radial basis function (RBF) is utilized. The model is tested for 12 speakers and the recognition rate that we achieved is 86.08%.

Dynamic Time Warping Assisted SVM Classifier for Bangla Speech Recognition

Exchange of private information over the public domain is very much susceptible to unauthorized access, therefore this necessitates the need for a cryptosystem to ensure the protection of information against forthcoming threats. Elliptical curve cryptography (ECC) has attracted the eyes of many scientists due to its smaller key size and high-speed operation. In this paper, an efficient hardware implementation using field programmable gate array (FPGA) on elliptical curve processor (ECP) over a prime field has been proposed. Jacobian coordinate has been used to avoid modular inversion, which is regarded as the most costly operation. To minimize the area and delay in modular multiplication, an interleaved modular multiplier has been proposed. Point doubling (PD) and point addition (PA) architecture was designed with the minimum arithmetic unit using the efficient modular multiplier algorithm. Furthermore, an efficient elliptical curve point multiplication (ECPM) module has been proposed using the high-performance PD and PA architecture. It was found that the proposed ECPM module has a minimal delay and is very area efficient and thus it has immense potentiality in its application in modern day cryptosystem.

Debopriya Roy Dipta

Papers

A Comprehensive Review of the Load Forecasting Techniques Using Single and Hybrid Predictive Models

Metrics and enhancement strategies for grid resilience and reliability during natural disasters

Design of a Convolutional Neural Network Based Smart Waste Disposal System

Dynamic Time Warping Assisted SVM Classifier for Bangla Speech Recognition

Efficient Hardware Implementation of 256-bit ECC Processor Over Prime Field