The integration of the smart grid with the cloud computing environment promises to develop an improved energy-management system for utility and consumers. New applications and services are being developed which generate huge requests to be processed in the cloud. As smart grids can dynamically be operated according to consumer requests (data), so, they can be called Data-Driven Smart Grids. Fog computing as an extension of cloud computing helps to mitigate the load on cloud data centers. This paper presents a cloud-fog-based system model to reduce Response Time (RT) and Processing Time (PT). The load of requests from end devices is processed in fog data centers. The selection of potential data centers and efficient allocation of requests on Virtual Machines (VMs) optimize the RT and PT. A New Service Broker Policy (NSBP) is proposed for the selection of a potential data center. The load-balancing algorithm, a hybrid of Particle Swarm Optimization and Simulated Annealing (PSO-SA), is proposed for the efficient allocation of requests on VMs in the potential data center. In the proposed system model, Micro-Grids (MGs) are placed near the fogs for uninterrupted and cheap power supply to clusters of residential buildings. The simulation results show the supremacy of NSBP and PSO-SA over their counterparts.

Medium-term electricity consumption and load forecasting in smart grids is an attractive topic of study, especially using innovative data analysis approaches for future energy consumption trends. Loss of electricity during generation and use is also a problem to be addressed. Both consumers and utilities can benefit from a predictive study of electricity demand and pricing. In this study, we used a new machine learning approach called AdaBoost to identify key features from an ISO-NE dataset that includes daily consumption data over eight years. Moreover, the DT classifier and RF are widely used to extract the best features from the dataset. Moreover, we predicted the electricity load and price using machine learning techniques including support vector machine (SVM) and deep learning techniques such as a convolutional neural network (CNN). Coronavirus herd immunity optimization (CHIO), a novel optimization approach, was used to modify the hyperparameters to increase efficiency, and it used classifiers to improve the performance of our classifier. By adding additional layers to the CNN and fine-tuning its parameters, the probability of overfitting the classifier was reduced. For method validation, we compared our proposed models with several benchmarks. MAE, MAPE, MSE, RMSE, the f1 score, recall, precision, and accuracy were the measures used for performance evaluation. Moreover, seven different forms of statistical analysis were given to show why our proposed approaches are preferable. The proposed CNN-CHIO and SVM techniques had the lowest MAPE error rates of 6% and 8%, respectively, and the highest accuracy rates of 95% and 92%, respectively.

No abstract

No abstract

Electricity theft harms smart grids and results in huge revenue losses for electric companies. Deep learning (DL), machine learning (ML), and statistical methods have been used in recent research studies to detect anomalies and illegal patterns in electricity consumption (EC) data collected by smart meters. In this paper, we propose a hybrid DL model for detecting theft activity in EC data. The model combines both a gated recurrent unit (GRU) and a convolutional neural network (CNN). The model distinguishes between legitimate and malicious EC patterns. GRU layers are used to extract temporal patterns, while the CNN is used to retrieve optimal abstract or latent patterns from EC data. Moreover, imbalance of data classes negatively affects the consistency of ML and DL. In this paper, an adaptive synthetic (ADASYN) method and TomekLinks are used to deal with the imbalance of data classes. In addition, the performance of the hybrid model is evaluated using a real-time EC dataset from the State Grid Corporation of China (SGCC). The proposed algorithm is computationally expensive, but on the other hand, it provides higher accuracy than the other algorithms used for comparison. With more and more computational resources available nowadays, researchers are focusing on algorithms that provide better efficiency in the face of widespread data. Various performance metrics such as F1-score, precision, recall, accuracy, and false positive rate are used to investigate the effectiveness of the hybrid DL model. The proposed model outperforms its counterparts with 0.985 Precision–Recall Area Under Curve (PR-AUC) and 0.987 Receiver Operating Characteristic Area Under Curve (ROC-AUC) for the data of EC.