Adaptive electricity theft detection method based on load shape dictionary of customers

With the application of the advanced measurement infrastructure in power grids,data driven electricity theft detection methods become the primary stream for pinpointing electricity thieves.However,owing to anomaly submergence,which shows that the usage patterns of electricity thieves may not always deviate from those of normal users,the performance of the existing usage-pattern-based method could be affected.In addition,the detection results of some unsupervised learning algorithm models are abnormal degrees rather than“0-1”to ascertain whether electricity theft has occurred.The detection with fixed threshold value may lead to deviation and would not be sufficiently flexible to handle the detection for different scenes and users.To address these issues,this study proposes a new electricity theft detection method based on load shape dictionary of users.A corresponding strategy for tunable threshold is proposed to optimize the detection effect of electricity theft,and the efficacy and applicability of the proposed adaptive electricity theft detection method were verified from numerical experiments.

RoGRUT: A Hybrid Deep Learning Model for Detecting Power Trapping in Smart Grids

Electricity theft is a widespread non-technical issue that has a negative impact on both power grids and electricity users.It hinders the economic growth of utility companies,poses electrical risks,and impacts the high energy costs borne by consumers.The development of smart grids is crucial for the identification of power theft since these systems create enormous amounts of data,including information on client consumption,which may be used to identify electricity theft using machine learning and deep learning techniques.Moreover,there also exist different solutions such as hardware-based solutions to detect electricity theft that may require human resources and expensive hardware.Computer-based solutions are presented in the literature to identify electricity theft but due to the dimensionality curse,class imbalance issue and improper hyper-parameter tuning of such models lead to poor performance.In this research,a hybrid deep learning model abbreviated as RoGRUT is proposed to detect electricity theft as amalicious and non-malicious activity.The key steps of the RoGRUT are data preprocessing that covers the problem of class imbalance,feature extraction and final theft detection.Different advanced-level models like RoBERTa is used to address the curse of dimensionality issue,the near miss for class imbalance,and transfer learning for classification.The effectiveness of the RoGRUTis evaluated using the dataset fromactual smartmeters.A significant number of simulations demonstrate that,when compared to its competitors,the RoGRUT achieves the best classification results.The performance evaluation of the proposed model revealed exemplary results across variousmetrics.The accuracy achieved was 88%,with precision at an impressive 86%and recall reaching 84%.The F1-Score,a measure of overall performance,stood at 85%.Furthermore,themodel exhibited a noteworthyMatthew correlation coefficient of 78%and excelled with an area under the curve of 91%.

Electricity Theft Detection Method Based on Ensemble Learning and Prototype Learning

With the development of advanced metering infrastructure(AMI),large amounts of electricity consumption data can be collected for electricity theft detection.However,the imbalance of electricity consumption data is violent,which makes the training of detection model challenging.In this case,this paper proposes an electricity theft detection method based on ensemble learning and prototype learning,which has great performance on imbalanced dataset and abnormal data with different abnormal level.In this paper,convolutional neural network(CNN)and long short-term memory(LSTM)are employed to obtain abstract feature from electricity consumption data.After calculating the means of the abstract feature,the prototype per class is obtained,which is used to predict the labels of unknown samples.In the meanwhile,through training the network by different balanced subsets of training set,the prototype is representative.Compared with some mainstream methods including CNN,random forest(RF)and so on,the proposed method has been proved to effectively deal with the electricity theft detection when abnormal data only account for 2.5%and 1.25%of normal data.The results show that the proposed method outperforms other state-of-the-art methods.

Energy Theft Identification Using Adaboost Ensembler in the Smart Grids

One of the major concerns for the utilities in the Smart Grid(SG)is electricity theft.With the implementation of smart meters,the frequency of energy usage and data collection from smart homes has increased,which makes it possible for advanced data analysis that was not previously possible.For this purpose,we have taken historical data of energy thieves and normal users.To avoid imbalance observation,biased estimates,we applied the interpolation method.Furthermore,the data unbalancing issue is resolved in this paper by Nearmiss undersampling technique and makes the data suitable for further processing.By proposing an improved version of Zeiler and Fergus Net(ZFNet)as a feature extraction approach,we had able to reduce the model’s time complexity.To minimize the overfitting issues,increase the training accuracy and reduce the training loss,we have proposed an enhanced method by merging Adaptive Boosting(AdaBoost)classifier with Coronavirus Herd Immunity Optimizer(CHIO)and Forensic based Investigation Optimizer(FBIO).In terms of low computational complexity,minimized over-fitting problems on a large quantity of data,reduced training time and training loss and increased training accuracy,our model outperforms the benchmark scheme.Our proposed algorithms Ada-CHIO andAda-FBIO,have the low MeanAverage Percentage Error(MAPE)value of error,i.e.,6.8%and 9.5%,respectively.Furthermore,due to the stability of our model our proposed algorithms Ada-CHIO and Ada-FBIO have achieved the accuracy of 93%and 90%.Statistical analysis shows that the hypothesis we proved using statistics is authentic for the proposed technique against benchmark algorithms,which also depicts the superiority of our proposed techniques.