Hyperparameter Optimization with Genetic Algorithms and XGBoost: A Step Forward in Smart Grid Fraud Detection

This study provides a comprehensive analysis of the combination of Genetic Algorithms (GA) and XGBoost, a well-known machine-learning model. The primary emphasis lies in hyperparameter optimization for fraud detection in smart grid applications. The empirical findings demonstrate a noteworthy enhancement in the model's performance metrics following optimization, particularly emphasizing a substantial increase in accuracy from 0.82 to 0.978. The precision, recall, and AUROC metrics demonstrate a clear improvement, indicating the effectiveness of optimizing the XGBoost model for fraud detection. The findings from our study significantly contribute to the expanding field of smart grid fraud detection. These results emphasize the potential uses of advanced metaheuristic algorithms to optimize complex machine-learning models. This work showcases significant progress in enhancing the accuracy and efficiency of fraud detection systems in smart grids.

Towards a sustainable edge computing framework for condition monitoring in decentralized photovoltaic systems

In recent times, the rapid advancements in technology have led to a digital revolution in urban areas, and new computing frameworks are emerging to address the current issues in monitoring and fault detection, particularly in the context of the growing renewable decentralized energy systems. This research proposes a novel framework for monitoring the condition of decentralized photovoltaic systems within a smart city infrastructure. The approach uses edge computing to overcome the challenges associated with costly processing through remote cloud servers. By processing data at the edge of the network, this concept allows for significant gains in speed and bandwidth consumption, making it suitable for a sustainable city environment. In the proposed edge-learning scheme, several machine learning models are compared to find the best suitable model achieving both high accuracy and low latency in detecting photovoltaic faults. Four light and rapid machine learning models, namely, CBLOF, LOF, KNN, ANN, are selected as best performers and trained locally in decentralized edge nodes. The overall approach is deployed in a smart solar campus with multiple distributed PV units located in the R&D platform Green & Smart Building Park. Several experiments were conducted on different anomaly scenarios, and the models were evaluated based on their supervision method, f1-score, inference time, RAM usage, and model size. The paper also investigates the impact of the type of supervision and the class of the model on the anomaly detection performance. The findings indicated that the supervised artificial neural network (ANN) had superior performance compared to other models, obtaining an f1-score of 80 % even in the most unfavorable conditions. The findings also showed that KNN was the most suitable unsupervised model for the investigated experiments achieving good f1-scores (100 %, 95 % and 92 %) in 3 out of 4 scenarios making it a good candidate for similar anomaly detection tasks.