• Energy Research

Room occupancy prediction based on indoor environmental quality may be the breakthrough to ensure energy efficiency and establish an interior ambience tailored to each user. Identifying whether temperature, humidity, lighting, and CO2 levels may be used as efficient predictors of room occupancy accuracy is needed to help designers better utilize the readings and data collected in order to improve interior design, in an effort to better suit users. It also aims to help in energy efficiency and saving in an ever-increasing energy crisis and dangerous levels of climate change. This paper evaluated the accuracy of room occupancy recognition using a dataset with diverse amounts of light, CO2, and humidity. As classification algorithms, K-nearest neighbors (KNN), hybrid Adam optimizer–artificial neural network–back-propagation network (AO–ANN (BP)), and decision trees (DT) were used. Furthermore, this research is based on machine learning interpretability methodologies. Shapley additive explanations (SHAP) improve interpretability by estimating the significance values for each feature for classifiers applied. The results indicate that the KNN performs better than the DT and AO-ANN (BP) classification models have 99.5%. Though the two classifiers are designed to evaluate variations in interpretations, we must ensure that they have accurate detection. The results show that SHAP provides successful implementation following these metrics, with differences detected amongst classifier models that support the assumption that model complexity plays a significant role when predictability is taken into account.

Smart power grids suffer from electricity theft cyber-attacks, where malicious consumers compromise their smart meters (SMs) to downscale the reported electricity consumption readings. This problem costs electric utility companies worldwide considerable financial burdens and threatens power grid stability. Therefore, several machine learning (ML)-based solutions have been proposed to detect electricity theft; however, they have limitations. First, most existing works employ supervised learning that requires the availability of labeled datasets of benign and malicious electricity usage samples. Unfortunately, this approach is not practical due to the scarcity of real malicious electricity usage samples. Moreover, training a supervised detector on specific cyberattack scenarios results in a robust detector against those attacks, but it might fail to detect new attack scenarios. Second, although a few works investigated anomaly detectors for electricity theft, none of the existing works addressed consumers’ privacy. To address these limitations, in this paper, we propose a comprehensive federated learning (FL)-based deep anomaly detection framework tailored for practical, reliable, and privacy-preserving energy theft detection. In our proposed framework, consumers train local deep autoencoder-based detectors on their private electricity usage data and only share their trained detectors’ parameters with an EUC aggregation server to iteratively build a global anomaly detector. Our extensive experimental results not only demonstrate the superior performance of our anomaly detector compared to the supervised detectors but also the capability of our proposed FL-based anomaly detector to accurately detect zero-day attacks of electricity theft while preserving consumers’ privacy.