Search

Medium voltage (MV) switchgear is a vital part of modern power systems, responsible for regulating the flow of electrical power and ensuring the safety of equipment and personnel. However, switchgear can experience various types of faults that can compromise its reliability and safety. Common faults in switchgear include arcing, tracking, corona, normal cases, and mechanical faults. Accurate detection of these faults is essential for maintaining the safety of MV switchgear. In this paper, we propose a novel approach for fault detection using a hybrid model (1D-CNN-LSTM) in both the time domain (TD) and frequency domain (FD). The proposed approach involves gathering a dataset of switchgear operation data and pre-processing it to prepare it for training. The hybrid model is then trained on this dataset, and its performance is evaluated in the testing phase. The results of the testing phase demonstrate the effectiveness of the hybrid model in detecting faults. The model achieved 100% accuracy in both the time and frequency domains for classifying faults in Switchgear, including arcing, tracking, and mechanical faults. Additionally, the model achieved 98.4% accuracy in detecting corona faults in the TD. The hybrid model proposed in this study provides an effective and efficient approach for fault detection in MV switchgear. By learning spatial and temporal features simultaneously, this model can accurately classify faults in both the TD and FD. This approach has significant potential to improve the safety of MV switchgear as well as other industrial applications.

Recent advances in deep learning (DL) have shown that data-driven insights can be used in smart healthcare applications to improve the quality of life for patients. DL needs more data and diversity to build a more accurate system. To satisfy these requirements, more data need to be pooled at the centralized server to train the model deeply, but the process of pooling faces privacy and regulatory challenges. To settle them, the concept of sharing model learning rather than sharing data through federated learning (FL) is proposed. FL creates a more reliable system without transferring data to the server, resulting in the right system with stronger security and access rights to data that protect privacy. This research aims to (1) provide a literature review and an in-depth study on the roles of FL in the fields of healthcare; (2) highlight the effectiveness of current challenges facing standardized FL, including statistical data heterogeneity, privacy and security concerns, expensive communications, limited resources, and efficiency; and (3) present lists of open research challenges and recommendations for future FL for the academic and industrial sectors in telemedicine and remote healthcare applications. An extensive review of the literature on FL from a data-centric perspective was conducted. We searched the Science Direct, IEEE Xplore, and PubMed databases for publications published between January 2018 and January 2023. A new crossover matching between the approaches that solve or mitigate all types of skewed data has been proposed to open up opportunities to other researchers. In addition, a list of various applications was organized by learning application task types such as prediction, diagnosis, and classification. We think that this study can serve as a helpful manual for academics and industry professionals, giving them guidance and important directions for future studies.

Response surface methodology (RSM) is used in this study to optimize the thermal characteristics of single graphene nanoplatelets and hybrid nanofluids utilizing the miscellaneous design model. The nanofluids comprise graphene nanoplatelets and graphene nanoplatelets/cellulose nanocrystal nanoparticles in the base fluid of ethylene glycol and water (60:40). Using response surface methodology (RSM) based on central composite design (CCD) and mini tab 20 standard statistical software, the impact of temperature, volume concentration, and type of nanofluid is used to construct an empirical mathematical formula. Analysis of variance (ANOVA) is applied to determine that the developed empirical mathematical analysis is relevant. For the purpose of developing the equations, 32 experiments are conducted for second-order polynomial to the specified outputs such as thermal conductivity and viscosity. Predicted estimates and the experimental data are found to be in reasonable arrangement. In additional words, the models could expect more than 85% of thermal conductivity and viscosity fluctuations of the nanofluid, indicating that the model is accurate. Optimal thermal conductivity and viscosity values are 0.4962 W/m-K and 2.6191 cP, respectively, from the results of the optimization plot. The critical parameters are 50 °C, 0.0254%, and the category factorial is GNP/CNC, and the relevant parameters are volume concentration, temperature, and kind of nanofluid. From the results plot, the composite is 0.8371. The validation results of the model during testing indicate the capability of predicting the optimal experimental conditions.

We evaluate the performance of the penalized vector autoregression (VAR), diffusion index (DI), and regression tree-based ensemble learning models to forecast Singapore's macroeconomy using high-dimensional data. Our dataset consists of 220 monthly time series that capture the economy of Singapore and 20 monthly times series that capture the global economic environment. We find that the penalized VAR model and the ensemble learning model give an outstanding performance in both short and long horizons. On the other hand, the performance of the DI model depends crucially on the methods to select the number of factors. In particular, a conventional selection method may overestimate the true number of factors and thus deteriorate the forecasting performance of the DI model. Additionally, the VAR and DI models may utilize different information in forecasting. [ABSTRACT FROM AUTHOR]