22 September 2021 in IEEE Transactions on Industrial Electronics
AbstractThe rapid introduction of electric vehicles (EVs) in the transportation market has initiated the concept of vehicle-to-grid (V2G) technology in smart grids. However, where V2G technology is intended to facilitate the power grid ancillary services, it could also have an adverse effect on the aging of battery packs in EVs. This is due to the instant depletion of power during the charge and discharge cycles, which could eventually impact the structural complexity and electrochemical operations in the battery pack. To address this situation, a median expectation-based regression approach is proposed for parameter estimation of vehicle batteries in V2G systems. The proposed method is built on the property of uncertainty prediction of Gaussian processes for parameter estimation while considering the cell variations as an exogenous function. Firstly, a median expectation-based Gaussian process model is derived to predict the fused and individual cell variations of a battery pack. Secondly, a magnitude-squared coherence model is developed by the error matrix to detect and isolate each variation. This is obtained by extracting the cross-spectral densities for the measurements. The proposed regression-based approach is evaluated using experimental measurements collected from lithium-ion (Li-ion) battery pack in EVs. The parametric analysis of the battery pack has been verified using D-SAT Chroma 8000ATS hardware platform.
23 January 2023 in Sustainable Energy, Grids and Networks
14 February 2025 in IEEE Access
The need for accurate and efficient malaria diagnosis has driven research into automated solutions using deep learning. This study presents a comparative analysis of five convolutional neural network (CNN) models for malaria detection in microscopic blood cell images. We evaluated models ranging from a basic architecture to advanced iterations employing data augmentation and VGG16 transfer learning. Using a dataset of 24,958 training and 2,600 test images, we assessed performance based on accuracy, precision, recall, F1-score, and computational efficiency. Our results demonstrate that a simpler CNN model (Model_1) achieved greater than 99% accuracy, surpassing the performance of more complex models while requiring significantly less computational resources. This finding underscores the potential of efficient deep learning strategies for developing scalable and cost-effective malaria diagnostic tools.
1 June 2009 in IEEE Sensors Journal
In this paper, a bio-inspired pattern recognition system for tin-oxide gas sensor applications is proposed. To mimic the biological olfactory system, temperature modulation is first used to virtually increase the number of sensors by periodically sampling the sensors' response at different temperatures. A convex microhotplate is used in order to improve the thermal properties of the structure enabling efficient temperature modulation process to be carried out. Temperature modulation is shown to increase the number of effective sensors from 16 physically available sensors (integrated on a single chip) to 12 000 virtual sensors (VSs). This enables the emulation of a very large number of sensors typically found in biological systems. The response of each sensor is seen as a fingerprint map, which is further processed using various image processing techniques. Self organized maps (SOMs) algorithm is used to create a 2D map for each gas and to combine the huge number of VSs in order to reduce the dimensionality. Image moments are used as a feature enabling to characterize the spatial distribution within the image lattice and to retrieve the brighter regions in the SOM' nodes exhibiting high activity to the input gas. Experiments on real sensors data show improved performance (96%) as compared with standard gas discrimination algorithms.
10 March 2025 in Nonlinear Dynamics
14 October 2008 in Signal, Image and Video Processing