Loading...
3 results
Search Results
Now showing 1 - 3 of 3
- SpectraNet–53: A deep residual learning architecture for predicting soluble solids content with VIS–NIR spectroscopyPublication . A. Martins, J.; Guerra, Rui Manuel Farinha das Neves; Pires, R.; Antunes, M.D.; Panagopoulos, T.; Brázio, A.; Afonso, A.M.; Silva, L.; Lucas, M.R.; Cavaco, A.M.This work presents a new deep learning architecture, SpectraNet-53, for quantitative analysis of fruit spectra, optimized for predicting Soluble Solids Content (SSC, in Brix). The novelty of this approach resides in being an architecture trainable on a very small dataset, while keeping a performance level on-par or above Partial Least Squares (PLS), a time-proven machine learning method in the field of spectroscopy. SpectraNet-53 performance is assessed by determining the SSC of 616 Citrus sinensi L. Osbeck 'Newhall' oranges, from two Algarve (Portugal) orchards, spanning two consecutive years, and under different edaphoclimatic conditions. This dataset consists of short-wave near-infrared spectroscopic (SW-NIRS) data, and was acquired with a portable spectrometer, in the visible to near infrared region, on-tree and without temperature equalization. SpectraNet-53 results are compared to a similar state-of-the-art architecture, DeepSpectra, as well as PLS, and thoroughly assessed on 15 internal validation sets (where the training and test data were sampled from the same orchard or year) and on 28 external validation sets (training/test data sampled from different orchards/years). SpectraNet-53 was able to achieve better performance than DeepSpectra and PLS in several metrics, and is especially robust to training overfit. For external validation results, on average, SpectraNet-53 was 3.1% better than PLS on RMSEP (1.16 vs. 1.20 Brix), 11.6% better in SDR (1.22 vs. 1.10), and 28.0% better in R2 (0.40 vs. 0.31).
- A comprehensive approach for optimizing controller placement in Software-Defined NetworksPublication . Schutz, G.; Martins, JaimeSoftware-Defined Networks (SDNs) are characterized by dividing a network architecture in a data plane (i.e., any packet-relaying nodes like switches or routers) and a control plane, where specialized controllers assign forwarding decisions to the underlying data plane, and must do so in a very short timeframe. Thus, controllers play a key role in SDNs and the Controller Placement Problem (CPP) becomes a critical issue, affecting network delays and synchronization. If there are significant propagation delays between controllers and nodes, or among controllers, their ability to quickly react to network events is affected, degrading reliability. In this work, we propose a comprehensive mathematical formalization of the CPP, which constrains propagation latency and controller capacity, and determines simultaneously the minimum number of controllers, their location and the assignment of nodes to each, while keeping a balanced load distribution among controllers. As CPP is NP-hard, a heuristic approach is also presented. Simulations for 60 network scenarios show that this approach obtains balanced and resilient solutions, in negligible time, which are proven to be optimal or near optimal for 90% of the evaluated cases.
- Intelligent monitoring systems for electric vehicle chargingPublication . Martins, Jaime; Rodrigues, JoaoFeatured Application This paper reviews EV charging challenges and existing monitoring methods to pinpoint key gaps. From our review, we propose a practical monitoring framework that leverages IoT sensors, edge computing, and cloud services for real-time oversight, predictive maintenance, and responsive analysis of user behavior.Abstract The growing adoption of electric vehicles (EVs) presents new challenges for managing parking infrastructure, particularly concerning charging station utilization and user behavior patterns. This review examines the current state-of-the-art in intelligent monitoring systems for EV charging stations in parking facilities. We specifically focus on two key inefficiencies: vehicles occupying charging spots beyond the optimal fast-charging range (80% state-of-charge) and remaining connected even after reaching full capacity (100%). We analyze the theoretical and practical foundations of these systems, summarizing existing research on intelligent monitoring architectures and commercial implementations. Building on this analysis, we also propose a novel monitoring framework that integrates Internet of things (IoT) sensors, edge computing, and cloud services to enable real-time monitoring, predictive maintenance, and adaptive control. This framework addresses both the technical aspects of monitoring systems and the behavioral factors influencing charging station management. Based on a comparative analysis and simulation studies, we propose performance benchmarks and outline critical research directions requiring further experimental validation. The proposed architecture aims to offer a scalable, adaptable, and secure solution for optimizing EV charging infrastructure utilization while addressing key research gaps in the field.