The taste electroencephalogram (EEG) evoked by the taste stimulation can reflect different brain patterns and be used in applications such as sensory evaluation of food. However, considering the computational cost and efficiency, EEG data with many channels has to face the critical issue of channel selection. This paper proposed a channel selection method called class activation mapping with attention (CAM-Attention). The CAM-Attention method combined a convolutional neural network with channel and spatial attention (CNN-CSA) model with a gradient-weighted class activation mapping (Grad-CAM) model. The CNN-CSA model exploited key features in EEG data by attention mechanism, and the Grad-CAM model effectively realized the visualization of feature regions. Then, channel selection was effectively implemented based on feature regions. Experimental results showed that the proposed CAM-Attention method achieved an accuracy of 97.85% and an F1-score of 97.74% when the selected channel number was 12, which were only 0.25% and 0.33% lower, respectively, compared to using all channels. This demonstrates that the CAM-Attention method can significantly reduce computational burden while maintaining excellent classification performance. In short, it has excellent recognition performance and provides effective technical support for taste sensory evaluation.
Traditional taste evaluation relies heavily on manual sensory analysis, which is highly subjective and inefficient with poor cross-individual generalization, limiting its application in industrial flavor detection. To achieve accurate cross-subject taste recognition, this paper proposes an electroencephalogram (EEG) classification method based on a meta-learning wavelet graph convolutional neural network (ML-WGCNet) under sweet- and bitter-taste stimuli. Sucrose (sweetness) and quinine (bitterness) were used as stimulation sources, each prepared at six concentration gradients, including a water control. EEG signals were detected from 20 subjects. First, the Morlet wavelet transform was applied to decompose the EEG signals in the time–frequency domain, extracting the maximum and average energy values from five frequency bands as core features. A graph structure was then constructed using electrodes as nodes and Pearson correlation coefficients between electrodes as edge weights. A lightweight graph convolutional neural network (GCN) is employed to model spatial correlations among brain regions. Finally, by integrating a meta-learning framework and adopting leave-one-subject-out cross-validation, the model can rapidly adapt to new subjects. The experimental results show that the proposed method achieves average accuracies of 76.03% and 77.01% in cross-subject classification of sweet and bitter tastes, respectively. The corresponding precision values are 79.94% and 79.53%, the recall values are 75.77% and 78.51%, and the F1-scores are 78.24% and 78.08%, respectively, demonstrating that the proposed model significantly outperforms existing mainstream EEG classification methods.
Detecting and identifying consumers’ perception of food sweetness can help guide the optimization of food formulations. Electroencephalogram (EEG) detection can capture changes in brain electrical activity in response to different sweet taste stimuli. In this work, we employ EEG detection and propose an EEG Feature Calculation and Classification Network (EFCC-Net) to recognize taste EEG signals under different sweetness concentration stimuli. First, taste-related EEG data from a subject group under varying sweetness concentration stimuli are collected. Then, an EEG Feature Calculation Module (EFCM) is proposed, which utilizes convolutional kernels of different sizes to compute local features from both temporal and spatial dimensions of EEG data. A lightweight self-attention mechanism is employed to compute global features, and a multi-branch computation approach is adopted to enhance feature extraction capability. Next, based on EEG topographic maps, qualitative analysis is conducted to examine differences in brain region activation under varying taste concentrations. Finally, leveraging the proposed EFCM, the EFCC-Net is designed to classify EEG data corresponding to different sweetness levels. Through structural optimization, ablation experiments, and comparisons with state-of-the-art EEG classification methods, EFCC-Net achieves the best classification performance, with an accuracy of 96.57%, a precision of 96.58%, and a recall of 96.53%, while also demonstrating superior stability.
As distributed generators and new loads spread, grid operations have grown more complex, transitioning from AC to hybrid AC/DC and pure DC distribution grids [1] .
1 September 2026 in Recent Advances in Electrical & Electronic Engineering (Formerly Recent Patents on Electrical & Electronic Engineering)
Introduction: High penetration of distributed photovoltaic generation can cause simultaneous overvoltage, undervoltage, and increased voltage fluctuations in distribution feeders. This study proposes a voltage-quality-assessment-based strategy for coordinating the configuration and operation of step voltage regulators. Methods: An SVR tap-ratio model and a modified Newton-Raphson power-flow formulation are first established. Seven voltage-quality and user-side indicators are then weighted using improved entropy weighting and IAHP, and bus-level regulation demand is mapped to weak branches. A bilevel model coordinates SVR and capacitor-bank planning with daily SVR taps, capacitor switching, and PV-inverter reactive-power operation. The mixed discrete-continuous problem is solved using an improved particle swarm optimization algorithm. Results: In the modified IEEE 33-bus case, the coordinated strategy narrows the voltage range from 0.9073-1.0995 p.u. to 0.9641-1.0358 p.u. and reduces the average network loss from 0.1924 to 0.1603 MW. It also avoids the PV curtailment required by the curtailment-only comparison scenario. Discussion: The results show that assessment-based candidate screening can be integrated with coordinated SVR-CB-PV operation. The present model assumes deterministic PV and load profiles and does not include tap- or switching-wear costs; these limitations should be addressed through uncertainty modeling and validation on actual feeders. Conclusion: The proposed framework improves the voltage profile and network-loss performance in the tested radial feeders without relying on PV curtailment. Field validation and uncertainty-aware extensions are required before practical deployment.