Epilepsy is a chronic neurological disorder characterized by recurrent seizures,posing significant challenges to patients’quality of life.Accurate classification of seizure states is crucial for effective interventio...Epilepsy is a chronic neurological disorder characterized by recurrent seizures,posing significant challenges to patients’quality of life.Accurate classification of seizure states is crucial for effective intervention.This paper presents a deep learning-based approach for epileptic seizure classification by integrating multi-feature analysis of electroencephalogram(EEG)signals.The proposed method begins with signal preprocessing,including denoising,segmentation,and label construction.Subsequently,a comprehensive set of temporal,spectral,and wavelet-based features—such as signal mean,power,heart rate,and wavelet coefficients—is extracted.Feature selection is then performed using the Maximal Information Coefficient(MIC)to identify the most discriminative inputs.A hybrid model combining a Transformer encoder and a Long Short-Term Memory(LSTM)network is developed to effectively capture both long-range dependencies and temporal dynamics in EEG sequences for seizure classification.Evaluated on the Bonn dataset using 5-fold cross-validation,the proposed method achieves an accuracy of 96.43%in distinguishing between epileptic patients and healthy subjects,with a sensitivity of 97.53%in detecting seizure states.It also attains a multi-class classification accuracy of 90.14%across different epileptic signal types.Ablation studies confirm that MICbased feature selection improves accuracy by over 2O%compared to using raw features without selection.The results demonstrate that the integration of multi-feature analysis with the Transformer-LSTM architecture offers an effective and reliable solution for EEG-based seizure classification.展开更多
Accurate estimation of battery health status plays a crucial role in battery management systems.However,the lack of operational data still affects the accuracy of battery state of health(SOH)estimation.For this reason...Accurate estimation of battery health status plays a crucial role in battery management systems.However,the lack of operational data still affects the accuracy of battery state of health(SOH)estimation.For this reason,a SOH estimation method is proposed based on charging data reconstruction combined with image processing.The charging voltage data is used to train the least squares generative adversarial network(LSGAN),which is validated under different levels of missing data.From a visual perspective,the Gram angle field method is applied to convert one-dimensional time series data into image data.This method fully preserves the time series characteristics and nonlinear evolution patterns,which avoids the difficulties and limited expressive power associated with manual feature extraction.At the same time,the Swin Transformer model is introduced to extract global structures and local details from images,enabling better capture of sequence change trends.Combined with the long short-term memory network(LSTM),this enables accurate estimation of battery SOH.Two different types of batteries are used to validate the test.The experimental results show that the proposed method has good estimation accuracy under different training proportions.展开更多
基金supported by the National Key Research and Development Program of China(2020AAA0104905)in part by National Natural Science Foundation of China underGrant(62341118,62503241)+1 种基金in part by the Natural Science Foundation of Jiangsu Province of China under Grant(BK20250664)in part by Foundation of recruiting talents of HYIT under Grant(Z301B25508).
摘要Epilepsy is a chronic neurological disorder characterized by recurrent seizures,posing significant challenges to patients’quality of life.Accurate classification of seizure states is crucial for effective intervention.This paper presents a deep learning-based approach for epileptic seizure classification by integrating multi-feature analysis of electroencephalogram(EEG)signals.The proposed method begins with signal preprocessing,including denoising,segmentation,and label construction.Subsequently,a comprehensive set of temporal,spectral,and wavelet-based features—such as signal mean,power,heart rate,and wavelet coefficients—is extracted.Feature selection is then performed using the Maximal Information Coefficient(MIC)to identify the most discriminative inputs.A hybrid model combining a Transformer encoder and a Long Short-Term Memory(LSTM)network is developed to effectively capture both long-range dependencies and temporal dynamics in EEG sequences for seizure classification.Evaluated on the Bonn dataset using 5-fold cross-validation,the proposed method achieves an accuracy of 96.43%in distinguishing between epileptic patients and healthy subjects,with a sensitivity of 97.53%in detecting seizure states.It also attains a multi-class classification accuracy of 90.14%across different epileptic signal types.Ablation studies confirm that MICbased feature selection improves accuracy by over 2O%compared to using raw features without selection.The results demonstrate that the integration of multi-feature analysis with the Transformer-LSTM architecture offers an effective and reliable solution for EEG-based seizure classification.
基金supported in part by the National Natural Science Foundation of China(under Grant 62473309,62203352)the Shaanxi Outstanding Youth Science Fund Project(under Grant 2024JC-JCQN-68)+1 种基金the Xi’an Science and Technology Plan Project(under Grant 24GXFW0050)the Xi’an Key Laboratory(under Grant 24ZDSY0015).
摘要Accurate estimation of battery health status plays a crucial role in battery management systems.However,the lack of operational data still affects the accuracy of battery state of health(SOH)estimation.For this reason,a SOH estimation method is proposed based on charging data reconstruction combined with image processing.The charging voltage data is used to train the least squares generative adversarial network(LSGAN),which is validated under different levels of missing data.From a visual perspective,the Gram angle field method is applied to convert one-dimensional time series data into image data.This method fully preserves the time series characteristics and nonlinear evolution patterns,which avoids the difficulties and limited expressive power associated with manual feature extraction.At the same time,the Swin Transformer model is introduced to extract global structures and local details from images,enabling better capture of sequence change trends.Combined with the long short-term memory network(LSTM),this enables accurate estimation of battery SOH.Two different types of batteries are used to validate the test.The experimental results show that the proposed method has good estimation accuracy under different training proportions.