Deep learning methods have been widely applied in motor imagery(MI)-based brain-computer interfaces(BCI)for decoding electroencephalogram(EEG)signals.High temporal resolution and asymmetric spatial activation are fund...Deep learning methods have been widely applied in motor imagery(MI)-based brain-computer interfaces(BCI)for decoding electroencephalogram(EEG)signals.High temporal resolution and asymmetric spatial activation are fundamental properties of EEG during MI processes.However,due to the limited receptive field of convolutional kernels,traditional convolutional neural networks(CNNs)often focus only on local features,and are insufficient to cover neural processes across different frequency bands and duration scales.This limitation hinders the effective characterization of rhythmic activity changes in MI-EEG signals over time.Additionally,MI-EEG signals exhibit significant asymmetric activation between the left and right hemispheres.Traditional spatial feature extraction methods overlook the interaction between global and local regions at the spatial scale of EEG signals,resulting in inadequate spatial representation and ultimately limiting decoding accuracy.To address these limitations,in this study,a novel deep learning network that integrates multi-modal temporal features with spatially asymmetric feature modeling was proposed.The network first extracts multi-modal temporal information from EEG data channels,and then captures global and hemispheric spatial features in the spatial dimension and fuses them through an advanced fusion layer.Global dependencies are captured using a self-attention module,and a multi-scale convolutional fusion module is introduced to explore the relationships between the two types of temporal features.The fused features are classified through a classification layer to accomplish motor imagery task classification.To mitigate the issue of limited sample size,a data augmentation strategy based on signal segmentation and recombination is designed.Experimental results on the BCI Competition IV-2a(bbic-IV-2a)and BCI Competition IV-2b(bbic-IV-2a)datasets demonstrated that the proposed method achieved superior accuracy in multi-class motor imagery classification compared with existing models.On the BCI-IV-2a dataset,it attained an average classification accuracy of 84.36%,while also showing strong performance on the binary classification BCI-IV-2b dataset.These outcomes validate the capability of the proposed network to enhance MI-EEG classification accuracy.展开更多
In this paper,an algorithm on measurement noise with adaptive strong tracking unscented Kalman filter(ASTUKF)is advanced to improve the precision of pose estimation and the stability for data computation.To suppress h...In this paper,an algorithm on measurement noise with adaptive strong tracking unscented Kalman filter(ASTUKF)is advanced to improve the precision of pose estimation and the stability for data computation.To suppress high-frequency noise,an infinite impulse response filter(IIRF)is introduced at the front end of ASTUKF to preprocess the original data.Then the covariance matrix of the error is corrected and the measurement noise is estimated in the process of filtering.After that,the data from the experiment were tested on the hardware experiment platform.The experimental results show that compared to the traditional extended Kalman filter(EKF)and unscented Kalman filter(UKF)algorithms,the root mean square error(RMSE)of the roll axis results from the algorithm proposed in this paper is respectively reduced by approximately 57.5%and 36.1%;the RMSE of the pitch axis results decreases by nearly 58.4%and 51.5%,respectively;and the RMSE of the yaw axis results decreases almost 62.8%and 50.9%,correspondingly.The above results indicate that the algorithm enhances the ability of resisting high-frequency vibration interference and improves the accuracy of attitude solution.展开更多
文摘Deep learning methods have been widely applied in motor imagery(MI)-based brain-computer interfaces(BCI)for decoding electroencephalogram(EEG)signals.High temporal resolution and asymmetric spatial activation are fundamental properties of EEG during MI processes.However,due to the limited receptive field of convolutional kernels,traditional convolutional neural networks(CNNs)often focus only on local features,and are insufficient to cover neural processes across different frequency bands and duration scales.This limitation hinders the effective characterization of rhythmic activity changes in MI-EEG signals over time.Additionally,MI-EEG signals exhibit significant asymmetric activation between the left and right hemispheres.Traditional spatial feature extraction methods overlook the interaction between global and local regions at the spatial scale of EEG signals,resulting in inadequate spatial representation and ultimately limiting decoding accuracy.To address these limitations,in this study,a novel deep learning network that integrates multi-modal temporal features with spatially asymmetric feature modeling was proposed.The network first extracts multi-modal temporal information from EEG data channels,and then captures global and hemispheric spatial features in the spatial dimension and fuses them through an advanced fusion layer.Global dependencies are captured using a self-attention module,and a multi-scale convolutional fusion module is introduced to explore the relationships between the two types of temporal features.The fused features are classified through a classification layer to accomplish motor imagery task classification.To mitigate the issue of limited sample size,a data augmentation strategy based on signal segmentation and recombination is designed.Experimental results on the BCI Competition IV-2a(bbic-IV-2a)and BCI Competition IV-2b(bbic-IV-2a)datasets demonstrated that the proposed method achieved superior accuracy in multi-class motor imagery classification compared with existing models.On the BCI-IV-2a dataset,it attained an average classification accuracy of 84.36%,while also showing strong performance on the binary classification BCI-IV-2b dataset.These outcomes validate the capability of the proposed network to enhance MI-EEG classification accuracy.
基金supported by the Key Research and Development Program of Shaanxi Province(No.2024NC-YBXM-246)the Shaanxi Provincial Science and Technology Department(No.2024JC-YBQN-0725)+1 种基金the Education Department of Shaanxi Province(No.23JK0371)the Shaanxi University of Technology(No.SLGRCQD2318).
文摘In this paper,an algorithm on measurement noise with adaptive strong tracking unscented Kalman filter(ASTUKF)is advanced to improve the precision of pose estimation and the stability for data computation.To suppress high-frequency noise,an infinite impulse response filter(IIRF)is introduced at the front end of ASTUKF to preprocess the original data.Then the covariance matrix of the error is corrected and the measurement noise is estimated in the process of filtering.After that,the data from the experiment were tested on the hardware experiment platform.The experimental results show that compared to the traditional extended Kalman filter(EKF)and unscented Kalman filter(UKF)algorithms,the root mean square error(RMSE)of the roll axis results from the algorithm proposed in this paper is respectively reduced by approximately 57.5%and 36.1%;the RMSE of the pitch axis results decreases by nearly 58.4%and 51.5%,respectively;and the RMSE of the yaw axis results decreases almost 62.8%and 50.9%,correspondingly.The above results indicate that the algorithm enhances the ability of resisting high-frequency vibration interference and improves the accuracy of attitude solution.
