Frequency Band Of Signal Research Articles

Feature extraction of rolling bearing based on adaptive variational multi-harmonic mode extraction

Variable multi-harmonic mode extraction (VMHME) not only has the advantages of high computational efficiency and extraction accuracy similar to variational mode extraction (VME), but also could extract the multi-harmonic components of periodic narrowband impulse signals in frequency band as wide as possible, making it very suitable for feature extraction in the event of rolling bearing failure. VMHME needs to accurately estimate the fault characteristic frequency of rolling bearing as its prior parameter, and small errors in estimating the fault characteristic frequency will cause significant deviations in the target extraction components. At present, the theoretical fault characteristic frequency of rolling bearings is commonly used as the estimated fault characteristic frequency. However, due to the installation deformation of rolling bearings and the random sliding between the rolling elements and the raceway during operation, it can cause a deviation between the actual fault characteristic frequency and the theoretical fault characteristic frequency. The most scientific and effective method is to enable VMHME to adaptively obtain the fault characteristic frequency based on the characteristics of the analyzed signal itself. Therefore, this paper introduces the envelope harmonic product spectrum (EHPS) theory into VMHME and proposes an adaptive VMHME (AVMHME) method to effectively extract the multi harmonic components of the periodic narrowband impulse signal when rolling bearings fail. Feasibility of the proposed method is verified through simulation and rolling bearing’ early weak fault experiment, and its superiority is also verified through comparative analysis.

An experimental study on the effect of non-ionizing electromagnetic fields on honey bees

ABSTRACT Due to the increase in data rate in mobile communication and the widespread use of mobile internet, electromagnetic communication systems are increasing daily. This situation causes increases in the use of more mobile communication devices and environmental non-ionizing Electromagnetic Field (EMF) levels. The rise of bee deaths and colony losses in beekeeping parallel to the increase of the EMF sources cause the concept of “electromagnetic pollution” to be considered among the reasons. Therefore, studying the effects of non-ionizing Electromagnetic Radiation (EMR) on the health of living things is one of the most significant issues today. The bees determine their direction with the Earth’s magnetic field. Electromagnetic signals emitted by GSM base stations, etc. may affect the direction-finding capabilities of honey bees and constitute a stress factor. In this study, the aim was to determine the effect of EMF on honey bees and honey yield. Honey bee colonies were used, obtained from the same farm in the Trabzon region, and equalized in all respects. Moreover, these colonies were divided into five groups randomly as experiments and control groups. The experiment hives were exposed to the EMF in the frequency band of the Wi-Fi signals (2.4 GHz) and the high-voltage line (50 hz). The control hives are located far away from the EMR sources. The study was repeated in the second year to confirm the results. During the investigation, some physiological and behavioural effects of bees, such as aggressiveness, brood area, etc. were determined based on EMR exposure.

An innovative approach for FMCW radar vital sign monitoring with removal of respiratory harmonics

Frequency Modulated Continuous Wave (FMCW) radar shows broad research potential in monitoring vital signs. However, the harmonic components generated by the human respiratory process overlap with the frequency band of the heartbeat signal, posing challenges to precisely monitoring the heartbeat signal. To address this issue, this paper introduces the singular spectrum analysis method, which accurately identifies and removes respiratory harmonics by adding a sine signal with the same frequency as the respiratory harmonics to the original time series. Furthermore, the paper introduces energy entropy and the Pearson correlation coefficient to comprehensively decompose physiological signals to optimize the traditional variational mode decomposition algorithm, proposing the EE-VMD and PCC-VMD algorithms. Experimental results demonstrate that the SNR increased, and the accuracy of measuring respiratory rate and heartbeat rate has significantly improved. The results presented in this paper not only advance theoretical methodologies for the monitoring of human vital signs but also underscore the practical applicability of FMCW radar in this domain.

Experimental study on phase noise of terahertz quantum cascade laser frequency comb and dual-comb sources

Frequency combs with equally spaced frequency lines show great potentials for applications in spectroscopy, imaging, communications, and so on. In the terahertz frequency region, the quantum cascade laser (QCL) is an ideal radiation source for frequency comb and dual-comb operation. The systematic evaluation of phase noise characteristics of terahertz QCL frequency comb and dual-comb sources is of great importance for high precision measurements. In this work, we present detailed measurements and analysis of the phase noise characteristics of terahertz QCL frequency comb and dual-comb sources emitting around 4.2 THz with repetition frequencies of ~6.2 GHz. The measurement results for the current noise of the direct current (DC) sources (that are used to electrically pump the terahertz QCLs) indicate that at 100 Hz, the current noise for DC-1 and DC-2 is 0.3895 and 0.0982 nA/Hz1/2, respectively. Such levels of current noise can be safely disregarded. The phase noise of radio frequency (RF) generators (that are employed for injection locking and phase locking), intermode beatnotes, and dual-comb signals with and without phase-locked loop (PLL) are all measured and compared. The experimental results show that in the free-running mode, the phase noise of the intermode beatnote signals is always lower than that of the dual-comb signals across all frequencies. Additionally, the phase noise induced by the RF generators is negligible. By employing the phase locking technique, the phase noise of the intermode beatnote and dual-comb signals in the low offset frequency band can be significantly suppressed. At an offset frequency of 100 Hz, the measured phase noise values of the dual-comb line without and with phase locking are 15.026 and −64.801 dBc/Hz, respectively.

Стохастический резонанс при определении оценки характеристической функции Ляпунова

The purpose of the paper is to investigate the phenomenon of stochastic resonance in a digital filter under the conditions of exposure to an additive mixture of a quasi-deterministic signal and a concentrated interference. The phenomenon of stochastic resonance in a digital filter based on the Lyapunov characteristic function is considered. In the process of modelling of the digital filter the effect and indices of stochastic resonance at filtering of additive mixture of quasi-deterministic signal and concentrated noise are established. The spectral characteristics of the output signal of the digital filter under the conditions of stochastic resonance occurrence are investigated. The results of modelling of a digital filter under the influence of an additive mixture of a useful signal and a concentrated noise are given. Modelling of the digital filter is carried out for cases when the frequency of the interference is equal to the frequency of the useful signal, the frequency of the interference is in the frequency band of the useful signal, as well as when the frequency of the interference is outside the frequency band of the useful signal. The presence of the phenomenon of stochastic resonance under the influence of concentrated interference is shown. The positive influence of the concentrated noise on the characteristics of the digital filter is shown. The improvement of signal/noise ratio at the output of the system in more than 180 times in comparison with the signal/noise ratio at its input is obtained. A comparison of the results of the study of the phenomenon of stochastic resonance in the case of the effects of concentrated noise and «white» noise is given.

Comparative exploration on EEG signal filtering using window control methods

Electroencephalography (EEG) is used to monitor brain activity. The brain signals consist of different frequency band signals delta, theta, alpha, beta, and gamma waves. The signals are affected by external noise which reduces the quality of the EEG signal due to which it becomes difficult to do further processing of EEG signals like feature extraction or extraction of meaningful features from EEG signal. Therefore, it becomes important to filter the noise from the EEG signal before feature extraction or classification of the EEG signal. The research article presents an overview of different types of windowing filter techniques like Rectangular, Bartlett, Hamming, Hanning, and Kaiser windows applied for finite impulse response (FIR) behavior which is used for EEG signal processing for different brain waves processed in different frequency bands. The comparative analysis is carried out in terms of the response time of brain frequency bands for different windowing filter techniques using the MATLAB 2023 signal processing simulation tool. The novelty of the work lies in estimating minimum latency and appropriate filter selection for various typical EEG waves, since the EEG signals are pre-supposed in the hardware chip design, noise elimination is the first step in high-performance computing applications. The Bartlett window band stop has an optimal response time of 12.666 s for delta waves, a highpass filter with a response time of 16.187 s for theta waves, a bandpass with a response time of 13.122 s for alpha waves, a highpass filter with a response time of 17.866 s for beta waves, and a highpass filter with a response time of 13.797 s for gamma waves. The Barlett window FIR filter is well-suited for EEG applications.

Detecting Instantaneous Tidal Signals in Ocean Models Utilizing Streaming Band‐Pass Filters

AbstractThrough the implementation of a streaming filter, output of numerical ocean simulations can be band‐pass filtered at tidal frequencies while the model is running, yielding time series of sinusoidal motions consisting of tidal signals in the filter's target frequency band. The filtering algorithm is developed from a system of two ordinary differential equations that represents the motion of a damped harmonic oscillator. The filter's response to a broadband input signal is unity at its target frequency but vanishes toward the low and high frequency limits. The decay of the filter response is controlled by a dimensionless parameter, which determines the filter's bandwidth. As a result, the filter allows signals within a small frequency band around its target frequency to pass through, while blocking signals outside of its target frequency band. In this work, the filtering algorithm is implemented into the barotropic solver of the Modular Ocean Model version 6 (MOM6) for determining the instantaneous tidal velocities of the semi‐diurnal and diurnal tides. Utilizing the filters, the frequency‐dependent internal wave drag is applied to the semi‐diurnal and diurnal frequency bands separately. The simulation results suggest that the performance of the algorithm is consistent with the filter transfer function in Fourier space. Potential applications of the algorithm also include de‐tiding the model output for nested regional ocean models, especially those for the purpose of operational forecasting.

Wireless optically pumped magnetometer MEG

The current magnetoencephalography (MEG) systems, which rely on cables for control and signal transmission, do not fully realize the potential of wearable optically pumped magnetometers (OPM). This study presents a significant advancement in wireless OPM-MEG by reducing magnetization in the electronics and developing a tailored wireless communication protocol. Our protocol effectively eliminates electromagnetic interference, particularly in the critical frequency bands of MEG signals, and accurately synchronizes the acquisition and stimulation channels with the host computer's clock. We have successfully achieved single-channel wireless OPM-MEG measurement and demonstrated its reliability by replicating three well-established experiments: The alpha rhythm, auditory evoked field, and steady-state visual evoked field in the human brain. Our prototype wireless OPM-MEG system not only streamlines the measurement process but also represents a major step forward in the development of wearable OPM-MEG applications in both neuroscience and clinical research.

Spectral Detection of a Weak Frequency Band Signal Based on the Pre-Whitening Scale Transformation of Stochastic Resonance in a Symmetric Bistable System in a Parallel Configuration

The spectral detection of weak frequency band signals poses a serious problem in many applications, especially when the target is within a certain frequency band under low signal-to-noise ratio (SNR) conditions. A kind of novel technique based on the pre-whitening scale transformation of stochastic resonance (SR) in a symmetric bistable system in a parallel configuration is proposed to solve the problem. Firstly, pre-whitening can ensure the Gaussian distribution of the receiving signal fits the requirements for SR processing. Secondly, scale transformation can help to effectively utilize the properties of a weak signal, especially under a low-frequency band. Thirdly, the SR in a symmetric bistable system in a parallel configuration can try to smoothly reduce the variances in the clutter and additive noise. Fourthly, by subtracting the steady state response of the SR in the selected symmetric bistable system from the parallel output, the spectral detection of a weak signal can be realized successfully. Experiment results based on actual sea clutter radar data guarantee the effectiveness and applicability of the proposed symmetric bistable PSR processing approach.

A Fault Diagnosis Method for Electric Vehicle Lithium Power Batteries Based on Dual Feature Extraction from the Time and Frequency Domains

Abstract This study focuses on the safety and reliability issues of lithium-ion batteries, proposing a fault diagnosis strategy that leverages dual feature extraction in the time-frequency domains. Additionally, by modifying the traditional autoencoder, the study proposes a feature-guided autoencoder as an unsupervised model for extracting features in the time domain. Initially, wavelet packet decomposition and its energy denoising treatment are employed to refine fault information within battery voltage signals. Subsequently, the reconstruction error outputted by the Feature-Guided Autoencoder is utilized as the time-domain fault feature, while the cosine similarity of the energy of signals in various frequency bands obtained after wavelet packet decomposition serves as the frequency-domain fault feature. Ultimately, this paper selects the Isolation Forest algorithm for two-dimensional outlier detection of time-frequency features. Experimental results demonstrate that the Feature-Guided Autoencoder proposed in this study not only enhances the sensitivity of time-domain fault features compared to traditional autoencoders and their variants but also optimizes issues related to training time and computational load. The effectiveness of the proposed dual feature fault diagnosis method in both the time and frequency domains is validated through data from two actual vehicles, showing superior early fault detection capability relative to single-feature fault diagnosis methods.

Frequency-Aware Divide-and-Conquer for Efficient Real Noise Removal.

Deep-learning-based approaches have achieved remarkable progress for complex real scenario denoising, yet their accuracy-efficiency tradeoff is still understudied, particularly critical for mobile devices. As real noise is unevenly distributed relative to underlay signals in different frequency bands, we introduce a frequency-aware divide-and-conquer strategy to develop a frequency-aware denoising network (FADN). FADN is materialized by stacking frequency-aware denoising blocks (FADBs), in which a denoised image is progressively predicted by a series of frequency-aware noise dividing and conquering operations. For noise dividing, FADBs decompose the noisy and clean image pairs into low-and high-frequency representations via a wavelet transform (WT) followed by an invertible network and recover the final denoised image by integrating the denoised information from different frequency bands. For noise conquering, the separated low-frequency representation of the noisy image is kept as clean as possible by the supervision of the clean counterpart, while the high-frequency representation combining the estimated residual from the successive FADB is purified under the corresponding accompanied supervision for residual compensation. Since our FADN progressively and pertinently denoises from frequency bands, the accuracy-efficiency tradeoff can be controlled as a requirement by the number of FADBs. Experimental results on the SIDD, DND, and NAM datasets show that our FADN outperforms the state-of-the-art methods by improving the peak signal-to-noise ratio (PSNR) and decreasing the model parameters. The code is released at https://github.com/NekoDaiSiki/FADN.

The influence of mental calculations on brain regions and heart rates

Performing mathematical calculations is a cognitive activity that can affect biological signals. This study aims to examine the changes in electroencephalogram (EEG) and electrocardiogram (ECG) signals of 36 healthy subjects during the performance of arithmetic tasks. To process EEG signals in different frequency bands, the energy and entropy of entropy (EoE) were extracted from the power spectrum and phase spectrum, respectively. Statistical analysis was conducted to determine meaningful features. These features were sent into support vector machine (SVM) and multi-layer perception (MLP) classifiers to assess their capability in separating math and rest classes. Results indicated the highest classification accuracy of 98.4% for classifying good counters in math and rest state using the MLP method. Based on the majority of features selected for each EEG channel, discriminative brain areas were identified. Analyzing EEG signals proved that math calculation may have multiple influences on various parts of the brain. By comparing good counters’ brain activities to those in a resting state, prominent changes were observed in the F4, C4, T4, T5, P3, and O2 areas. However, O1 and O2 channels showed significant changes in the brain of bad counters compared to the resting state. Considering ECG signals also demonstrated that during math calculation the number of heart rates per minute surpasses the rest state. These alterations can occur due to cognitive abilities or emotional processes which were observed to be prominent in subjects who performed more accurate calculations.

MOVING: A Multi-Modal Dataset of EEG Signals and Virtual Glove Hand Tracking.

Brain-computer interfaces (BCIs) are pivotal in translating neural activities into control commands for external assistive devices. Non-invasive techniques like electroencephalography (EEG) offer a balance of sensitivity and spatial-temporal resolution for capturing brain signals associated with motor activities. This work introduces MOVING, a Multi-Modal dataset of EEG signals and Virtual Glove Hand Tracking. This dataset comprises neural EEG signals and kinematic data associated with three hand movements-open/close, finger tapping, and wrist rotation-along with a rest period. The dataset, obtained from 11 subjects using a 32-channel dry wireless EEG system, also includes synchronized kinematic data captured by a Virtual Glove (VG) system equipped with two orthogonal Leap Motion Controllers. The use of these two devices allows for fast assembly (∼1 min), although introducing more noise than the gold standard devices for data acquisition. The study investigates which frequency bands in EEG signals are the most informative for motor task classification and the impact of baseline reduction on gesture recognition. Deep learning techniques, particularly EEGnetV4, are applied to analyze and classify movements based on the EEG data. This dataset aims to facilitate advances in BCI research and in the development of assistive devices for people with impaired hand mobility. This study contributes to the repository of EEG datasets, which is continuously increasing with data from other subjects, which is hoped to serve as benchmarks for new BCI approaches and applications.

SFT-Net: A Network for Detecting Fatigue From EEG Signals by Combining 4D Feature Flow and Attention Mechanism.

Fatigued driving is a leading cause of traffic accidents, and accurately predicting driver fatigue can significantly reduce their occurrence. However, modern fatigue detection models based on neural networks often face challenges such as poor interpretability and insufficient input feature dimensions. This article proposes a novel Spatial-Frequency-Temporal Network (SFT-Net) method for detecting driver fatigue using electroencephalogram (EEG) data. Our approach integrates EEG signals' spatial, frequency, and temporal information to improve recognition performance. We transform the differential entropy of five frequency bands of EEG signals into a 4D feature tensor to preserve these three types of information. An attention module is then used to recalibrate the spatial and frequency information of each input 4D feature tensor time slice. The output of this module is fed into a depthwise separable convolution (DSC) module, which extracts spatial and frequency features after attention fusion. Finally, long short-term memory (LSTM) is used to extract the temporal dependence of the sequence, and the final features are output through a linear layer. We validate the effectiveness of our model on the SEED-VIG dataset, and experimental results demonstrate that SFT-Net outperforms other popular models for EEG fatigue detection. Interpretability analysis supports the claim that our model has a certain level of interpretability. Our work addresses the challenge of detecting driver fatigue from EEG data and highlights the importance of integrating spatial, frequency, and temporal information.

Spatial Localization of Transformer Inspection Robot Based on Adaptive Denoising and SCOT-β Generalized Cross-Correlation.

In the detection process of the internal defects of large oil-immersed transformers, due to the huge size of large transformers and metal-enclosed structures, the positional localization of miniature inspection robots inside the transformer faces great difficulties. To address this problem, this paper proposes a three-dimensional positional localization method based on adaptive denoising and the SCOT weighting function with the addition of the exponent β (SCOT-β) generalized cross-correlation for L-type ultrasonic arrays of transformer internal inspection robots. Aiming at the strong noise interference in the field, the original signal is decomposed by an improved Empirical Mode Decomposition (EMD) method, and the optimal center frequency and bandwidth of each mode are adaptively searched. By extracting the modes in the frequency band of the positional localization signal, suppressing the modes in the noise frequency band, and reconstructing the Intrinsic Mode Function (IMF) of the independently selected superior modal components, a signal with a high signal-to-noise ratio is obtained. In addition, for the traditional mutual correlation algorithm with a large delay estimation error at a low signal-to-noise ratio, this paper adopts an improved generalized joint weighting function, SCOT-β, which improves the anti-jamming ability of the generalized mutual correlation method at a low signal-to-noise ratio by adding an exponential function to the denominator term of the SCOT weighting function's generalized cross-correlation. Finally, the accurate positional localization of the transformer internal inspection robot is realized based on the quadratic L-array and search-based maximum likelihood estimation method. Simulation and experimental results show the following: the improved EMD denoising method better improves the signal-to-noise ratio of the positional localization signal with a lower distortion rate; in the transformer test tank, which is 120 cm in length, 100 cm in width, and 100 cm in height, based on the positional localization method in this paper, the average relative positional localization error of the transformer internal inspection robot in three-dimensional space is 2.27%, and the maximum positional localization error is less than 2 cm, which meets the requirements of engineering positional localization.

EEG-Based Detection of Mild Cognitive Impairment Using DWT-Based Features and Optimization Methods.

In recent years, electroencephalography (EEG) has been investigated for identifying brain disorders. This technique involves placing multiple electrodes (channels) on the scalp to measure the brain's activities. This study focuses on accurately detecting mild cognitive impairment (MCI) from the recorded EEG signals. To achieve this, this study first introduced discrete wavelet transform (DWT)-based approaches to generate reliable biomarkers for MCI. These approaches decompose each channel's signal using DWT into a set of distinct frequency band signals, then extract features using a non-linear measure such as band power, energy, or entropy. Various machine learning approaches then classify the generated features. We investigated these methods on EEGs recorded using 19 channels from 29 MCI patients and 32 healthy subjects. In the second step, the study explored the possibility of decreasing the number of EEG channels while preserving, or even enhancing, classification accuracy. We employed multi-objective optimization techniques, such as the non-dominated sorting genetic algorithm (NSGA) and particle swarm optimization (PSO), to achieve this. The results show that the generated DWT-based features resulted in high full-channel classification accuracy scores. Furthermore, selecting fewer channels carefully leads to better accuracy scores. For instance, with a DWT-based approach, the full-channel accuracy achieved was 99.84%. With only four channels selected by NSGA-II, NSGA-III, or PSO, the accuracy increased to 99.97%. Furthermore, NSGA-II selects five channels, achieving an accuracy of 100%. The results show that the suggested DWT-based approaches are promising to detect MCI, and picking the most useful EEG channels makes the accuracy even higher. The use of a small number of electrodes paves the way for EEG-based diagnosis in clinical practice.

Chaotic Orthogonal Composite Sequence for 5G NR Time Service Signal Capture Algorithm

Establishing a national comprehensive PNT (Positioning, Navigation, and Timing) system has become a consensus among major countries worldwide. As a crucial component in completing the entire PNT system, the 5G NR (new radio) time service signal plays a vital role. This paper proposes a 5G NR time service signal that uses a spread spectrum system, shares the 5G signal frequency band, but does not occupy the bandwidth of the 5G communication signal. This timing service signal has relatively low power, making it appear “submerged” within the power of the 5G communication signal. The spread spectrum code for this timing signal employs the chaotic orthogonal composite sequence proposed in this paper. Compared to traditional spread spectrum sequences, this sequence offers better security than m-sequences, improved autocorrelation than Walsh sequences, and an effective suppression of the short-period characteristics exhibited when the Skew Tent-Map chaotic sequence takes special values. This paper simulates the capture of the 5G NR time service signal in an environment with a signal-to-noise ratio of 10 dB using an FFT-based parallel code phase search algorithm, successfully capturing the 5G NR time service signal and verifying the feasibility of the proposed chaotic orthogonal composite sequence as a spread spectrum code.

Newly identified Phonocardiography frequency bands for psychological stress detection with Deep Wavelet Scattering Network

The timely psychological stress detection can improve the quality of human life by preventing stress-induced behavioral and pathological consequences. This paper presents a novel framework that eliminates the need of Electrocardiography (ECG) signals-based referencing of Phonocardiography (PCG) signals for psychological stress detection. This stand-alone PCG-based methodology uses wavelet scattering approach on the data acquired from twenty-eight healthy adult male and female subjects to detect psychological stress. The acquired PCG signals are asynchronously segmented for the analysis using wavelet scattering transform. After the noise bands removal, the optimized segmentation length (L), scattering network parameters namely-invariance scale (J) and quality factor (Q) are utilized for computation of scattering features. These scattering coefficients generated are fed to K-nearest neighbor (KNN) and Extreme Gradient Boosting (XGBoost) classifier and the ten-fold cross validation-based performance metrics obtained are-accuracy 94.30 %, sensitivity 97.96 %, specificity 88.01 % and area under the curve (AUC) 0.9298 using XGBoost classifier for detecting psychological stress. Most importantly, the framework also identified two frequency bands in PCG signals with high discriminatory power for psychological stress detection as 270–290 Hz and 380–390 Hz. The elimination of multi-modal data acquisition and analysis makes this approach cost-efficient and reduces computational complexity.

Research on the Gunn Oscillation Effect of GaN HEMT with Field Plate Structure in the Terahertz Frequency Band

Based on the enormous application potential of GaN-based high electron mobility transistors (HEMT) in high-frequency and high-power scenarios, this article focuses mainly on the study of the Gunn oscillation effect of GaN-based HEMT devices. From the perspective of electric field regulation, a sandwich structure GaN HEMT device model with field plate structure is proposed, and a hydrodynamic physical model is established. The negative resistance characteristics in the GaN HEMT are obtained by the finite element method and the influence of the gate field plate on the Gunn oscillation frequency in the device channel is studied. The numerical simulation results show that the suitable field plate structure can modulate the distribution of the channel electric field below the gate, promote the electric field to enter the negative differential mobility region, undergo valley to valley electron transfer, form electron domains, and generate the Gunn oscillation currents in the terahertz band. Meanwhile, the length of the field plate regulates the oscillation current frequency of the device, and the stable and usable terahertz frequency band signal can be realized. This research opens up the possibility for semiconductor solid-state devices to realize terahertz frequency band radiation, and provides the basis for realizing new breakthroughs in HEMT for terahertz applications.

Classification of motor imagery using chaotic entropy based on sub-band EEG source localization

Objective. Electroencephalography (EEG) has been widely used in motor imagery (MI) research by virtue of its high temporal resolution and low cost, but its low spatial resolution is still a major criticism. The EEG source localization (ESL) algorithm effectively improves the spatial resolution of the signal by inverting the scalp EEG to extrapolate the cortical source signal, thus enhancing the classification accuracy. Approach. To address the problem of poor spatial resolution of EEG signals, this paper proposed a sub-band source chaotic entropy feature extraction method based on sub-band ESL. Firstly, the preprocessed EEG signals were filtered into 8 sub-bands. Each sub-band signal was source localized respectively to reveal the activation patterns of specific frequency bands of the EEG signals and the activities of specific brain regions in the MI task. Then, approximate entropy, fuzzy entropy and permutation entropy were extracted from the source signal as features to quantify the complexity and randomness of the signal. Finally, the classification of different MI tasks was achieved using support vector machine. Main result. The proposed method was validated on two MI public datasets (brain–computer interface (BCI) competition III IVa, BCI competition IV 2a) and the results showed that the classification accuracies were higher than the existing methods. Significance. The spatial resolution of the signal was improved by sub-band EEG localization in the paper, which provided a new idea for EEG MI research.