An essential technique for diagnosing brain disorders is electrophysiological source imaging (ESI). While modelbased optimization and deep learning methods have achieved promising results in this field, the accurate selection and refinement of features remains a central challenge for precise ESI. This paper proposes FAIR-ESI, a novel framework that adaptively refines feature importance across different views, including FFTbased spectral feature refinement, weighted temporal feature refinement, and self-attention-based patch-wise feature refinement. Extensive experiments on two simulation datasets with diverse configurations and two real-world clinical datasets validate our framework's efficacy, highlighting its potential to advance brain disorder diagnosis and offer new insights into brain function.

Medical time-series analysis is crucial for the diagnosis and treatment of various diseases. As modern medical sensors evolve, the complexity and dimensionality of medical signals have increased, posing challenges for data labeling and classification. Self-supervised learning has emerged as a promising solution by automatically extracting meaningful feature representations from raw data without manually annotated labels. However, existing research has often focused on the time-point or instance level input and overlook the intricate relationships among different channels, which are crucial for precise data representation. In this study, we propose CLHi-MTS, a novel Contrastive Learning-Based Hierarchical Framework for Masked Medical Time-Series Modeling that leverages three distinct levels of proximity: channel interdependencies, temporal fluctuations, and instance-level correlations, to guide the reconstruction of masked segments enhanced with contrastive learning. We also introduce a new augmentation strategy for generating positive examples in contrastive learning, further enhancing the model’s representation learning ability. Evaluated on five datasets for both in-domain and cross-domain medical time-series classification tasks, our framework outperforms six state-of-the-art self-supervised methods.

Magnetoencephalography (MEG) is a vital non-invasive tool for epilepsy analysis, as it captures high-resolution signals that reflect changes in brain activity over time. The automated detection of epileptic spikes within these signals can significantly reduce the labor and time required for manual annotation of MEG recording data, thereby aiding clinicians in identifying epileptogenic foci and evaluating treatment prognosis. Research in this domain often utilizes the raw, multi-channel signals from MEG scans for spike detection, commonly neglecting the multi-channel spiking patterns from spatially adjacent channels. Moreover, epileptic spikes share considerable morphological similarities with artifact signals within the recordings, posing a challenge for models to differentiate between the two. In this paper, we introduce a multimodal fusion framework that addresses these two challenges collectively. Instead of relying solely on the signal recordings, our framework also mines knowledge from their corresponding topography-map images, which encapsulate the spatial context and amplitude distribution of the input signals. To facilitate more effective data fusion, we present a novel multimodal feature fusion technique called CrossConvPyramid, built upon a convolutional pyramid architecture augmented by an attention mechanism. It initially employs cross-attention and a convolutional pyramid to encode inter-modal correlations within the intermediate features extracted by individual unimodal networks. Subsequently, it utilizes a self-attention mechanism to refine and select the most salient features from both inter-modal and unimodal features, specifically tailored for the spike classification task. Our method achieved the average F1 scores of 92.88% and 95.23% across two distinct real-world MEG datasets from separate centers, respectively outperforming the current state-of-the-art by 2.31% and 0.88%. We plan to release the code on GitHub later.

A Magnetoencephalography (MEG) time-series recording consists of multi-channel signals collected by superconducting sensors, with each signal’s intensity reflecting magnetic field changes over time at the sensor location. Automating epileptic MEG spike detection significantly reduces manual assessment time and effort, yielding substantial clinical benefits. Existing research addresses MEG spike detection by encoding neural network inputs with signals from all channel within a time segment, followed by classification. However, these methods overlook simultaneous spiking occurred from nearby sensors. We introduce a simple yet effective paradigm that first clusters MEG channels based on their sensor’s spatial position. Next, a novel convolutional input module is designed to integrate the spatial clustering and temporal changes of the signals. This module is fed into a custom MEEG-ResNet3D developed by the authors, which learns to extract relevant features and classify the input as a spike clip or not. Our method achieves an F1 score of 94.73% on a large real-world MEG dataset Sanbo-CMR collected from two centers, outperforming state-of-the-art approaches by 1.85%. Moreover, it demonstrates efficacy and stability in the Electroencephalographic (EEG) seizure detection task, yielding an improved weighted F1 score of 1.4% compared to current state-of-the-art techniques evaluated on TUSZ, whch is the largest EEG seizure dataset.

Data-driven approaches are promising to address the modeling issues of modern power electronics-based power systems, due to the black-box feature. Frequency-domain analysis has been applied to address the emerging small-signal oscillation issues caused by converter control interactions. However, the frequency-domain model of a power electronic system is linearized around a specific operating condition. It thus requires measurement or identification of frequency-domain models repeatedly at many operating points (OPs) due to the wide operation range of the power systems, which brings significant computation and data burden. This article addresses this challenge by developing a deep learning approach using multilayer feedforward neural networks (FNNs) to train the frequency-domain impedance model of power electronic systems that is continuous of OP. Distinguished from the prior neural network designs relying on trial-and-error and sufficient data size, this article proposes to design the FNN based on latent features of power electronic systems, i.e., the number of system poles and zeros. To further investigate the impacts of data quantity and quality, learning procedures from a small dataset are developed, and K-medoids clustering based on dynamic time warping is used to reveal insights into multivariable sensitivity, which helps improve the data quality. The proposed approaches for the FNN design and learning have been proven simple, effective, and optimal based on case studies on a power electronic converter, and future prospects in its industrial applications are also discussed.

Current control of grid-connected converters may result in harmonic instability when grid impedance changes. To prevent this issue, current controller parameters can be tuned adaptively according to different short-circuit ratios (SCRs). It is thus important to estimate the grid impedance in real-time. Unlike traditional FFT-based impedance measurement methods, a more efficient estimation approach based on neural networks is proposed in this paper. This method does not require a fixed and relatively long sampling window, making it possible for real-time impedance measurement. Further, a step-by-step design method of the feedforward neural network (FNN) used for grid impedance estimation is developed. Time-domain simulation results validate the effectiveness of the approach. Based on the designed FNN, adaptive current control is implemented and verified through simulation.

Multi-sampling control provides an attractive way to reduce the control delays in LCL grid connected inverters, then the bandwidth and stability margin will be improved. However, high frequency switching harmonics (SHs) are introduced in the control loop, which may cause low frequency aliasing and deteriorate the power quality. In this paper, the multi-sampled currents are reconstructed through the double-sampled average currents, which is simple and easy to implement in a microprocessor. As a result, most of the SHs are removed and the power quality is better than the conventional multi-sampling control. Finally, a single-loop pseudo-derivative-feedback control with inverter current feedback is employed, and the effectiveness of the proposed method is verified through simulation.

Due to the increasing performance and decreasing price of microcontrollers, applying a high sampling frequency becomes more feasible in modern control. This approach is known as multisampling technology. This paper tries to give an overview of the multisampling application in power electronics converters as it enables a lot of opportunities. First, the multisampling can reduce control delays and thereby break bandwidth limitations in power electronic controllers. Secondly, a smarter control strategy can be realized with more current, voltage and system information acquired through multisampling techniques. However, this approach may also introduce two major drawbacks: risk of dead bands and high-frequency switching ripples, which may degrade the system stability margin. To address these issues, different solutions are explored and illustrated in this paper.

This paper proposes a physics-level modeling method for analyzing the primary to secondary-side (common-mode) parasitic capacitance of the transformer for the Medium-voltage SiC MOSFETs gate drivers. The lumped circuit-based physics-model of the turn-to-turn capacitance, turn-to-core capacitance, and self capacitance of the core are derived, and it is found that the turn-to-core capacitance mainly contributes to the total equivalent parasitic common-mode capacitance. The measured common-mode impedance of the transformer shows high agreements with the calculated value, where the accuracy of the proposed modeling method can be proved based on the experimental results.

In this paper, an eigenvalue-based harmonic stability analysis method for offshore wind farm is proposed. Considering the internal cable connection layout, a component connection method (CCM) is adopted to divide the system into individual blocks as current controller of converters, LCL filters, collection system cable impedance, transmission cable impedance and grid impedance. By using this method, each block can be modelled independently and then be integrated into a whole state space matrix which contains sparse and diagonal matrix. Compared with the traditional state space formulation process, this method is superior in reducing the computational cost. A 4 by 4 wind turbines wind farm is selected as the study case and the proposed method is validated by comparing the results obtained from time domain simulation software - Power System Computer Aided Design (PSCAD).

Continuously expanding deployments of distributed power-generation systems (DPGSs) are transforming the conventional centralized power grid into a mixed distributed electrical network. The modern power grid requires flexible energy utilization but presents challenges in the case of a high penetration degree of renewable energy, among which wind and solar photovoltaics are typical sources. The integration level of the DPGS into the grid plays a critical role in developing sustainable and resilient power systems, especially with highly intermittent renewable energy resources. To address the challenging issues and, more importantly, to leverage the energy generation, stringent demands from both utility operators and consumers have been imposed on the DPGS. Furthermore, as the core of energy conversion, numerous power electronic converters employing advanced control techniques have been developed for the DPGS to consolidate the integration. In light of the above, this paper reviews the power-conversion and control technologies used for DPGSs. The impacts of the DPGS on the distributed grid are also examined, and more importantly, strategies for enhancing the connection and protection of the DPGS are discussed.

Compared with conventional dc-link drive, slim dc-link drive is expected to achieve lower cost and longer life time. However, harmonics distortion problem may occur in such drive systems. This paper proposes to use an active damping control method to suppress the harmonic distortion with the benefit of low cost and also low loss. A new analysis method, based on the frequency domain impedance model, is presented to explore the mechanism of harmonics suppression. Also, a general method is presented to build the impedance model of a PMSM drive system using Field Oriented Control (FOC) method. Some design issues, including power levels, current control bandwidth and harmonic interaction, are discussed when the drive system is fed by a weak grid. Case studies on a two-drive system composed by two slim dc-link drive units are provided to verify the proposed analysis method.

This paper presents a new method to measure the phase dependent impedance from an experimental set up. Though most of power electronics based system is gradually migrating to IGBT based voltage source converter due to their controllability, the rectifier composed of diode or thyristor components are still widely used in AC-DC applications because of their cost effectiveness and reliability. However, these topologies generate harmonic problems in their network due to their switching instant variation caused by the frequency and phase of grid voltage. Hence, a lot of solutions have been proposed to provide an optimized solution for better power quality. However, the phase dependent impedance, which is driven by both switching instant variation and frequency coupling between ac and dc network, has not been treated in the design of passive filter or harmonic compensator design for power electronics application. It is found that the phase dependent impedance shows different properties with the impedance profiles, which have been proposed in the research. This paper explains a method to measure the phase dependent impedance profile from an experimental set up. Furthermore, the results are compared with results from time-domain simulations and results from an analytical model developed in the Harmonic State-Space (HSS).

The present paper addresses the AC microgrid control issue using the hierarchical control structure and droop controllers for load sharing. Once the droop controllers impose an operation with frequency and voltage deviations, depending on the load and droop parameters, a hierarchical control structure must be added to change the droop controller operating points. The hierarchical controllers operate with local measurements and shared signals from communication links among the distributed generation systems connected to the microgrid. Depending on the geographical size of the microgrid, the communication links can be economically unviable. This paper thus proposes a fuzzy secondary controller for AC microgrids to reduce the link communication dependency by using only local measurements. The simulation results show that the deviations as happened with the conventional secondary controllers can be compensated when the microgrid is operating as stand-alone mode, and can provide a certain amount of power to the connected utility bus when operating as grid-connected mode.

This paper addresses a state-space-based harmonic stability analysis of paralleled grid-connected inverters system. A small signal model of individual inverter is developed, where LCL filter, the equivalent delay of control system, and current controller are modeled. Then, the overall small signal model of paralleled grid-connected inverters is built. Finally, the state-space-based stability analysis approach is developed to explain the harmonic resonance phenomenon. The eigenvalue traces associated with time delay and coupled grid impedance are obtained, which accounts for how the unstable inverter produces the harmonic resonance and leads to the instability of whole paralleled system. The proposed approach reveals the contributions of the grid impedance as well as the coupled effect on other grid-connected inverters under different grid conditions. Simulation and experimental results are provided to verify the proposed harmonic stability assessment method.

This paper presents an eigenvalue-based harmonic stability analysis method for inverter-fed power systems. A full-order small-signal model for a droop-controlled Distributed Generation (DG) inverter is built first, including the time delay of digital control system, inner current and voltage control loops, and outer droop-based power control loop. Based on the inverter model, an overall small-signal model of a two-inverter-fed system is then established, and the eigenvalue-based stability analysis is subsequently performed to assess the influence of controller parameters on the harmonic resonance and instability in the power system. Eigenvalues associated with time delay of inverter and inner controller parameters is obtained, which shows the time delay has an important effect on harmonic instability of inverter-fed power systems. Simulation results are given for validating the proposed harmonic stability analysis method.

This paper addresses the harmonic compensation error problem existing with parallel connected inverter in the same grid interface conditions by means of impedance-based analysis and modeling. Unlike the single grid connected inverter, it is found that multiple parallel connected inverters and grid impedance can make influence to each other if they each have a harmonic compensation function. The analysis method proposed in this paper is based on the relationship between the overall output impedance and input impedance of parallel connected inverter, where controller gain design method, which can minimize the compensation error under various conditions, is also proposed.