An Approach for EEG Denoising Based on Wasserstein Generative Adversarial Network.

Electroencephalogram (EEG) recordings often contain artifacts that would lower signal quality. Many efforts have been made to eliminate or at least minimize the artifacts, and most of them rely on visual inspection and manual operations, which is time/labor-consuming, subjective, and incompatible to filter massive EEG data in real-time. In this paper, we proposed a deep learning framework named Artifact Removal Wasserstein Generative Adversarial Network (AR-WGAN), where the well-trained model can decompose input EEG, detect and delete artifacts, and then reconstruct denoised signals within a short time. The proposed approach was systematically compared with commonly used denoising methods including Denoised AutoEncoder, Wiener Filter, and Empirical Mode Decomposition, with both public and self-collected datasets. The experimental results proved the promising performance of AR-WGAN on automatic artifact removal for massive data across subjects, with correlation coefficient up to 0.726±0.033, and temporal and spatial relative root-mean-square error as low as 0.176±0.046 and 0.761±0.046, respectively. This work may demonstrate the proposed AR-WGAN as a high-performance end-to-end method for EEG denoising, with many on-line applications in clinical EEG monitoring and brain-computer interfaces.

Using microneedle array electrodes for non-invasive electrophysiological signal acquisition and sensory feedback evoking.

Introduction: Bidirectional transmission of information is needed to realize a closed-loop human-machine interaction (HMI), where electrophysiological signals are recorded for man-machine control and electrical stimulations are used for machine-man feedback. As a neural interface (NI) connecting man and machine, electrodes play an important role in HMI and their characteristics are critical for information transmission. Methods: In this work, we fabricated a kind of microneedle array electrodes (MAEs) by using a magnetization-induced self-assembly method, where microneedles with a length of 500-600 µm and a tip diameter of ∼20 µm were constructed on flexible substrates. Part of the needle length could penetrate through the subjects' stratum corneum and reach the epidermis, but not touch the dermis, establishing a safe and direct communication pathway between external electrical circuit and internal peripheral nervous system. Results: The MAEs showed significantly lower and more stable electrode-skin interface impedance than the metal-based flat array electrodes (FAEs) in various testing scenarios, demonstrating their promising impedance characteristics. With the stable microneedle structure, MAEs exhibited an average SNR of EMG that is more than 30% higher than FAEs, and a motion-intention classification accuracy that is 10% higher than FAEs. The successful sensation evoking demonstrated the feasibility of the MAE-based electrical stimulation for sensory feedback, where a variety of natural and intuitive feelings were generated in the subjects and thereafter objectively verified through EEG analysis. Discussion: This work confirms the application potential of MAEs working as an effective NI, in both electrophysiological recording and electrical stimulation, which may provide a technique support for the development of HMI.

Investigation of Fracture Behavior and Mechanism in High-Speed Precise Shearing for Metal Bars with Prefabricated Fracture-Start Kerfs.

A laser-assisted high-speed shearing (LAHSS) method has been proposed for metal bars, which prefabricates equally spaced fracture-start kerfs by Nd:Yag laser to make stress concentration, and applies a high-speed load to complete fracture separation. Comparative tests were conducted for Q235, 40Cr, and 304 steel bars, and the effects of fracture-start kerfs and axial clearance were investigated on the fracture section. Moreover, the fracture behavior was demonstrated by numerical simulation, and the micro-fracture mechanism was revealed by fractographic analysis. The numerical simulation results show that the material damage concentrates along with the kerf tips with peak equivalent plastic strain, and the corresponding stress triaxiality drops to almost zero at the kerf tip, which reveals that the material is subjected to pure shearing at kerf tip; the Max. loading force is reduced by 15.2%-29.6%, and the impact energy is decreased by 29.8%-46.9% for the three types of bar material. The experimental results showed that the fracture-start kerfs effectively inhibited the plastic deformation stage, and higher precision blanks were obtained in the LAHSS test: roundness error improved from 2.7%-10.9% to 1.1%-2.6%, Max. bending deflection decreased from 1.3-3.4 mm to 0.4-1.0 mm, and flatness error dropped from 0.9-3.3 mm to 0.3-0.7 mm. The fractographic analysis reveals that the crack initiation is related to alternative V-shape micro-notches at the laser-affected zone; the predominant fracture mechanism involves mode II microvoid coalescence at the main fracture plane; smaller and less elongated dimples were formed in 40Cr steels due to higher number density of grains and pinning effect of second-phase particles compared to Q235 and 304 steel bars.

Investigation of Notch-Induced Precise Splitting of Different Bar Materials under High-Speed Load.

A notch-induced high-speed splitting method was developed for high-quality cropping of metal bars using a new type of electric-pneumatic counter hammer. Theoretical equations and FE models were established to reveal the crack initiation and fracture mode. Comparative tests were conducted for notched and unnotched bars of four types of steels, i.e., AISI 1020, 1045, 52100, and 304, and the section quality and microfracture mechanism were further investigated. The results show that damage initiates at the bilateral notch tips with peak equivalent plastic strain, and propagates through the plane induced by the notch tip; the stress triaxiality varies as a quasi-sine curve, revealing that the material is subjected to pure shearing at the notch tip, and under compression at the adjacent region. High precision chamfered billets were obtained with roundness errors of 1.1-2.8%, bending deflections of 0.5-1.5mm, and angles of inclination of 0.7°-3.4°. Additionally, the notch effectively reduced the maximum impact force by 21.6-23.9%, splitting displacement by 7.6-18.6%, and impact energy by 27.8-39.1%. The crack initiation zone displayed quasi-parabolic shallow dimples due to shear stress, and the pinning effect was larger in AISI 52100 and 1045 steel; the final rupture zone was characterized by less elongated and quasi-equiaxial deeper dimples due to the combination of shear and normal stress.