In this paper, a novel remedial direct torque control (DTC) scheme is proposed for T type NPC (T- NPC) three-level asymmetric six-phase PMSM drives under open-phase faults. A simplified space vector modulation (SVM) is designed and... more
In this paper, a novel remedial direct torque control (DTC) scheme is proposed for T type NPC (T- NPC) three-level asymmetric six-phase PMSM drives under open-phase faults. A simplified space vector modulation (SVM) is designed and applied for the DTC controlled asymmetric six-phase drive, in such a way that both good current harmonic performance and small DC link voltage fluctuation are available. The relationship between the stator fluxes and the stator voltages including the faulty phase is derived after the fault. Thus, the perturbation could be added to the voltage references, and the control system under is kept almost unchanged after fault. The experiments on a laboratory prototype are carried out to verify that the proposed remedial scheme can provide smooth torque while good harmonic performance without changing the control structure.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
In this paper, we consider the problem of formation control of multiple unmanned aerial vehicles (MUAVs). Based on a self-organizing approach, we propose a solution for both circle and arbitrary polygon formations. The proposed solution... more
In this paper, we consider the problem of formation control of multiple unmanned aerial vehicles (MUAVs). Based on a self-organizing approach, we propose a solution for both circle and arbitrary polygon formations. The proposed solution is a distributed one only using the neighbouring UAV information. Simulation is conducted to verify two types of the formations as well as UAV failure and re-formation process.
Research Interests:
Research Interests:
The design experience in the Shenzhen side engineering of Shenzhen-Hong Kong western corridor is introduced,including the integrated design of pipelines at Dongbin Road and the design of drainage system of the subway which is the longest... more
The design experience in the Shenzhen side engineering of Shenzhen-Hong Kong western corridor is introduced,including the integrated design of pipelines at Dongbin Road and the design of drainage system of the subway which is the longest urban subway up to now in China.All these can be used for reference in design of similar engineering.
Research Interests:
ABSTRACT In this experiment, we record the EEG data during the sleeping and the awakening for the drowsiness cognition. As is well known, analyzing the frequency components of the EEG is important for the sleeping. There are many studies... more
ABSTRACT In this experiment, we record the EEG data during the sleeping and the awakening for the drowsiness cognition. As is well known, analyzing the frequency components of the EEG is important for the sleeping. There are many studies to analyze the EEG frequency data for recognizing the sleeping quality, and so on. However, there is no established method for the analysis of the sleeping EEG. From these reason, we are going to apply the rhythmic component extraction (RCE), proposed by Tanaka et al., to extract the rhythmic component in the brain. RCE finds a component extracted by using a weighted sum of observed channel signals. The weights are optimized by maximizing the power in a certain frequency range of interest. In the experiment, the subjects lie back, relax in a chair, and sleep. We take the EEG recording at 30 channels. From the results, the EEG features, such as the alpha and the theta wave, are different between the sleeping and the waking. These rhythmic components are extracted by the Fourier transform and the RCE. By using the RCE, we confirmed the alpha wave and the theta wave when they are not extracted in a single channel signal. Furthermore, we confirmed the effective measurement locations by analyzing the RCE weights.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Locomotion of animals is highly coordinating, efficient and full of maneuverability. Such outstanding and unique characteristics are acquired through millions of years of evolution. It is highly desirable to enhance robots with such... more
Locomotion of animals is highly coordinating, efficient and full of maneuverability. Such outstanding and unique characteristics are acquired through millions of years of evolution. It is highly desirable to enhance robots with such characteristics, which is one of the ultimate objectives of biomimetic research. To achieve this goal, in this paper, a biomimetic learning approach for locomotion generation of bionic robots is proposed. The main feature of the learning approach is adopting multiple general internal models (GIMs) to learn and regenerate coordinated animal behaviors. This work discusses the basic mechanism of the proposed GIM-based learning approach. Moreover, two cases with different bionic robots are studied to explore the effectiveness and the generality of the proposed approach. This paper summarizes the recent development of biomimetic learning. It is noted that this work is devoted to the contribution of Late Professor Jian-xin Xu in this area.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Intracranial electroencephalogram (iEEG) recorded at cerebral cortex contains a lot of important information for the diagnosis of epilepsy. Currently, the diagnosis of epilepsy must be performed by multiple clinical experts through visual... more
Intracranial electroencephalogram (iEEG) recorded at cerebral cortex contains a lot of important information for the diagnosis of epilepsy. Currently, the diagnosis of epilepsy must be performed by multiple clinical experts through visual judgment on the long term interictal iEEG signals. However, it is a time consuming and extremely difficult process. In this paper, we introduce the feature extraction method based on the several different entropies evaluated on the different frequency bands, which can thus be formed as a 2D feature map. Then, we employ the convolutional neural network (CNN) to train a binary classifier based on the labels provided by clinical experts. The experimental results on public benchmark and real-world iEEG recorded from patients demonstrate that our method can achieve 99.0% classification performance. Hence, it is a promising technique to reduce the workload of clinical experts for automatic detection of epileptic focal.
Research Interests:
Abstract Static models for a single row and a double row four-point contact pitch bearing, taking into account the clearances therein, were presented. A precise computation method was proposed for the static load-carrying capacity curves... more
Abstract Static models for a single row and a double row four-point contact pitch bearing, taking into account the clearances therein, were presented. A precise computation method was proposed for the static load-carrying capacity curves of such pitch bearings, which can be used to pre-select pitch bearings and slewing bearings supposing rigid bearing rings. The effects on the static load-carrying capacity induced by changing the clearance, the raceway groove radius of curvature, and initial contact angle were analysed. The clearance has a significant effect on the static load-carrying capacity of the bearing only in load cases with a rather small axial force and a large tilting moment. When the coefficient of raceway groove curvature radius increases, the static load-carrying capacity decreases. The smaller the radial load, the more significant the effect of the coefficient of raceway groove curvature radius on the static load-carrying capacity of the bearing. When radial loads range from 0 to 800 kN, the load-carrying capacity increases with increasing initial contact angle. When the radial loads are above 800 kN, the load-carrying capacity decreases with increasing initial contact angle.
Research Interests:
Dielectric elastomer actuators (DEAs) exhibit interesting muscle-like attributes including large voltage-induced deformation and high energy density, thus can function as artificial muscles for soft robots/devices. This paper focuses on... more
Dielectric elastomer actuators (DEAs) exhibit interesting muscle-like attributes including large voltage-induced deformation and high energy density, thus can function as artificial muscles for soft robots/devices. This paper focuses on soft planar DEAs, which have extensive applications such as artificial muscles for jaw movement, stretchers for cell mechanotransduction, and vibration shakers for tactile feedback, etc. Specifically, we develop a soft planar DEA, in which compression springs are employed to make the entire structure freestanding. This soft freestanding actuator can achieve both linear actuation and turning without increasing the size, weight, or structural complexity, which makes the actuator suitable for driving a soft crawling robot. Furthermore, its simple structure and homogeneous deformation allow for analytic modeling, which can be used to interpret the large voltage-induced deformation and interesting mechanics phenomenon (i.e., wrinkling and electromechanical instability). A preliminary demonstration showcases that this soft planar actuator can be employed as an artificial muscle to drive a soft crawling robot.
Research Interests:
Electroencephalogram (EEG) recording is relatively safe for the patients who are in deep coma or quasi brain death, so it is often used to verify the diagnosis of brain death in clinical practice. The objective of this paper is to apply... more
Electroencephalogram (EEG) recording is relatively safe for the patients who are in deep coma or quasi brain death, so it is often used to verify the diagnosis of brain death in clinical practice. The objective of this paper is to apply deep learning method to EEG signal analysis in order to confirm clinical brain death diagnosis. A novel approach using spectrogram images produced from EEG signals as the input dataset of Convolution Neural Network (CNN) is proposed in this paper. A deep CNN was trained to obtain the similarity degree of the patients’ EEG signals with the clinical diagnosed symptoms. This method can evaluate the condition of the brain damage patients and can be a reliable reference of quasi brain death diagnosis.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Although brain death has been accepted by most countries, how to diagnose brain death quickly and accurately is still a very challenging task. Electroencephalography (EEG) is considered to be one of the most effective methods for... more
Although brain death has been accepted by most countries, how to diagnose brain death quickly and accurately is still a very challenging task. Electroencephalography (EEG) is considered to be one of the most effective methods for clinically diagnosing the brain state. In the present study, we investigated if it is possible to find a robust neuro-marker to help doctors diagnose brain death from the perspective of brain information interaction. With 6-channel EEG data collected from prefrontal lobe of 30 patients (deep coma: 13, brain death 17), PLV was used to measure the phase synchronization and quantify the interaction between brain regions. Firstly, we found that there exists significant difference in brain synchronization between brain death patients and deep coma patients. Then, an interesting phenomenon was observed. In all the three low frequency bands (delta, theta, alpha), synchronization between the left hemisphere and the right hemisphere (IHPS) is significantly stronger than that only in left (LHPS) or right hemisphere (RHPS) for deep coma patients. However, this phenomenon has not been found in high frequency bands (beta and gamma) and for brain death patients. More importantly, it was verified for almost every patient (only 1 exception). This phenomenon might provide a robust neuro-marker for brain death diagnosis. Finally, we tried to distinguish between brain death and brain coma by monitoring brain synchronization in real time. The results also show the effectiveness and accuracy of our method. The method proposed in this paper can not only provide doctors with more stable and effective information, but also hope to develop a new visualization tool to assist brain death diagnosis.
Research Interests:
The five components of a stream corridor ecological restoration project, including planning, implementation, performance assessment, adaptive management, and product dissemination, were introduced. Based on the summary of experiences by... more
The five components of a stream corridor ecological restoration project, including planning, implementation, performance assessment, adaptive management, and product dissemination, were introduced. Based on the summary of experiences by United States Environment Protection Agency, the principles for the ecological restoration including stream corridor, were explained in detail.
Research Interests:
Disturbances in process plants may usually widely propagate because of the interconnection in process equipment. It is necessary to get the correct source of loop oscillations. The application of transfer entropy method has now been... more
Disturbances in process plants may usually widely propagate because of the interconnection in process equipment. It is necessary to get the correct source of loop oscillations. The application of transfer entropy method has now been proved effective. These existing methods need too much process knowledge and the results are affected by different parameters. In this work, spectral independent component analysis (Spectral ICA) are used to select the oscillatory process loop variables and reduce the number of variables which need to be analyzed by transfer entropy method, then a normalized transfer entropy method with non-parametric is used to isolate the root-cause of plant-wide and identify the propagation paths. The successful application of the methods has been demonstrated through two cases.
Research Interests:
Research Interests:
Research Interests:
In this paper, we propose a tensor-based non-local filtering technique for image and MRI denoising using Bayesian CP factorization (BCPF). This approach simply groups together similar sub-tensors (e.g., 3D tensors) selected from a noisy... more
In this paper, we propose a tensor-based non-local filtering technique for image and MRI denoising using Bayesian CP factorization (BCPF). This approach simply groups together similar sub-tensors (e.g., 3D tensors) selected from a noisy tensor and forms a 4D stack, then decomposes this stack into latent factors by employing BCPF, resulting in a filtered group of 3D sub-tensors. This procedure is repeated across the entire tensor in sliding window fashion to obtain the denoised result of original tensor. Our Bayesian CP factorization can learn CP-rank as well as noise variance solely from the observed noisy tensor data, which can also avoid overfitting problem by employing a fully Bayesian treatment for latent factor inference. The main advantage of our method is that the standard deviation of Gaussion noise can be automatically inferred and not necessary to be fixed. The experimental results on image and MRI denoising demonstrate the superiorities of our method in terms of flexibility and performance, as compared to other tensor-based denoising methods.
Research Interests:
This paper presents a novel algorithm of spatial and temporal equalization for multiuser detection with an antenna array based on a robust subspace method and a self-adaptive step-size training technique. Our robust approach includes two... more
This paper presents a novel algorithm of spatial and temporal equalization for multiuser detection with an antenna array based on a robust subspace method and a self-adaptive step-size training technique. Our robust approach includes two procedures. In the first procedure, a robust subspace technique is utilized to reduce the high-power of additive noise and the correlation among the channels. Moreover,
Research Interests:
Brain Computer Interface (BCI) aims to translate the brain signals, reflecting the neural activities of brain evoked by external stimuli or mental tasks, into the corresponding commands, which thus provides a direct communication between... more
Brain Computer Interface (BCI) aims to translate the brain signals, reflecting the neural activities of brain evoked by external stimuli or mental tasks, into the corresponding commands, which thus provides a direct communication between human brain and machine. P300 based BCI has demonstrated to be one of the most reliable and subject independent paradigm. However, the existing P300 based BCI only uses single modality, i. e., visual stimuli evoked potential. In this paper, to further improve the reliability of BCI system, we develop a hybrid BCI by using auditory and visual stimulus simultaneously. Experimental results demonstrate that the event-related potentials evoked by hybrid stimulus are significantly different with single visual evoked potentials, and our hybrid BCI shows more reliable performance than the traditional P300 BCI system.
Research Interests:
EEG (electroencephalography) can be analyzed quantitatively by introducing the EEG energy index in EEG preliminary examination of determination of brain death. EMD (empirical mode decomposition), MEMD (multivariate empirical mode... more
EEG (electroencephalography) can be analyzed quantitatively by introducing the EEG energy index in EEG preliminary examination of determination of brain death. EMD (empirical mode decomposition), MEMD (multivariate empirical mode decomposition), and 2T-EMD (turning tangent empirical mode decomposition) can be used to analyze coma and quasi-brain-death patients' EEG energy. In this paper, EMD, MEMD, and 2T-EMD are compared from algorithm principle and experimental aspects in order to obtain the optimal algorithm for supporting determination of brain death, where experiments are carried out based on standard artificial signals and patients' EEG. The analysis results illustrate that 2T-EMD is the optimal algorithm with the relative superior computational performance for both single and multi-channel signals.
Research Interests:
Herein, we report the crucial importance of C‐defective sites on the CO adsorption over ϵ‐Fe2C and η‐Fe2C Fischer‐Tropsch catalysts via systematic DFT calculations. The simulated XRD and Wulff construction show the significant differences... more
Herein, we report the crucial importance of C‐defective sites on the CO adsorption over ϵ‐Fe2C and η‐Fe2C Fischer‐Tropsch catalysts via systematic DFT calculations. The simulated XRD and Wulff construction show the significant differences in their equilibrium shapes and most exposed surfaces. It is observed that the ϵ‐Fe2C exposes a high proportion (89 %) of facets (1 1) with similar structure to that of η‐Fe2C (011) which has been proved to be the active surface of CO activation.
Research Interests:
Research Interests:
Research Interests: Mathematics, Statistics, Biomedical Engineering, Multivariate Statistics, Medicine, and 11 moreMultivariate Analysis, Discriminant Analysis, Humans, Linear Discriminant Analysis, Female, Male, Differential Diagnosis, Electromyography, Indexation, Electrical And Electronic Engineering, and Motor neuron disease
Research Interests:
Research Interests:
As the novel suspension bearing, Magnetic-Liquid Double Suspension Bearing (MLDSB) is mainly supported by magnetic suspension and supplemented by a liquid hydrostatic bearing. Due to its great bearing capacity and stiffness, rapid... more
As the novel suspension bearing, Magnetic-Liquid Double Suspension Bearing (MLDSB) is mainly supported by magnetic suspension and supplemented by a liquid hydrostatic bearing. Due to its great bearing capacity and stiffness, rapid response, great active control, and so on, MLDSB is suitable for medium speed heavy loads and has a large carrying capacity and high operating stability. In addition, the radial inertia coupling and gyroscopic coupling between radial 4-DOF control channels can reduce control precision, operation stability, and reliability of MLDSB. Therefore, a mathematical model of radial 4-DOF rotor-dynamics of MLDSB is established in this paper, and the inherent coupling mechanism is explored. Taking inertial coupling, gyroscopic coupling, and external disturbance loads as lumped disturbances, a decoupled controller based on Generalized Extended State Observer (GESO) is established. The influence of the GESO controller on the decoupling and control performance of radial 4-DOF control channels is simulated. The results indicate that the decoupling effect of the GESO controller is great. Under the action of step signal, the steady displacement, maximum displacement, adjustment time, and peak time of the rotor after decoupling are all reduced, among which the steady displacement and maximum displacement are the most obvious. Under the sinusoidal signal, the steady displacement and maximum displacement are reduced by 90%, which can effectively avoid the “gap-impact” fault. Under the pulse signal, the steady displacement, maximum displacement, adjustment time, and peak time are all reduced, among which the maximum displacement is the most obvious. The research in this paper can provide a theoretical reference for the stable support and decoupling control of MLDSB.
Research Interests:
In real-time electroencephalography (EEG) analysis, the problem of observing dynamic changes and the problem of binary classification is a promising direction. EEG energy and complexity are important evaluation metrics in brain death... more
In real-time electroencephalography (EEG) analysis, the problem of observing dynamic changes and the problem of binary classification is a promising direction. EEG energy and complexity are important evaluation metrics in brain death determination in the field of EEG analysis. We developed two algorithms, dynamic turning tangent empirical mode decomposition to compute EEG energy and dynamic approximate entropy to compute EEG complexity for brain death determination. The developed algorithm is applied to analyze 50 EEG data of coma patients and 50 EEG data of brain death patients. The validity of the dynamic analysis is confirmed by the accuracy rate derived from the comparison with turning tangent empirical mode decomposition and approximate entropy algorithms. We evaluated the EEG data of three patients using the built diagnostic system. The experimental results visually showed that the EEG energy ratio was higher in a coma state than that in brain death, while the complexity was lower than that in brain death.
Research Interests:
To improve the reliability of perception, autonomous mobile robots often obtain environmental information from multiple sensors. However, the redundancy of sensors and extra fusion process increase the risks of system failure. In this... more
To improve the reliability of perception, autonomous mobile robots often obtain environmental information from multiple sensors. However, the redundancy of sensors and extra fusion process increase the risks of system failure. In this paper, a fault-tolerance architecture is proposed for mobile robot localization and a differential drive mobile robot is investigated. In the architecture, the relative/absolute localization methods are fused by Extended Kalman Filters (EKFs). Furthermore, fault detection and fault identification are accomplished by comparing the outputs of redundancy of fusing processes. Finally, the effectiveness of the fault-tolerance architecture is verified in several experiments conducted in the robot prototype.
Research Interests:
Soft actuators play an important role in producing motions in soft robots, and dielectric elastomers have shown great promise because of their considerable voltage-induced deformation. In particular, air-filled dielectric elastomer... more
Soft actuators play an important role in producing motions in soft robots, and dielectric elastomers have shown great promise because of their considerable voltage-induced deformation. In particular, air-filled dielectric elastomer actuators have been well studied, where the air inside provides prestretches to improve the actuation range. This paper proposes a network of inflated dielectric elastomer actuators, interconnected via a chamber, with the advantages to be highly deformable and continuously controllable. Theoretical analyses show that the networked design is able to largely postpone the occurrence of material failures of the actuators, resulting in a large and continuous actuation range for their control. We further carried out experiments for validation, and the results were largely in line with the theoretical predictions. These findings essentially provide insight into developing networked soft actuators, for achieving large actuation capability.
Research Interests:
Soft robots have attracted much interest recently, due to their potential capability to work effectively in unstructured environment. Soft actuators are key components in soft robots. Dielectric elastomer actuators are one class of soft... more
Soft robots have attracted much interest recently, due to their potential capability to work effectively in unstructured environment. Soft actuators are key components in soft robots. Dielectric elastomer actuators are one class of soft actuators, which can deform in response to voltage. Dielectric elastomer actuators exhibit interesting attributes including large voltage-induced deformation and high energy density. These attributes make dielectric elastomer actuators capable of functioning as artificial muscles for soft robots. It is significant to develop untethered robots, since connecting the cables to external power sources greatly limits the robots’ functionalities, especially autonomous movements. In this paper we develop a soft untethered robot based on dielectric elastomer actuators. This robot mainly consists of a deformable robotic body and two paper-based feet. The robotic body is essentially a dielectric elastomer actuator, which can expand or shrink at voltage on or off. In addition, the two feet can achieve adhesion or detachment based on the mechanism of electroadhesion. In general, the entire robotic system can be controlled by electricity or voltage. By optimizing the mechanical design of the robot (the size and weight of electric circuits), we put all these components (such as batteries, voltage amplifiers, control circuits, etc.) onto the robotic feet, and the robot is capable of realizing autonomous movements. Experiments are conducted to study the robot’s locomotion. Finite element method is employed to interpret the deformation of dielectric elastomer actuators, and the simulations are qualitatively consistent with the experimental observations.
Research Interests:
Carbon monoxide (CO) is notorious for its strong adsorption to poison platinum group metal catalysts in the chemical industry. Here, we conceptually distinguish and quantify the effects of the occupancy and energy of d electrons, emerging... more
Carbon monoxide (CO) is notorious for its strong adsorption to poison platinum group metal catalysts in the chemical industry. Here, we conceptually distinguish and quantify the effects of the occupancy and energy of d electrons, emerging as the two vital factors in d‐band theory, for CO poisoning of Pt nanocatalysts. The stepwise defunctionalization of carbon support is adopted to fine‐tune the 5d electronic structure of supported Pt nanoparticles. Excluding other promotional mechanisms, the increase of Pt 5d band energy strengthens the competitive adsorption of hydrogen against CO for the preferential oxidation of CO, affording the scaling relationship between Pt 5d band energy and CO/H2 adsorption energy difference. The decrease of Pt 5d band occupancy lowers CO site coverage to promote its association with oxygen for the total oxidation of CO, giving the scaling relationship between Pt 5d occupancy and activation energy. The above insights outline a molecular‐level understanding of CO poisoning.
Research Interests:
Research Interests: Mathematics, Computer Science, Independent Component Analysis, Blind Source Separation, Algorithm, and 11 moreTime-Frequency Analysis, Phase Synchronization, Sparse representation, Matrix Decomposition, Sparse Matrix, K Means, Wavelet Packet Transform, Experimental Data, LINEAR PROGRAM, Basis Pursuit, and Blind Signal Separation
ABSTRACT We present several electroencephalography (EEG) signal processing and statistical analysis methods for the purpose of clinical diagnosis of brain death, in which an EEG-based preliminary examination system was developed during... more
ABSTRACT We present several electroencephalography (EEG) signal processing and statistical analysis methods for the purpose of clinical diagnosis of brain death, in which an EEG-based preliminary examination system was developed during the standard clinical procedure. Specifically, given the real-life recorded EEG signals, a robust principal factor analysis (PEA) associated with independent component analysis (ICA) approach is applied to reduce the power of additive noise and to further separate the brain waves and interference signals. We also propose a few frequency-based and complexity-based statistics for quantitative EEG analysis with an aim to evaluate the statistical significance differences between the coma patients and quasi-brain-death patients. Based on feature selection and classification, the system may yield a binary decision from the classifier with regard to the patient’s status. Our empirical data analysis has shown some promising directions for real-time EEG analysis in clinical practice.