Search

Mild cognitive impairment (MCI) is a transitional stage between normal aging and Alzheimer’s disease (AD), with a high risk of converting to AD. We propose a classification framework with a data augment method to identify MCI converter (MCI-C) and MCI non-converter (MCI-NC). Resting-state functional magnetic resonance images (rs-fMRI) from Alzheimer’s Disease Neuroimaging Initiative (ADNI) are processed as augmented resting-state functional connectivity by staggered sliding window (SSW) method proposed by us under Human Connectome Project (HCP) multi-modal parcellation. The HCP brain atlas provides a more detailed cortical parcellation of the brain, allowing for more precise localization of brain regions related to MCI and AD. Finally, the framework archive 88% accuracy in the task of identifying MCI-C. 46 brain regions are suggested as potential MCI-to-AD biomarkers.

Background: Alzheimer’s disease (AD) is an incurable neurodegenerative disease primarily affecting the elderly population. The therapy of AD depends heavily on an early diagnosis. In this study, our primary objective is to evaluate the classification framework, which combines graph theory and machine learning techniques for functional magnetic resonance imaging (fMRI), to distinguish AD, early mild cognitive impairment (EMCI), late mild cognitive impairment (LMCI), and healthy control (HC). Methods: A novel multi-feature selection method, incorporating the dual graph theoretical approach, is proposed for classification. This method utilizes three different feature selection methods after brain areas selection through graph-theory analyses in 96 subjects with brain parcellation by using the joint human connectome project multimodal parcellation (J-HCPMMP) of 180 areas per hemisphere. Results: The classification results show that the optimal features selected by the minimal redundancy maximal relevance (MRMR) based on support vector machine linear (SVM-linear) from graph measures for 36 areas of 360 areas. The classification accuracies for identifying HC vs. EMCI, HC vs. LMCI, HC vs. AD, EMCI vs. LMCI, LMCI vs. AD, and EMCI vs. AD, are 85.60%, 92.90%, 96.80%, 83.30%, 84.90% and 89.50%, respectively. Conclusion: The results indicate that the combination of graph measures and machine learning in fMRI connectivity analysis might be helpful for the diagnosis of AD, especially the use of local measures, which may better reflect functional changes in local brain regions because of cognitive impairment.

Intelligence and cloudification are widely recognized as key driving forces in the evolution of 5G radio access network (RAN). This paper presents a promising architecture framework for the evolution of 5G radio access network, enabled by a deep integration with cloudification and artificial intelligence/machine learning (AI/ML) technologies. To accommodate the diversified scenarios and services and handle the complexity of the 5G network in a flexible and efficient manner, the architecture framework highlights three concepts: convergence of RAN and cloud, RAN empowered by hierarchical AI capabilities, and mutual awareness between RAN and services. The key design aspects and technologies that realize those concepts are discussed systematically. Two typical use cases including the RAN slice resource allocation optimization and RAN-aware video service assurance, are demonstrated along with the simulation or lab test results to validate the potential of the architecture framework.

This paper proposes a modified model predictive control method with the virtual model (VM-MPC) to generate a virtual reference of the position loop. The dynamic deviation is introduced into the virtual reference as a position advance. Applying this virtual reference as the input of position controllers, the dynamic response is greatly improved without changing the gain of position controllers. Therefore, both the dynamic response and robustness of the position loop are improved when the response delay of the speed loop is uncertain. In order to reduce the influence of model mismatch on model predictive control, instead of the actual feedback, a virtual model feedback of the expected position response is used. With this approach, control parameters of VM-MPC are optimized offline to obtain the fast position response without overshoot. A model compensator is proposed to reduce the mismatch between the virtual model and the real system. The effectiveness and feasibility of the proposed method are verified by simulation and experiment results.

Based on proposed joint human connectome project multi-modal parcellation (JHCPMMP), the study on the binary classification of Alzheimer’s disease was conducted. We tried to build a novel classification model, which can be interpretative and have the ability to deal with the complexity and individual differences of brain networks. The subclass weighted logistic regression (SWLR) based on logistic regression was proposed in this paper. We conducted five groups of experiments, in which the accuracy of HC vs. AD was 95.8%, HC vs. EMCI was 91.6%, HC vs. LMCI was 93.7%, EMCI vs. LMCI was 89.5%, and LMCI vs. AD was 91.6%. In addition, we conducted a follow-up analysis of the coefficient matrix and found that the distribution of core deterioration brain regions in different stages is different in the development of Alzheimer’s disease. We located these brain regions in two-dimensional images and found that they generally show a trend of continuous counterclockwise migration.

Reverse conduction loss of GaN high electron mobility transistors (HEMTs) in very high frequency (VHF) converters is non-neglectable due to the absence of body diode. To reduce the reverse conduction time, this paper proposes a linear equivalent model for class $\Phi _{2}$ inverter to derive optimal duty cycles for different working conditions. The proposed model is derived from frequency-domain perspective, which simplifies the derivation process and provides an intuitive physical insight for class $\Phi _{2}$ inverter. Firstly, the power switch is modeled as a current source according to time-domain expressions of inductor current, which simplifies the circuit to a linear network. Based on quantitative analysis of the current source spectrum characteristics, the linear network response to the current excitation is derived, which forms the linear equivalent model. Then, drain voltage of the main switch under different working conditions is obtained. With the zero-crossing point of the drain voltage, numerical solutions of the optimal duty cycles are calculated. Finally, the optimal duty cycles are implemented with a high-resolution duty cycle generation circuit in a 27.12MHz prototype, which achieves not only a peak efficiency of 93.6% at full load, but also higher efficiency over the whole load and input voltage range compared to conventional fixed duty cycle.

Nonideal switching characteristics of voltage source inverter (VSI) will result in distortions in both phase currents and reference voltages of a permanent magnet synchronous machine (PMSM) drive system, which will also consequently introduce ripples into the output speed and shaft torque, and result in a performance deterioration in the state observation control. To solve this issue, a method of directly extracting and compensating the nonlinear voltage disturbance due to VSI nonlinearity employed in PMSM drives is proposed in this paper. Firstly, the distorted voltage due to VSI nonlinearity is directly extracted via a harmonic separation scheme, which can be easily compensated by a proportional integral controller afterwards. Since both 6th harmonic and DC terms of dq-axis distorted voltages due to VSI nonlinearity are well compensated, the proposed method shows sufficiently good performance under both $i_{d}=0$ and maximum torque per ampere controls ( $i_{d} < 0$ ), as verified on two prototype PMSMs being salient pole and non-salient pole, respectively, both developed for a low-speed rotary actuator. In addition, the application of proposed method to state observation controls such as the model-based sensorless control and the deadbeat predictive current control, is finally investigated. Test results manifest that the proposed method can effectively improve the accuracy of observed rotor position/speed, and the ability of current tracking.

The Cucker-Smale (C-S) model describes an interacting particle system in which the connection weights decrease with increasing distance. This model features emergent behaviors by which the velocities of the particles converge to a common value without a central command. However, the consensus value of the original C-S flocking model is restricted to the leader-following consensus or average consensus. Moreover, for the short-range communication-based C-S model, consensus can only be obtained for specific initial configurations. In this paper, the short-range communication-based C-S model is extended to achieve distributed optimization, where the consensus value optimizes the objective function of the group for any bounded initial configuration. Simulation examples are provided to demonstrate the effectiveness of our approach.

Valley-switching (VS) and zero-voltage switching (ZVS) improves overall efficiency in critical conduction mode (CRM) boost converters. To achieve VS/ZVS, off-time of the main switch is often extended to match the resonance period by circuit inductor and parasitic capacitors of switching components. In this article, a piecewise equivalent model for parasitic capacitors is proposed to derive analytical solutions of the resonant process, by which the numerical solutions of VS/ZVS time is calculated. To precisely achieve VS/ZVS in a boost converter, corresponding extended off-time based on the numerical solutions is implemented with an extended off-time (EOT) controller. The EOT controller makes the inductor volt-second unbalance in continuous conduction mode (CCM) operation during one switching cycle, leading to the convergence to CRM operation. The analytical and experimental results correct the results derived by conventional equivalent model for parasitic capacitors, which leads to deviated VS/ZVS boundary and wrongly calculated off-time. The proposed model and controller are verified in a non-synchronous CRM boost converter based on GaN high electron mobility transistor (HEMT) and SiC diode. With the derived extended off-time, a peak efficiency of 99.15% is achieved at output power level of 200W.

In motor drives, distortions of phase voltages and currents are often caused by nonlinear effects of inverters such as dead time, turn-on delay, turn-off delay and voltage drop of power devices. To eliminate these distortions, the dead-time compensation voltage is usually investigated. Furthermore, the relationship between the dead-time compensation voltage and phase currents is nonlinear, which is related to not only the parameters mentioned above, but also the snubber and parasitic capacitance of inverters. A nonlinear function is constructed to model the nonlinear relationship in this paper. To identify the nonlinear function, a method based on artificial neural networks is proposed without inverter parameters. According to the criterion that the trajectory of voltage vector in α-β coordinate system is a circle, an adaptive law is constructed to modify the parameters of the nonlinear function. Therefore, the nonlinear dead-time compensation voltage model is obtained accurately, where the distortions of voltages and currents are reduced without any additional hardware. Applying this method to the current predictive control, the bandwidth of a current loop is increased by 500Hz. Effectiveness of the method is verified by experiments.

Identification of vulnerable lines in power systems has great meaning to predict cascading failures and prevent the blackouts. A novel approach to identify the vulnerable lines in power systems based on the weighted H-index is proposed. Firstly, to remove the Kirchhoff's law constraint in power system, a second-level correlation network is constructed considering the secondary cascading failure. Then, the relative node strength is defined in the second-level correlation network to improve the identification effectiveness. Secondly, a weighted H-index is proposed, aiming at the disadvantages of the classic H-index only suited for unweighted networks. The improved indicator is applied in the second-level correlation network to identify the vulnerable lines of the power grid. The proposed method is verified in the IEEE-39 bus system and a real-life regional power grid in China. The experimental results show that the proposed method has higher precision compared with other methods. Specifically, the identification accuracy is improved by at least 8.33% than existing methods in the IEEE-39 bus system. In addition, the weighted H-index has higher discrimination and robustness than the classic H-index.

In this paper, we propose a new hotspot ranking-based indoor mapping and mobility analysis approach based on the sporadically collected crowdsourced Wi-Fi received signal strength (RSS) data. This approach aims to construct the indoor mapping, as well as achieve the mobility analysis of the users following their daily motion patterns in target environment. First, we perform the wavelet analysis with respect to each RSS sequence to mitigate the noise interference to some extent. Second, we develop a new multidimensional scaling approach to map each RSS data into a linear one in the 2-D signal space, which is followed by the density clustering approach to merge the linear ones into different clusters based on the spatial correlation property. Finally, we construct the indoor mapping from the signal into physical spaces by the concept of hotspot ranking order, as well as the transfer relations among different RSS clusters and different physical sub-areas. The experimental results demonstrate that the proposed approach can achieve the superior performance in terms of indoor mapping and mobility analysis in an unknown indoor environment.