We propose an automatic diabetic retinopathy (DR) analysis algorithm based on two-stages deep convolutional neural networks (DCNN). Compared to existing DCNN-based DR detection methods, the proposed algorithm have the following advantages: (1) Our method can point out the location and type of lesions in the fundus images, as well as giving the severity grades of DR. Moreover, since retina lesions and DR severity appear with different scales in fundus images, the integration of both local and global networks learn more complete and specific features for DR analysis.(2) By introducing imbalanced weighting map, more attentions will be given to lesion patches for DR grading, which significantly improve the performance of the proposed algorithm. In this study, we label 12, 206 lesion patches and re-annotate the DR grades of 23, 595 fundus images from Kaggle competition dataset. Under the guidance of clinical ophthalmologists, the experimental results show that our local lesion detection net achieve comparable performance with trained human observers, and the proposed imbalanced weighted scheme also be proved to significantly improve the capability of our DCNN-based DR grading algorithm.Index Terms-Diabetic retinopathy, deep convolutional neural networks, fundus images, retinopathy lesions 1 End-to-end DCNN grading means that directly feed the input images into DCNN, then output the DR grades of the images.
Switched Reluctance Motor (SRM) drives conventionally use current control techniques at low speed and voltage control techniques at high speed. However, these conventional methods usually fail to restrain the torque ripple which is normally associated with this type of machine. Compared with conventional three-phase SRMs, higher phase SRMs have the advantage of lower torque ripple: to further reduce their torque ripple, this paper presents a control method for torque ripple reduction in six-phase SRM drives. A constant instantaneous torque is obtained by regulating the rotational speed of the stator flux linkage. This torque control method is subsequently developed for a conventional converter and a proposed novel converter with fewer switching devices. Moreover, modeling and simulation of this six-phase SRM drive system has been conducted in detail and validated experimentally using a 4.0-kW six-phase SRM drive system. Test results demonstrate that the proposed torque control method has outstanding performance of restraining the torque ripple with both converters for the six-phase SRM, showing superior performance to the conventional control techniques.
Switched-mode power converters are inherently nonlinear and piecewise smooth systems which may exhibit a series of undesirable operations that can greatly reduce the converter's efficiency and lifetime. This paper presents a nonlinear analysis technique to investigate the influence of system parameters on the stability of interleaved boost converters. In this approach, Monodromy matrix which contains all the comprehensive information of converter parameters and control loop can be employed to fully reveal and understand the inherent nonlinear dynamics of interleaved boost converters, including the interaction effect of switching operation. Thereby not only the boundary conditions but also the relationship between stability margin and the parameters given can be intuitively studied by the eigenvalues of this matrix. Furthermore, employing the knowledge gained from this analysis a real time cycle to cycle variable slope compensation method is proposed to guarantee a satisfactory performance of the converter with an extended range of stable operation. Outcomes show that systems can regain stability by applying the proposed method within a few time periods of switching cycles. The numerical and analytical results validate the theoretical analysis, and experimental results verify the effectiveness of the proposed approach.
Linear Ramp Slope Compensation (LRC) and Quadratic Slope Compensation (QSC) are commonly implemented in peak current mode controlled DC-DC converters in order to minimize subharmonic and chaotic oscillations. Both compensating schemes rely on the linearized state-space averaged model (LSSA) of the converter. LSSA ignores the impact that switching actions have on the stability of converters. In order to include switching events, the nonlinear analysis method based on Monodromy matrix was introduced to describe a complete-cycle stability. Analyses on analogue controlled DC-DC converters applying this method show that system stability is strongly dependent on the change of the derivative of the slope at the time of switching instant. However, in a mixed-signal controlled system, the digitalization effect contributes differently to system stability. This paper shows a full complete-cycle stability analysis using this nonlinear analysis method, which is applied to a mixed-signal controlled converter. Through this analysis, a generalized equation is derived that reveals for the first time the real boundary stability limits, for LRC and QSC. Furthermore, this generalized equation allows the design of a new compensating scheme which is able to increases system stability. The proposed scheme is called Polynomial Curve Slope Compensation (PCSC) and it is demonstrated that PCSC increases the stable margin by 30% compared to LRC and 20% to QSC. This outcome is proved experimentally by using an interleaved DC-DC converter that is built for this work.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.