Search

A new hybrid technique for grid integration of solar photovoltaic system using modified incremental conductance maximum power point tracking algorithm with multi-output converter and multilevel inverter is proposed in the paper. The proposed hybrid technique is the joint execution of improved dolphin echolocation algorithm and gradient boosting decision trees and this way the proposed technique is named as IDE-GBDT technique. The novelty of the proposed work is the dolphin echolocation algorithm, which is integrated by the crossover and the mutation function so it is named as IDE. In the proposed technique, the multi-output converter is the combination of boost converter and switched capacitor function to generate different self-balanced output voltages. The utilization of multilevel inverter in the proposed system provides better quality of output voltage and current waveform thereby reducing the size of passive filters. Also, eliminates the requirement of bulky transformers for grid integration. Multicarrier unipolar phase disposition pulse width modulation technique is employed for triggering the switches of the multilevel inverter. The maximum power point tracking algorithm uses the estimated active power output of the generator as its input and generates command speed so that maximum power is transferred to the dc link. This control system also incorporates a loss minimization approach to minimize the losses in the generator and hence to improve the efficiency of the photovoltaic system. Finally, the performance of the proposed maximum power point tracking control of wind and photovoltaic power generation schemes is executed in Matrix Laboratory/Simulink working platform and the execution is assessed with the existing techniques. The proposed technique is compared with the existing techniques and the observed total harmonic distortion of the proposed technique in all the cases is 0.67%, 0.51%, 0.58%, 0.63%, 0.92%, and 1.03% and the total harmonic distortion is found to be very less compared with existing techniques. [ABSTRACT FROM AUTHOR]

This article presents a analysis, design and implementation of non-linear controllers for fundamental negative super lift Luo-Converter (FNSLLC) for purposes needing the stable power source in battery operated portable devices, floppy/hard disk drives, LED TV, physiotherapy medical instrument, lap-top computers, mother board and fan in central processing unit (CPU) applications etc.,. The FNSLLC is a advanced DC-DC converter topology. The FNSLLC is erratic structure system (ESS) and its dynamic analysis is poor. The linear regulators for FNSLLC are poor operating analysis particularly during large source voltage and load modifications. With the aim of improve the dynamic analysis, load voltage and coil current controls of FNSLLC, a linear quadratic regulator (LQR) plus fuzzy logic controller (FLC) is designed. The LQR is designed for FNSLLC with their state space dynamic equations. The controller formation of this converter consists of two loops like, current loop (CL) and voltage loop (VL). In this study, LQR is act as a inner CL for manipulating the coil current of FNSLLC, but the FLC is act as a VL for controlling the load voltage of FNSLLC. The FLC is developed depending on the same system activities and qualitative linguistic control rules. The performance of FNSLLC using LQR plus FLC is verified at various operating states by building both in MATLAB/Simulink and prototype field programmable array (FPGA) models in comparisons with LQR plus proportional double integral controller (PDIC). The results and time domain specifications analyze are presented to prove the adroit of designed controller in different provinces. [ABSTRACT FROM AUTHOR]

Glaucoma is a condition that damages optic nerve head which is irreversible if it is not treated at the earliest. In this paper, Optical Coherence Tomography (OCT) image classification for glaucoma diagnosis is presented which uses texture statistical approach for classification. A set of common statistical features from the OCT image and another set of statistical features from the Gray Level Difference Matrix (GLDM) of OCT image are used. They are analyzed independently by a supervised classifier named Support Vector Machine (SVM). To obtain higher accuracy, two sets of features are combined linearly and tested. Experimental results show that the OCT image classification system provides an accuracy of 95.5% with sensitivity of 94% and specificity of 97% by the combination of features. [ABSTRACT FROM AUTHOR]