Scholarlink

<p>The negative-output KY boost DC-DC converter is used in LED lightings. LED lighting major problems are color shift and depreciation in luminance due to high ripples and low efficiency of the converter. In this paper, modified negative output KY boost converter (MNO-KYB) is proposed. In proposed MNO-KYB Converter, Bayesian optimized MLR Controller is used, which improves the efficiency and reduces the ripples through appropriate PWM tuning during high dissipation of LED. Interleaved topology KY converters with Bayesian optimization machine learning has R-square of about 0.9926 during the LED light luminance. This indicates that the Bayesian Optimized-Multiple Regression Learning (BO-MRL) model is highly suitable for converter operations. Moreover, the proposed MNO-KYB converter is comprehensively benchmarked against classical PI, fuzzy logic, and sliding mode control techniques. The simulation results demonstrate a ripple reduction of up to 98.21% compared to traditional negative-output KY boost converters, with hardware implementation achieving a 95.74% decrease in output voltage ripple (from 60 mV down to 2.5 mV). Experimental validation further confirms a peak efficiency of 96.20%, which surpasses the typical 89.9%–94% range reported in state-of-the-art designs. These advancements ensure not only minimized ripple and higher efficiency but also maintain uniform luminance and color stability in demanding LED lighting applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Bayesian Optimized Multiple Regression PWM Controller for LED Lighting to Eliminate Color Shift and Maintain Luminance Using Interleaved Negative KY Converter

  • 期刊论文

  • 发表日期：

    2025年9月29日

查看全文
  • K. Rameez Raja,
  • Kuppan Ramash Kumar

摘要

The negative-output KY boost DC-DC converter is used in LED lightings. LED lighting major problems are color shift and depreciation in luminance due to high ripples and low efficiency of the converter. In this paper, modified negative output KY boost converter (MNO-KYB) is proposed. In proposed MNO-KYB Converter, Bayesian optimized MLR Controller is used, which improves the efficiency and reduces the ripples through appropriate PWM tuning during high dissipation of LED. Interleaved topology KY converters with Bayesian optimization machine learning has R-square of about 0.9926 during the LED light luminance. This indicates that the Bayesian Optimized-Multiple Regression Learning (BO-MRL) model is highly suitable for converter operations. Moreover, the proposed MNO-KYB converter is comprehensively benchmarked against classical PI, fuzzy logic, and sliding mode control techniques. The simulation results demonstrate a ripple reduction of up to 98.21% compared to traditional negative-output KY boost converters, with hardware implementation achieving a 95.74% decrease in output voltage ripple (from 60 mV down to 2.5 mV). Experimental validation further confirms a peak efficiency of 96.20%, which surpasses the typical 89.9%–94% range reported in state-of-the-art designs. These advancements ensure not only minimized ripple and higher efficiency but also maintain uniform luminance and color stability in demanding LED lighting applications.