Scholarlink

<p>Voltage regulation in self-excited induction generator (SEIG) systems equipped with electronic load controllers (ELCs) remains a persistent challenge, particularly under fluctuating load conditions and nonlinear system dynamics. Such variations adversely affect the point of common coupling (PCC) parameters, underscoring the need for a proactive prediction framework that can anticipate voltage deviations and ensure reliable system stability in standalone micro-hydro applications. Thus, this work presents a novel machine learning (ML)-based predictive framework for estimating the PCC voltage in an SEIG–ELC system. The training data were generated using the MATLAB/Simulink model of the system. Switching time, main load, and dump load were input parameters, while PCC voltage was the target. Four supervised ML algorithms were trained and evaluated using standard statistical performance metrics. These included artificial neural networks (ANN), support vector machines (SVM), k-nearest neighbors (kNN), and random forests (RF). The results indicate that the RF model achieved the highest prediction accuracy with minimal error metrics. Additionally, strategies for proactive maintenance have been proposed to enhance the model’s practical applicability.</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

Machine learning-based predictive framework for PCC voltage estimation in SEIG–ELC systems for rural micro-hydro applications

  • 期刊论文

  • 发表日期：

    2025年12月22日

查看全文
  • Shalini Sinha,
  • Rajen Pudur

摘要

Voltage regulation in self-excited induction generator (SEIG) systems equipped with electronic load controllers (ELCs) remains a persistent challenge, particularly under fluctuating load conditions and nonlinear system dynamics. Such variations adversely affect the point of common coupling (PCC) parameters, underscoring the need for a proactive prediction framework that can anticipate voltage deviations and ensure reliable system stability in standalone micro-hydro applications. Thus, this work presents a novel machine learning (ML)-based predictive framework for estimating the PCC voltage in an SEIG–ELC system. The training data were generated using the MATLAB/Simulink model of the system. Switching time, main load, and dump load were input parameters, while PCC voltage was the target. Four supervised ML algorithms were trained and evaluated using standard statistical performance metrics. These included artificial neural networks (ANN), support vector machines (SVM), k-nearest neighbors (kNN), and random forests (RF). The results indicate that the RF model achieved the highest prediction accuracy with minimal error metrics. Additionally, strategies for proactive maintenance have been proposed to enhance the model’s practical applicability.