Scholarlink

<p>Hybrid energy systems are increasingly critical in addressing the growing demand for sustainable and efficient power solutions. In this paper, a novel converter for a hybrid energy system with the capability to integrate two power sources of different characteristics, namely AC and DC is proposed. This paper aims to enhance the efficiency of hybrid energy systems that involve multiple power conversions and necessitate multiple power converters. The pivotal aspect of the proposed converter lies in its ability to connect photovoltaic (PV) and grid power sources. Diverging from conventional setups, this converter eliminates the need for a diode rectifier, streamlining the power conversion process and mitigating complexities associated with multiple stage conversions between DC and AC power stages. The proposed converter shows versatility by operating solely on grid power, solar power, or a combination of grid and solar power, and it is able to change its operating mode by adapting dynamically to varying power availability. The proposed converter with proposed flat-topped waveform has 5.69% voltage THD, which is 87.50% less than conventional system voltage waveform. Various regression models, such as trees, Gaussian processes regression (GPR), ensembles of trees, support vector machine, and neural network were trained and tested to forecast the PV power. Among these, the squared exponential GPR model outperforms other regression models, exhibiting the least root mean square error of 0.16745 and mean square error of 0.02841. The paper further analyses the behavior of the proposed converter in water pumping systems used for residential, commercial, and irrigation applications. The operating modes of the converter are determined by machine learning-based power predictions, influencing transitions between grid and solar power as well as the concurrent utilization of both sources. This research provides insights into the transient behaviors during these operational mode changes, contributing to a comprehensive understanding of the converter's performance.</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

A next-generation hybrid energy converter empowered by machine learning: pioneering sustainable integration of photovoltaic and grid power

  • 期刊论文

  • 发表日期：

    2025年2月18日

查看全文
  • Nikhil Agrawal,
  • Anshul Agarwal,
  • Tirupathiraju Kanumuri

摘要

Hybrid energy systems are increasingly critical in addressing the growing demand for sustainable and efficient power solutions. In this paper, a novel converter for a hybrid energy system with the capability to integrate two power sources of different characteristics, namely AC and DC is proposed. This paper aims to enhance the efficiency of hybrid energy systems that involve multiple power conversions and necessitate multiple power converters. The pivotal aspect of the proposed converter lies in its ability to connect photovoltaic (PV) and grid power sources. Diverging from conventional setups, this converter eliminates the need for a diode rectifier, streamlining the power conversion process and mitigating complexities associated with multiple stage conversions between DC and AC power stages. The proposed converter shows versatility by operating solely on grid power, solar power, or a combination of grid and solar power, and it is able to change its operating mode by adapting dynamically to varying power availability. The proposed converter with proposed flat-topped waveform has 5.69% voltage THD, which is 87.50% less than conventional system voltage waveform. Various regression models, such as trees, Gaussian processes regression (GPR), ensembles of trees, support vector machine, and neural network were trained and tested to forecast the PV power. Among these, the squared exponential GPR model outperforms other regression models, exhibiting the least root mean square error of 0.16745 and mean square error of 0.02841. The paper further analyses the behavior of the proposed converter in water pumping systems used for residential, commercial, and irrigation applications. The operating modes of the converter are determined by machine learning-based power predictions, influencing transitions between grid and solar power as well as the concurrent utilization of both sources. This research provides insights into the transient behaviors during these operational mode changes, contributing to a comprehensive understanding of the converter's performance.