Scholarlink

<p>The development of renewable energy systems depends on accurate biomass estimation, especially when it comes to locating sustainable sources for the production of heat and electricity. Although biomass has a lot of promise as a carbon–neutral energy source, traditional measurement techniques are frequently expensive and time-consuming. The current study investigates the application of sophisticated machine learning (ML) models to forecast three important biomass types: total biomass (TB), below-ground biomass (BGB), and above-ground biomass (AGB). Extreme Learning Machine (ELM), Kernel Ridge Regression (KR), and Gaussian Process (GP) regression are the models used; they were selected based on their individual prowess in handling nonlinear patterns, regularization in small datasets, and quick training. A dataset of 175 destructively sampled trees from 27 plots in the Central Highlands of Vietnam was used to train these models, which included important input variables like soil type, forest type, tree height, diameter at breast height (DBH), and annual temperature. With an R<sup>2</sup> of 0.9999 on the hold-out test set (25 percent of the data), GP performed better than the other models, while ELM’s was 0.9741. Additionally, the GP model’s Max Error (3.2031E-09) for TB prediction was incredibly low. The suggested method supports more effective resource planning in renewable energy systems by providing a quick, affordable, and extremely accurate approach to biomass forecasting.</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

The optimization of biomass production forecasting via machine learning techniques

  • 期刊论文

  • 发表日期：

    2025年9月9日

查看全文
  • Xiaohua Li,
  • Zhenhua Yang

摘要

The development of renewable energy systems depends on accurate biomass estimation, especially when it comes to locating sustainable sources for the production of heat and electricity. Although biomass has a lot of promise as a carbon–neutral energy source, traditional measurement techniques are frequently expensive and time-consuming. The current study investigates the application of sophisticated machine learning (ML) models to forecast three important biomass types: total biomass (TB), below-ground biomass (BGB), and above-ground biomass (AGB). Extreme Learning Machine (ELM), Kernel Ridge Regression (KR), and Gaussian Process (GP) regression are the models used; they were selected based on their individual prowess in handling nonlinear patterns, regularization in small datasets, and quick training. A dataset of 175 destructively sampled trees from 27 plots in the Central Highlands of Vietnam was used to train these models, which included important input variables like soil type, forest type, tree height, diameter at breast height (DBH), and annual temperature. With an R2 of 0.9999 on the hold-out test set (25 percent of the data), GP performed better than the other models, while ELM’s was 0.9741. Additionally, the GP model’s Max Error (3.2031E-09) for TB prediction was incredibly low. The suggested method supports more effective resource planning in renewable energy systems by providing a quick, affordable, and extremely accurate approach to biomass forecasting.