Scholarlink

<p>The global transition toward sustainable energy requires rapid deployment of renewable infrastructure supported by reliable tools for enterprise evaluation. Conventional assessment approaches fall short because they do not capture the environmental, social, technological, and economic dimensions of sustainability. This study proposes a machine learning (ML)–based framework to evaluate the sustainability performance of new power system companies. The objective is to design a data-driven ranking model that integrates environmental, financial, technological, and social indicators, enabling investors, regulators, and policymakers to make informed decisions. The Sustainable Power Enterprise Dataset was utilized, incorporating key performance indicators (KPIs) such as carbon offset, energy efficiency, renewable energy share, financial performance, innovation, and workforce diversity. Preprocessing was performed using Min–Max normalization to standardize feature scales and Z-score normalization for outlier removal. Linear Discriminant Analysis (LDA) was applied for feature extraction, highlighting critical sustainability dimensions including energy efficiency, carbon offsetting, and scalability. The proposed method employs a hybrid Scalable Shuffled Frog Leaping Algorithm–adapted Logistic Regression (SSFLA-LR). Here, SSFLA provides global search and optimization for handling high-dimensional feature spaces, while Logistic Regression (LR) ensures interpretability and probabilistic classification. To demonstrate robustness, comparative evaluations were also conducted with standalone SSFLA-LR models. Experimental results demonstrated that the proposed method significantly outperformed conventional models, achieving the lowest MSE (0.005) and MAE (0.045), alongside the highest E<sub>ns</sub> (0.768) and E<sub>lm</sub> (0.968). These findings demonstrate the effectiveness of SSFLA-LR in identifying high-potential sustainable enterprises, thereby supporting data-driven decision-making for advancing global energy sustainability.</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

AI driven sustainable development business evaluation system using machine learning model for new type power systems

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月3日

查看全文
  • Xiaona Gao,
  • Tinghao Lei,
  • Xin Zhou,
  • Xiaozhu Lin,
  • Xiting Chen

摘要

The global transition toward sustainable energy requires rapid deployment of renewable infrastructure supported by reliable tools for enterprise evaluation. Conventional assessment approaches fall short because they do not capture the environmental, social, technological, and economic dimensions of sustainability. This study proposes a machine learning (ML)–based framework to evaluate the sustainability performance of new power system companies. The objective is to design a data-driven ranking model that integrates environmental, financial, technological, and social indicators, enabling investors, regulators, and policymakers to make informed decisions. The Sustainable Power Enterprise Dataset was utilized, incorporating key performance indicators (KPIs) such as carbon offset, energy efficiency, renewable energy share, financial performance, innovation, and workforce diversity. Preprocessing was performed using Min–Max normalization to standardize feature scales and Z-score normalization for outlier removal. Linear Discriminant Analysis (LDA) was applied for feature extraction, highlighting critical sustainability dimensions including energy efficiency, carbon offsetting, and scalability. The proposed method employs a hybrid Scalable Shuffled Frog Leaping Algorithm–adapted Logistic Regression (SSFLA-LR). Here, SSFLA provides global search and optimization for handling high-dimensional feature spaces, while Logistic Regression (LR) ensures interpretability and probabilistic classification. To demonstrate robustness, comparative evaluations were also conducted with standalone SSFLA-LR models. Experimental results demonstrated that the proposed method significantly outperformed conventional models, achieving the lowest MSE (0.005) and MAE (0.045), alongside the highest Ens (0.768) and Elm (0.968). These findings demonstrate the effectiveness of SSFLA-LR in identifying high-potential sustainable enterprises, thereby supporting data-driven decision-making for advancing global energy sustainability.