Scholarlink

Purpose <p>Life Cycle Assessment (LCA) is an essential tool for evaluating the environmental impacts of products and processes, yet its integration with Machine Learning (ML) remains underexplored. This paper addresses the critical gap in the literature by analyzing how AI-enabled tools are incorporated into LCA methodology to predict potential environmental impacts. Our research aims to provide clear guidance on selecting the most appropriate ML methods tailored to the unique goals and specificities of LCA studies.</p> Methods <p>The research methodology involves a comprehensive literature review using targeted search strings to identify relevant studies. A bibliometric analysis was conducted to understand the publication trends. The identified studies were screened and categorized based on ML methods and their applications in LCA. A detailed content analysis was performed to evaluate the integration of ML in different phases of LCA, focusing on regression, classification, clustering, and other tasks. The findings were compiled into a synthesis table to highlight the effectiveness and computational efficiency of various ML models.</p> Results and discussion <p>The analysis revealed that Artificial Neural Networks (ANN) and Random Forest (RF) are the most frequently used ML models in LCA integration, primarily for regression and classification tasks. The study also found a significant gap in real-time dynamic LCA models, with existing models largely static. The synthesis table provides a comparative overview of ML models, indicating their effectiveness, sensitivity, and computational efficiency. ANN emerged as the most effective model, while Bayesian networks showed potential for further exploration due to their specific characteristics.</p> Conclusions <p>Integrating ML with LCA enhances the accuracy and efficiency of environmental impact predictions. However, the current use of ML in LCA is limited by data quality and the static nature of traditional LCA models. Future research should focus on developing dynamic LCA models and improving data quality to leverage the full potential of ML. Recommendations include exploring underutilized ML models and extending AI-enabled LCA applications to various sectors, including social contexts and the tertiary sector.</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

Integrating machine learning with life cycle assessment: a comprehensive review and guide for predicting environmental impacts

  • 期刊论文

  • 发表日期：

    2025年2月21日

查看全文
  • João Victor Encide Salla,
  • Tiago Agostinho de Almeida,
  • Diogo Aparecido Lopes Silva

摘要

Purpose

Life Cycle Assessment (LCA) is an essential tool for evaluating the environmental impacts of products and processes, yet its integration with Machine Learning (ML) remains underexplored. This paper addresses the critical gap in the literature by analyzing how AI-enabled tools are incorporated into LCA methodology to predict potential environmental impacts. Our research aims to provide clear guidance on selecting the most appropriate ML methods tailored to the unique goals and specificities of LCA studies.

Methods

The research methodology involves a comprehensive literature review using targeted search strings to identify relevant studies. A bibliometric analysis was conducted to understand the publication trends. The identified studies were screened and categorized based on ML methods and their applications in LCA. A detailed content analysis was performed to evaluate the integration of ML in different phases of LCA, focusing on regression, classification, clustering, and other tasks. The findings were compiled into a synthesis table to highlight the effectiveness and computational efficiency of various ML models.

Results and discussion

The analysis revealed that Artificial Neural Networks (ANN) and Random Forest (RF) are the most frequently used ML models in LCA integration, primarily for regression and classification tasks. The study also found a significant gap in real-time dynamic LCA models, with existing models largely static. The synthesis table provides a comparative overview of ML models, indicating their effectiveness, sensitivity, and computational efficiency. ANN emerged as the most effective model, while Bayesian networks showed potential for further exploration due to their specific characteristics.

Conclusions

Integrating ML with LCA enhances the accuracy and efficiency of environmental impact predictions. However, the current use of ML in LCA is limited by data quality and the static nature of traditional LCA models. Future research should focus on developing dynamic LCA models and improving data quality to leverage the full potential of ML. Recommendations include exploring underutilized ML models and extending AI-enabled LCA applications to various sectors, including social contexts and the tertiary sector.