Path selection for green and low-carbon economic industrial clustering based on machine learning algorithms

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:Demonstrates how ML (GNN, RL) can optimize industrial cluster path selection for decarbonization, offering a rigorous evaluation framework.

🏢実務担当者:Utilities and industrial developers can use the framework to plan renewable-integrated clusters and reduce carbon footprint.

🏛政策担当者:Supports evidence-based planning of low-carbon industrial zones and renewable energy integration.

In response to the shortcomings of traditional green and low-carbon economic industrial clustering path selection in data processing efficiency, path optimization, industrial collaboration, and policy-making, this paper innovatively applies automated data preprocessing, graph neural network (GNN) modeling, reinforcement learning (RL) dynamic optimization, and multi-objective optimization methods to solve problems, providing scientific and efficient decision-making support for achieving the “carbon peaking and carbon neutrality” goals. Renewable energy integration and sustainable energy transition form the core context of this investigation. First, multi-source data are collected and preprocessed. Then, K-means clustering and principal component analysis are used to optimize industrial cluster site selection. Model prediction capabilities are improved through recursive feature elimination and expert knowledge. Next, an enterprise relationship graph is constructed, and the GNN is applied to identify key nodes and collaboration patterns in the industrial chain. The RL is simultaneously adopted to simulate the dynamic development of the industrial chain and learn the optimal strategy. In addition, regression models and time-series analysis (long short-term memory) are used to predict the impact of policies on carbon emissions and future trends, and to establish a risk warning mechanism. Finally, a multi-objective optimization algorithm is used to balance environmental and economic benefits, and the enterprise layout is optimized to reduce transportation costs and carbon emissions. Experimental results show that the proposed machine learning framework improves data preprocessing efficiency by 85%, reduces cluster error by 60%, shortens path optimization time by 84%, and achieves a 66.7% reduction in carbon emissions within the simulation environment, demonstrating a measurable advance in the rigor and computational tractability of path selection for green, low-carbon economic industrial clustering.

• semanticscholar https://doi.org/10.1063/5.0323311first seen 2026-06-26 05:43:17