Scholarlink

<p>Designing effective solar energy policies requires adaptive decision-making that accounts for both socioeconomic conditions and public perception. This paper proposes a sentiment-informed causal reinforcement learning framework for adaptive solar energy policy design. Public sentiment toward solar energy is first extracted from large-scale social media and news data using a domain-adapted RoBERTa transformer model. Structural causal models are then employed to estimate the causal impact of sentiment on solar adoption while adjusting for key economic and policy confounders, including GDP per capita, electricity prices, and existing subsidy levels. These causal estimates inform a Proximal Policy Optimization (PPO) agent that learns adaptive and cost-aware policy interventions, such as subsidy adjustments and targeted awareness campaigns, to maximize long-term adoption under fiscal constraints. Explainability and robustness are ensured through SHAP-based feature attribution, policy entropy analysis, counterfactual simulations, and sensitivity analysis. Experiments on multi-regional data demonstrate consistent improvements over static and heuristic baselines in both adoption gains and cost efficiency, highlighting the value of sentiment-aware and causally grounded reinforcement learning for energy policy design.</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

Sentiment-informed causal reinforcement learning for solar energy policy optimization

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月7日

查看全文
  • Asif Jabbar,
  • Jingbo Yuan,
  • Salma Idris,
  • Muhammad I. Khan,
  • Tariq Mahmood

摘要

Designing effective solar energy policies requires adaptive decision-making that accounts for both socioeconomic conditions and public perception. This paper proposes a sentiment-informed causal reinforcement learning framework for adaptive solar energy policy design. Public sentiment toward solar energy is first extracted from large-scale social media and news data using a domain-adapted RoBERTa transformer model. Structural causal models are then employed to estimate the causal impact of sentiment on solar adoption while adjusting for key economic and policy confounders, including GDP per capita, electricity prices, and existing subsidy levels. These causal estimates inform a Proximal Policy Optimization (PPO) agent that learns adaptive and cost-aware policy interventions, such as subsidy adjustments and targeted awareness campaigns, to maximize long-term adoption under fiscal constraints. Explainability and robustness are ensured through SHAP-based feature attribution, policy entropy analysis, counterfactual simulations, and sensitivity analysis. Experiments on multi-regional data demonstrate consistent improvements over static and heuristic baselines in both adoption gains and cost efficiency, highlighting the value of sentiment-aware and causally grounded reinforcement learning for energy policy design.