Scholarlink

<p>This study presents an artificial neural network combined with particle swarm optimization (ANN+PSO) for the short-term forecasting of the solar radio flux index F10.7. The proposed model outperforms conventional backpropagation-trained ANN across all forecasting horizons (from F10.7(t+1) to F10.7(t+10)), achieving lower statistical error values and higher correlation coefficients. It was validated during the maximum phase of Solar Cycle 24, where it exhibited robust performance during episodes of high solar activity (F10.7 &gt; 100 sfu). When compared with state-of-the-art methods reported in the literature, the ANN+PSO achieved a lower RMSE, demonstrating its competitiveness as a simpler yet efficient alternative to more complex deep-learning approaches. Overall, the incorporation of PSO optimization enhances ANN learning and generalization, enabling richer internal representations that improve the capture of nonlinear F10.7 dynamics.</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

Hybrid ANN+PSO Technique for Accurate Short-Term F10.7 Index Forecasting

  • 期刊论文

  • 发表日期：

    2026年4月15日

查看全文
  • J. A. Lazzús,
  • I. Salfate

摘要

This study presents an artificial neural network combined with particle swarm optimization (ANN+PSO) for the short-term forecasting of the solar radio flux index F10.7. The proposed model outperforms conventional backpropagation-trained ANN across all forecasting horizons (from F10.7(t+1) to F10.7(t+10)), achieving lower statistical error values and higher correlation coefficients. It was validated during the maximum phase of Solar Cycle 24, where it exhibited robust performance during episodes of high solar activity (F10.7 > 100 sfu). When compared with state-of-the-art methods reported in the literature, the ANN+PSO achieved a lower RMSE, demonstrating its competitiveness as a simpler yet efficient alternative to more complex deep-learning approaches. Overall, the incorporation of PSO optimization enhances ANN learning and generalization, enabling richer internal representations that improve the capture of nonlinear F10.7 dynamics.