Scholarlink

<p>Wind speed forecasting becomes a tedious task due to its stochastic and fluctuating behavior. To handle these challenges and ensure a stable energy supply, reliable wind speed forecasting is required. This paper presents a novel hybrid classical-quantum model that combines the classical feedforward neural networks (FNN) with the quantum neural networks (QNN), implemented via parameterized quantum circuits. Notably, the FNN-Q1 model exhibited the best performance, with a mean absolute error (MAE) of 0.1285, underscoring the efficacy of integrating classical and quantum neural network methodologies for enhanced wind speed prediction. Additionally, the FNN-Q1 model achieved a root mean square error (RMSE) of 0.2550 and a mean absolute percentage error (MAPE) of 8.20%. Further proposed hybrid model is compared against benchmark models for multistep ahead forecasting. The performance and accuracy of proposed model is found better compared to benchmark model, proves superior computational capabilities over classical machine learning. In addition to this, uncertainty quantification is included in&#xa0;this study&#xa0;and performance of model is evaluated through metrics such as Brier Score (1.8123 for FNN-Q1, 2.3114 for FNN-Q2), Log Likelihood, Prediction Interval Coverage Probability (PICP: 0.8521 for FNN-Q1, 0.7028 for FNN-Q2), and mean squared log error (MSLE), highlighting the robustness of the proposed hybrid models in predictive reliability. The results show that blending classical and quantum methods can make wind forecasts more reliable and could help manage renewable energy more efficiently.</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

Wind speed forecasting using hybrid classical-quantum-infused model

  • 期刊论文

  • 发表日期：

    2026年1月9日

查看全文
  • Ajay Kumar,
  • A. J. Singh,
  • Sanjay Kumar

摘要

Wind speed forecasting becomes a tedious task due to its stochastic and fluctuating behavior. To handle these challenges and ensure a stable energy supply, reliable wind speed forecasting is required. This paper presents a novel hybrid classical-quantum model that combines the classical feedforward neural networks (FNN) with the quantum neural networks (QNN), implemented via parameterized quantum circuits. Notably, the FNN-Q1 model exhibited the best performance, with a mean absolute error (MAE) of 0.1285, underscoring the efficacy of integrating classical and quantum neural network methodologies for enhanced wind speed prediction. Additionally, the FNN-Q1 model achieved a root mean square error (RMSE) of 0.2550 and a mean absolute percentage error (MAPE) of 8.20%. Further proposed hybrid model is compared against benchmark models for multistep ahead forecasting. The performance and accuracy of proposed model is found better compared to benchmark model, proves superior computational capabilities over classical machine learning. In addition to this, uncertainty quantification is included in this study and performance of model is evaluated through metrics such as Brier Score (1.8123 for FNN-Q1, 2.3114 for FNN-Q2), Log Likelihood, Prediction Interval Coverage Probability (PICP: 0.8521 for FNN-Q1, 0.7028 for FNN-Q2), and mean squared log error (MSLE), highlighting the robustness of the proposed hybrid models in predictive reliability. The results show that blending classical and quantum methods can make wind forecasts more reliable and could help manage renewable energy more efficiently.