Accurate wind‐power forecasting is essential for the successful integration of renewable energy into smart city grids. Traditional long short‐term memory (LSTM) networks face significant challenges in parameter optimization, leading to suboptimal forecasting performance. This paper introduces a novel Bernstein dynamic horned lizard optimization algorithm (BDHLOA) to optimize LSTM parameters for wind power prediction in smart cities. BDHLOA incorporates three key enhancements: a Bernstein‐assisted oppositional‐multiple learning (BOML) strategy that improves exploration–exploitation balance, Bernstein‐based adaptive differential (BAD) strategy for better solution refinement, and a dynamic drift search (DDS) mechanism that prevents premature convergence. The proposed BDHLOA‐LSTM framework is evaluated for one‐step‐ahead (10‐min horizon) wind‐power forecasting using four real‐world datasets from La Haute Borne wind turbines in France, under a rolling‐origin expanding‐window cross‐validation protocol that strictly prevents data leakage. Results demonstrate exceptional performance across all four stations, with mean R 2 = 0.9695, RMSE = 0.0011, and MAE = 0.0007. BDHLOA‐LSTM reduces MAE by 94% compared to persistence forecasting and 91% compared to autoregressive AR(24) models. Against the best competing optimizer (PSO‐LSTM), BDHLOA‐LSTM achieves 42% lower MAE on average across sites. Performance is assessed using R 2 , RMSE, MAE, symmetric mean absolute percentage error (sMAPE), mean absolute scaled error (MASE), and prediction interval coverage probability (PICP) to capture both accuracy and calibrated uncertainty. Furthermore, the proposed model is interpreted using SHapley Additive exPlanations (SHAP) technique. SHAP analysis confirms that BDHLOA‐LSTM learns physically meaningful relationships dominated by wind speed statistics and direction, rather than exploiting spurious correlations. The superior accuracy and stability of BDHLOA‐LSTM make it highly suitable for real‐time grid management and sustainable energy planning in smart cities.

Abstract Accurately predicting crop yield is essential for optimizing agricultural practices and ensuring food security. However, existing approaches often struggle to capture the complex interactions between various environmental factors and crop growth, leading to suboptimal predictions. Consequently, identifying the most important feature is vital when leveraging Support Vector Regressor (SVR) for crop yield prediction. In addition, the manual tuning of SVR hyperparameters may not always offer high accuracy. In this paper, we introduce a novel framework for predicting crop yields that address these challenges. Our framework integrates a new hybrid feature selection approach with an optimized SVR model to enhance prediction accuracy efficiently. The proposed framework comprises three phases: preprocessing, hybrid feature selection, and prediction phases. In preprocessing phase, data normalization is conducted, followed by an application of K-means clustering in conjunction with the correlation-based filter (CFS) to generate a reduced dataset. Subsequently, in the hybrid feature selection phase, a novel hybrid FMIG-RFE feature selection approach is proposed. Finally, the prediction phase introduces an improved variant of Crayfish Optimization Algorithm (COA), named ICOA, which is utilized to optimize the hyperparameters of SVR model thereby achieving superior prediction accuracy along with the novel hybrid feature selection approach. Several experiments are conducted to assess and evaluate the performance of the proposed framework. The results demonstrated the superior performance of the proposed framework over state-of-art approaches. Furthermore, experimental findings regarding the ICOA optimization algorithm affirm its efficacy in optimizing the hyperparameters of SVR model, thereby enhancing both prediction accuracy and computational efficiency, surpassing existing algorithms.