• • The SCO-BiLSTM model achieves a 10.313% reduction in eMAPE compared with PSO-BiLSTM, directly addressing the local optima stagnation that plagues conventional metaheuristic optimizers in renewable forecasting. This improvement translates to more reliable day-ahead scheduling and reduced imbalance penalties in electricity markets.
• • Joint wind-PV power forecasting under SCO-BiLSTM lowers the eRMSE by 27.443% relative to standalone PV power forecasting. This magnitude of error reduction is industrially significant because it enables grid operators to lower reserve margins, thereby decreasing curtailment and improving asset utilization.
• • Forecasting accuracy exhibits a monotonic relationship with wind-solar complementarity: stronger complementarity yields higher joint forecasting precision. This empirical threshold effect implies that site selection and portfolio pairing should prioritize negative Pearson correlation coefficients to maximize forecasting gains.
• • Extending the forecasting horizon degrades joint forecasting accuracy, establishing a clear operational boundary for the method's applicability. For short-term dispatch (e.g., minutes to hours), the complementarity benefit is maximized; beyond this window, uncertainty propagation dominates and the advantage diminishes.