Abstract: Traditional deterministic point forecasting methods struggle to effectively address the strong randomness and high volatility of renewable energy output, resulting in high prediction uncertainty, slow convergence, and insufficient generalization capabilities. To enhance the accuracy and stability of wind and solar output forecasting, this paper proposes a forecasting model based on the crown porcupine optimization algorithm and a bidirectional long short-term memory network. This model employs CPO for adaptive optimization of BiLSTM hyperparameters, accelerating model convergence and enhancing global search capabilities. Simultaneously, it leverages BiLSTM’s bidirectional temporal feature extraction mechanism to fully exploit temporal dependencies within wind and solar output data. Experimental results based on actual wind/solar power and meteorological data from a region demonstrate that the CPO-BiLSTM model outperforms other comparison models in evaluation metrics including RMSE, MAE, accuracy, and pass rate. This validates the effectiveness and superiority of the proposed model in wind and solar power output forecasting.