Short-term Photovoltaic Power Prediction Based on Complete Ensemble Empirical Mode Decomposition With Adaptive Noise-sample Entropy-bidirectional Long Short-term Memory

Photovoltaic power generation exhibits the characteristics of intermittence and great fluctuation, posing challenges for the traditional single model to achieve accurate prediction. Therefore, a prediction model is proposed based on a combination of complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), sample entropy (SE) and bi-directional long-short-term memory (Bi-LSTM). Firstly, the historical power sequences are decomposed using CEEMDAN to mitigate its non-stationarity. The subsequent sequences are reorganized by incorporating sample entropy to address the issue of increased data size in subsequent prediction after the decomposition. Secondly, the reorganized sequences are fed into a Bi-LSTM network for training and prediction. Finally, the final prediction results are obtained by linearly summing up the prediction results of each reorganized sequence. The case validation demonstrates that the constructed combined model is suitable for PV power prediction under diverse weather conditions and exhibits higher prediction accuracy.