Optimization of BiLSTM for PV Output Prediction Based on Hybrid Bat Algorithm

Photovoltaic (PV) power generation is affected by the climatic environment and has a large degree of uncertainty. Accurate PV output prediction can reduce its uncertainty and ensure the safe operation of the power system. The paper introduces a predictive model for photovoltaic (PV) output that leverages a bidirectional long-short-term memory neural network (BiLSTM) optimized using the Hybrid Bat Algorithm (HBA). Firstly, the correlation analysis between environmental factors and PV output data is carried out using gray correlation analysis, and the environ mental factors with higher correlation coefficients are taken as the input features of the predictive model; secondly, the HBA is employed to optimize the optimal parameters within the BiLSTM network, resulting in the creation of the HBA-LSTM prediction model. Finally, the prediction analysis is carried out with the actual data of a certain region, and the findings demonstrate that the predictive precision of the approach introduced in this article is higher compared with the BiLSTM prediction method.