Short-Term PV Power Prediction Based on NGO-BILSTM

The accuracy of load forecasting can be impacted by changes in external elements, such as the environment, in short-term PV power forecasts. This research suggests the NGO-BILSTM prediction model, which combines the Bi-directional Long Short-term Memory (BILSTM) network and the Northern Goshawk Optimization (NGO) algorithm, to solve this issue. First, high correlation features are selected as input data by Pearson correlation analysis, then the hyperparameters of BILSTM are optimized by the NGO algorithm. The NGO-BILSTM prediction model is then established based on the optimized parameters, and the prediction is carried out on the dataset. The experimental prediction findings demonstrate that the NGO-BILSTM model's mean absolute error, root mean square error, and linear regression coefficient index are, respectively, 1.434, 1.809, and 0.972, which are better than those of other comparable models, demonstrating the model's efficacy.