An Effective Hybrid NARX-LSTM Model for Point and Interval PV Power Forecasting

This paper proposes an effective Photovoltaic (PV) Power Forecasting (PVPF) technique based on hierarchical learning combining Nonlinear Auto-Regressive Neural Networks with exogenous input (NARXNN) Long Short-Term Memory (LSTM) model. First, the NARXNN model acquires data to generate a residual error vector. Then, stacked LSTM model, optimized by Tabu search algorithm, uses correction associated original produce point and interval PVPF. The performance of proposed PVPF was investigated using two real datasets different scales locations. comparative analysis NARX-LSTM twelve existing benchmarks confirms its superiority in terms accuracy measures. In summary, has following major achievements: 1) Improves prediction models; 2) Evaluates uncertainties forecasts high accuracy; 3) Provides generalization capability for PV systems scales. Numerical results comparison method real-world Australia USA demonstrate improved accuracy, outperforming benchmark approaches overall normalized Rooted Mean Squared Error (nRMSE) 1.98% 1.33% respectively.