Development and Performance Analysis of Aquila Algorithm Optimized SPV Power Imputation and Forecasting Models

Solar photovoltaic (SPV) generators play a vital role in the global pursuit of sustainable energy. However, the intermittent nature of SPV generators poses a challenge to power system planning. Accurate long-term SPV power forecasting is essential, but obtaining high-quality training dsecata for many years is cumbersome due to faults in sensors, communication, or databases, leading to missing values. This paper proposes an attention network (AN) for SPV power imputation to address missing historical data. In addition, the SPV power forecasting model using BDLSTM NN is proposed for forecasting the SPV power in the long-term horizon. During training, the weights and biases of the imputation and forecasting models are proposed to be optimized using the Aquila optimization algorithm (AOA). Training and testing results substantiate that the proposed imputation and forecasting models consistently attain superior accuracy with a reduction in training time regardless of the missing rates, climatic patterns, and seasons as compared to benchmark models.