Photovoltaic Power Forecasting Based on Nutcracker Optimization Algorithm Improved Bidirectional Long Short-Term Memory

With the rapid expansion of photovoltaic power plants in recent years over the world, accurate power prediction of photovoltaic generation is essential for ensuring the safe operation of power grids, developing effective dispatching plans, emission reductions, smart grid integration, and enhancing the efficiency of renewable energy usage. To improve the accuracy of photovoltaic power prediction, this paper proposes a nutcracker optimization algorithm to finely tune the extended parameters and enhance the information-capturing ability of the bidirectional long short-term memory method. The results indicate that the proposed approach demonstrates superior performance in the photovoltaic power prediction task. Compared with the other recurrent neural network methods, the mean squared error does not exceed 546.81 kW2. The mean absolute error is 15.35 kW. The root mean square error is 45.93 kW, and the R2 value is greater than 0.98.