Optimized wind power prediction and energy storage scheduling using genetic algorithm and backpropagation neural network

• Published in: International journal of renewable energy development Vol. 14; no. 1; pp. 146 - 157
• Main Authors: Wu, Peng, Li, Zongze
• Format: Journal Article
• Language: English
• Published: Diponegoro University 01.01.2025
• Subjects: backpropagation neural network; energy storage scheduling; genetic algorithm; support vector machine; wind power prediction
• ISSN: 2252-4940
• DOI: 10.61435/ijred.2025.60434

As renewable energy continues to rise in the global energy mix, wind energy is gradually increasing its share in the power system as a clean, renewable form of energy. However, the volatility and uncertainty of wind power bring new challenges to power system operation, making the need for its efficient prediction and intelligent dispatch more and more urgent. Based on this, a method combining genetic algorithm and backpropagation neural network is proposed for wind power prediction and energy storage scheduling. In this study, the improved genetic algorithm-backpropagation algorithm was generated by optimizing the weights and thresholds of the backpropagation neural network through the genetic algorithm, and optimizing the crossover and mutation processes of the genetic algorithm using similar block-order single-point crossover operator and shift mutation operator. Moreover, the improved genetic algorithm-backpropagation Neural Network wind energy prediction model was successfully constructed. Subsequently, the improved genetic algorithm was applied to search for the parameters of support vector machine and an improved genetic algorithm-support vector machine photovoltaic power generation prediction model was established. The experimental results showed that the average absolute percentage error of the improved genetic algorithm backpropagation neural network algorithm was 2.4%, and the accuracy was significantly higher than that of the traditional backpropagation neural network algorithm. The maximum photovoltaic prediction error of the autoregressive integral moving average model was about 80MW, while the photovoltaic prediction error of the improved genetic algorithm support vector machine photovoltaic prediction model was only about 12kW. In addition, the average absolute percentage error of the improved genetic algorithm support vector machine photovoltaic prediction model was only 1.53%, which was only 0.2% higher than the support vector machine prediction model. This study not only improves the stability of the power grid but also provides a practical and feasible method for realizing the large-scale application of clean energy, making a positive contribution to the sustainable development of the energy industry.