改进BBO优化BP神经网络的短期风电功率预测模型

Short-Term Wind Power Prediction Model Based on Improved BBO and Optimized BP Neural Network

为了提高预测模型在处理风电功率时间序列数据中的复杂模式和非线性特征时的识别能力,提出了一种新的预测模型。通过改进完全自适应噪声集合经验模态分解算法进行信号处理,然后根据改进生物地理学优化算法对反向传播神经网络进行初始权重优化,进一步提升短期风电功率预测的准确度和稳定性。通过实际应用案例表明,与其他优化算法相比,提出的模型在MAE、RMSE和MAPE上的表现分别平均提高了43.21%、37.98%和36.84%,显示出更高的预测准确度,仿真结果验证了本方法在短期风电功率预测领域的效果及其明显的优势。

To improve the recognition ability of the prediction model in dealing with complex patterns and nonlinear features in wind power time series data,a new prediction model is proposed in this study.Signal processing is performed by the improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise algorithm,and then the initial weights of the Back Propagation Neural Network are optimized according to the improved BBO algorithm to further improve the accuracy and stability of short-term wind power prediction.Practical application cases show that compared with other optimization algorithms,the performance of the proposed model on MAE,RMSE and MAPE is improved by 43.21%,37.98%and 36.84%on average,respectively,showing higher prediction accuracy.Simulation results validate the effect and obvious advantages of the proposed method in the field of short-term wind power prediction.