基于浅层方法和深度网络集成的短期风电功率预测

Short⁃term wind power prediction based on shallow method and deep network integration

风电功率的预测精度受到多种因素的影响,为进一步提高预测精度,提出一种基于浅层方法和深度网络集成的短期风电功率预测模型(GRA⁃GWO⁃SVR⁃AdaBoost⁃GRU),集成灰色关联度分析(GRA)、支持向量回归机(SVR)、自适应提升集成(AdaBoost)和门控循环单元(GRU)等多种模型/方法。首先采用GRA计算变量之间的相关程度,选择相关性高的3个特征作为模型的输入;其次利用GWO算法对SVR的惩罚参数和核函数参数进行优化,建立GWO⁃SVR预测模型;然后采用AdaBoost集成模型构建强回归器进行预测;最后采用GRU模型对预测误差进行修正,将修正后的误差与预测结果进行叠加,得到最终预测值。仿真结果表明,该模型的预测结果的均方根误差和R⁃Square显著优于其他传统模型,有效提高了风电功率的预测精度。

The wind power prediction accuracy is affected by many factors.In order to improve the prediction accuracy,a short⁃term wind power prediction model combining gray correlation analysis(GRA),support vector regression(SVR),adaptive boosting model(AdaBoost)and gated recurrent unit(GRU)is proposed,which is based on the shallow method and deep network integration.GRA is used to calculate the correlation between variables and three features with high correlation are selected as the input of the model.The GWO algorithm is used to optimize the penalty parameter and kernel function parameter of SVR to establish the GWO⁃SVR prediction model.Then,the AdaBoost integrated model is used to construct a strong regressive device for wind power prediction.The GRU model is used to correct the prediction error.The corrected error is superimposed with the prediction value to get the final prediction value.Simulation results show that the root mean square error and R⁃Square of the prediction results of this model are significantly better than that of the other traditional models,which effectively improves the accuracy of wind power prediction.