基于深度置信网络的风功率预测模型

A Deep Confidence Network-Based Wind Power Prediction Model

准确的风功率预测可以缓解风电并网对电力系统的负面影响.本文提出一种新的组合模型来提高风电功率预测精度.首先,对特征变量进行时间延迟实现数据重构,利用完全集成经验模态分解(CEEMDAN)将重构后的数据分解为趋势分量和波动分量.其次,利用最大互信息(MIC)选出低维敏感的变量.最后,低维变量输入置信神经网络(DBN)中进行风电功率预测.基于风机实际运行数据的实验结果表明,所建立模型预测结果的MAPE为3.41%,相比于对比模型取得了更高的预测性能.

Accurate wind power prediction can alleviate the negative impact of wind power grid connection on the power system.In this paper,a new combined model is proposed to improve the prediction accuracy of wind power.First,the feature variables are time-delayed to achieve data reconstruction.The reconstructed data were decomposed into trend and fluctuation components using complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN).Then,the maximum mutual information(MIC)is used to select low-dimensional sensitive variables.Finally,the low-dimensional variables are input into a deep belief networks(DBN)for wind power prediction.The experimental results based on the actual operation data of the fan show that the MAPE of the prediction results of the established model is 3.41%,which is higher than the comparison model.