基于样本熵和极端学习机的超短期风电功率组合预测模型

Combined Model for Ultra Short-term Wind Power Prediction Based on Sample Entropy and Extreme Learning Machine

该文提出一种经验模态分解(empirical mode decomposition,EMD)–样本熵(sample entropy,SE)和极端学习机(extreme learning machine,ELM)相结合的风电功率超短期预测方法。该方法首先利用EMD-SE将风电功率时间序列分解为一系列复杂度差异明显的风电子序列;其次利用最小二乘支持向量机(least squares support vector machine,LSSVM)、极端学习机和经原始岭回归(primal ridgeregression,PRR)改进的极端学习机(PRR-ELM)对各子序列建立组合预测模型,并采用交叉验证法和重构相空间法确定各模型的参数和输入向量维数,以提高各组合模型的预测精度;最后以某一风电场实际采集的数据为算例,结果表明基于EMD-SE理论的ELM和PRR-ELM组合预测模型在预测精度和训练速度上都明显优于EMD-SE理论和LSSVM的组合模型,且其预测结果更接近于真实值,为实现风电功率在线的较高精度超短期预测提供了可能。

An ultra short-term wind power combined prediction approach based on empirical mode decomposition （EMD）-sample entropy （SE） and extreme learning machine （ELM） was proposed. Firstly, the wind power time series was decomposed into a series of wind power subsequences with obvious differences in complex degree by using EMD-SE. Secondly, the prediction models of each subsequence were constructed with least squares support vector machine （LSSVM）, extreme learning machine （ELM） and ELM improved by primal ridge regression （PRR-ELM）, of which the parameters and the input vector dimensions were determined by cross validation and chaotic phase space theory to improve the forecasting accuracy of each prediction modek Finally, taking the actual collecting data of certain a wind farm for an example, the simulation results illustrate that ELM and PRR-ELM prediction model based on EMD-SE are much better than the combined LSSVM model based on EMD-SE on forecasting accuracy and training speed, and the prediction results of ELM are closer to the actual value, by which it is possible to achieve the online ultra short-term wind power combined prediction with higher precision.