基于EWT-KMPMR组合模型的光伏电站短期功率预测

发布时间：2018-12-19 21:28

【摘要】：为提高光伏电站短期功率预测的精度,提出一种基于经验小波变换(empirical wavelet transform,EWT))和核最小最大概率回归机(kernel mini max probability machine regression,KMPMR)的组合预测模型,对晴天、阴天和雨天3种天气类型下的光伏电站出力分别进行了预测分析。该文首先采用EWT将相似日光伏功率序列分解为具有特征差异的AM-FM分量,然后根据各AM-FM分量的变化特点建立相应的KMPMR预测模型分别进行预测并叠加得到最终预测结果。试验结果表明,相比SVM方法,该文方法在晴天、阴天和雨天可提高预测精度(MAE)分别为56.19%、54.15%和76.33%;相比EMD-KMPMR方法,在降低近一半左右计算规模的同时,可提高预测精度(MAE)分别为9.42%、38.74%和64.52%。以阿克苏地区光伏电站实际运行数据进行试验验证表明,该文方法在3种天气类型下均可取得较高的预测精度。
[Abstract]:In order to improve the accuracy of short-term power prediction for photovoltaic power plants, a combined prediction model based on empirical wavelet transform (empirical wavelet transform,EWT) and kernel minimum maximum probability regression machine (kernel mini max probability machine regression,KMPMR) is proposed. The output of photovoltaic power stations under cloudy and rainy weather types are predicted and analyzed respectively. In this paper, the similar diurnal photovoltaic power series is decomposed into AM-FM components with characteristic difference by using EWT, then the corresponding KMPMR prediction model is established according to the variation characteristics of each AM-FM component and the final prediction results are obtained by superposition. The experimental results show that compared with SVM method, this method can improve the prediction accuracy on sunny, cloudy and rainy days by 56.19% and 76.33% respectively. Compared with the EMD-KMPMR method, the calculation scale is reduced by about half, and the prediction accuracy is improved by 9.42% and 64.52%, respectively, with the (MAE) of 9.42% and 64.52% respectively. The experimental results of the actual operation data of the photovoltaic power station in Aksu area show that the proposed method can achieve high prediction accuracy under three weather types.
【作者单位】：国网新疆电力有限公司电力科学研究院电网技术中心;新疆铁道职业技术学院;
【基金】：新疆维吾尔自治区重大科技专项发展计划项目:光伏发电关键技术研究与应用(2016A02004) 国家电网公司发展计划项目:基于串阻型逆变器的光伏电站并网特性实证性研究与测试(5230DK160006)
【分类号】：TM615

【相似文献】