大规模风电场超短期风电功率预测的误差分析与评价.docVIP

摘 要 摘 要 风力发电输出功率具有波动性、不可准确预测性的特点，当风电大规模接入后，将提升该区域发电计划难度，给电力系统的安全、经济运行带来了巨大的挑战。对风电场风电功率进行高精度预测可以在一定程度上缓解风电大规模并网对电网造成的一系列问题，其中超短期风电功率预测对含大规模风电的电力系统安全运行更是有着极重要的意义。 由于风电功率预测受地形、地貌、天气及风电场运行状态等因素影响，所以无论选取何种预测算法很难准确反映风电场的输出功率波动特性，难以避免要存在误差。因此，对风电场风电功率预测的误差进行分析与评价就显得至关重要，但是目前对预测结果的评价均停留在常规统计学指标上，缺乏合理的评价体系来评价某特定风电场所选取预测模型的优劣，无法全面挖掘预测误差中更有价值的信息。 本文以吉林省某风电场实测的风电功率数据为基础，首先对风电功率时间序列进行分析，即预测前对风电功率数据进行预处理，尽量掌握风电功率时间序列特性，为风电功率预测提供理论基础；其次结合预测误差的存在形式与原因，重点剖析误差与预测模型、误差与建模域长度之间的关系，以此全面了解风电功率预测误差的组成；在此基础上，提出了一种包含误差评价及预报考核的风电场输出功率实时预测效果评估方法，旨在从各个方面了解所用预测方法的优缺点，提出改进意见，从而更好地将预测结果应用到风电功率预测中去；在文章最后对该风电场未来 4h 输出功率进行实时预测，并用提出的预测效果评价方法对不同预测模型的预测结果进行了分析评价，结果表明该评价方法可以实现不同地区风电场根据其风电输出功率变化的特点选择预测模型，并且为风电场输出功率效果的工程检验提供依据。关键词：风电功率预测；超短期；误差；多指标分析 I 东北电力大学硕士学位论文 Abstract The wind power output power has the characteristics of fluctuation and it is difficult topredict accurately when large scale wind power access it will promote the difficulty of theregional power plan bring great challenge to the power system security and economicaloperation. The high precision of the wind power prediction can alleviate a series of problemswhich are caused by the integration of the large-scale wind on a certain extent in which thesuper short-term wind power forecasting of a very important significance for the large-scalewind power system safety operation. As the wind power forecasting is affected by topography weather and wind farm operationstatus and other factors so no matter choose which prediction algorithm it is very difficult toaccurately reflect the fluctuation of the output power characteristics of wind farms and the erroris hardly to avoid. Therefore the wind farm wind power prediction error analysis and evaluationbecomes particularly important however at present the evaluation of the predicted results arestuck on the conventional statistical indicators lacking a reasonable evaluation system toevaluate the merits of the prediction model which is selected by a particular wind farms andcannot be fully excavated t