基于BP神经网络与WRF模式的风电功率预测系统设计与应用

发布时间：2018-02-10 10:58

本文关键词： BP神经网络 WRF模式风电功率预测数据同化风速订正　出处：《中国科学院大学(中国科学院工程管理与信息技术学院)》2017年硕士论文　论文类型：学位论文

【摘要】：当前,作为一种储量丰富成本低廉的清洁能源,风能日益受到了世界各国的关注。然而风能资源本身固有的随机性、间歇性和不可控性等特点,又极大的制约着风力发电的大规模开发和利用。在分析和总结国内外相关风电功率预报多种实践案例的基础上,贴合本地化的风电功率预报特点,设计和实现了一套基于BP神经网络和WRF模式的可交互视图的预报与分析工具,力求对风电输出功率及时准确的预测能改善电网接纳风电能力,缓解大容量的风电给当地区域电网的稳定运行带来严峻的挑战方面。本文在论述风电功率预测系统的设计和实现过程中,总结分析了国内外风电功率预报模型发展和现状以及在实际应用所需的关键技术;在风电功率预测的业务需求分析的基础上给出风电功率预测系统的总体框架设计以及相关的业务预测模型;提出了开发风电功率预测系统的整体解决方案和技术路线,并详细描述了风电功率预报模型的实现过程,以及最终达到的实际应用效果,最后给出了结论和展望。论文的成果在于,风电功率预测系统中短期功率预测模型和超短期功率预测模型分别应用了数值气象计算和人工神经网络等关键技术,其中不仅包括了中尺度WRF(Weather Research Forecast)模式、数据同化、风速订正等短期有限区域数值气象预报模式的应用;同时超短期预报模型中还深入应用了BP(Back Propagation)神经网络以及曲线拟合等数据挖掘算法,并将两个预测模型整合进了同一个业务预测系统,做成了一个独立通用的预测数据生成引擎,满足了业务预测模型在应对灵活多样的业务需求上的扩展性,它不但可以进行风速定量预测,还可以进行多种气象要素的定量预测,能很好的适应了开展本地化的风电功率预报,以及影响风电功率的多种气象要素的客观定量分析这一课题目标。其全面完整的架构设计和技术实现,是BP神经网络和中尺度WRF模式在风电功率预报领域综合应用的一种很好尝试。
[Abstract]:At present, as a clean energy with abundant reserves and low cost, wind energy has attracted more and more attention from all over the world. However, wind energy resources are inherently random, intermittent and uncontrolled. It also restricts the large-scale development and utilization of wind power generation. On the basis of analyzing and summarizing many practical cases of wind power prediction at home and abroad, the characteristics of localized wind power prediction are combined. A set of interactive view prediction and analysis tools based on BP neural network and WRF model are designed and implemented. This paper discusses the design and implementation of wind power forecasting system in this paper. The development and present situation of wind power prediction model at home and abroad and the key technologies needed in practical application are summarized and analyzed. Based on the analysis of operational requirements for wind power forecasting, the overall framework design and related business prediction model of wind power forecasting system are presented, and the overall solution and technical route of developing wind power forecasting system are presented. The realization process of wind power prediction model and its practical application effect are described in detail. Finally, the conclusion and prospect are given. In the wind power prediction system, the short-term power prediction model and the ultra-short-term power prediction model apply the key techniques of numerical weather calculation and artificial neural network, respectively, which include not only the mesoscale WRF(Weather Research forecasting model, but also the data assimilation. At the same time, the data mining algorithms such as BP(Back propagation neural network and curve fitting are also used in the ultra-short-term forecasting model. The two forecasting models are integrated into the same business forecasting system, and an independent and general forecasting data generation engine is made, which satisfies the expansibility of the business forecasting model in dealing with flexible and diverse business requirements. It not only can make quantitative prediction of wind speed, but also can carry out quantitative prediction of various meteorological elements. It can well adapt to the localized wind power forecast. And the objective of objective quantitative analysis of various meteorological elements affecting wind power. Its comprehensive and complete architecture design and technical realization, It is a good attempt to apply BP neural network and mesoscale WRF model in wind power prediction.
【学位授予单位】：中国科学院大学(中国科学院工程管理与信息技术学院)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM614;TP183

【相似文献】