基于组合模型的光伏功率预测系统研究

发布时间：2018-11-08 10:29

【摘要】：随着化石能源的枯竭,研究新能源的使用已然迫在眉睫。太阳能作为一种清洁新能源,已经在很多领域中得到了推广。光伏发电就是太阳能资源的典型应用,已经正在从独立系统向大规模并网发展。太阳能发电与电站的地理环境以及气象条件有着密切的关系,而且光伏发电具有一定的波动性和不稳定性,这对大规模并网带来了一定的压力,所以对光伏功率输出进行合理的预测将会对并网以及合理的调度带来理论依据。目前,对光伏功率预测的研究主要有两个方向。一种是通过建立本地太阳能辐照预测模型,通过气象数据对辐照度进行预测,最后再通过拟合进而求得预测功率数据。另外一种,则是通过光伏电站的历史运行数据进行统计分析,对功率数据直接进行预测,这种方法简单且不用建立复杂的预测模型。从算法角度看,神经网络、小波分析、支持向量机都是预测模型普遍采用的训练方法,而单一的预测算法具有一定的局限性,并不适合在多种场景下应用,所以本文将提出一种基于组合模型的光伏功率预测系统,提高预测的精确度,减轻光伏发电对并网的影响。
[Abstract]:With the depletion of fossil energy, it is urgent to study the use of new energy. As a clean new energy, solar energy has been popularized in many fields. Photovoltaic power generation is a typical application of solar energy resources, and has been developing from independent system to large-scale grid-connected. Solar power generation is closely related to the geographical environment and meteorological conditions of the power station, and photovoltaic power generation has certain volatility and instability, which brings certain pressure to large-scale grid connection. Therefore, reasonable prediction of photovoltaic power output will bring theoretical basis to grid connection and reasonable scheduling. At present, the research on photovoltaic power prediction has two main directions. One is to establish the local solar radiation prediction model, to predict the irradiance by meteorological data, and finally to obtain the predicted power data by fitting. On the other hand, the power data is predicted directly through the statistical analysis of the historical operation data of the photovoltaic power station. This method is simple and does not need to establish a complicated prediction model. From the point of view of algorithm, neural network, wavelet analysis and support vector machine are generally used training methods for prediction model, but the single prediction algorithm has some limitations and is not suitable for application in many scenarios. Therefore, this paper proposes a PV power prediction system based on combinatorial model to improve the accuracy of prediction and reduce the impact of photovoltaic power generation on grid connection.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM615

【相似文献】