光伏发电功率预测方法的研究

发布时间：2018-05-15 05:24

本文选题：光伏发电 + 短期预测　；参考：《西华大学》2014年硕士论文

【摘要】：光伏发电输出功率的不稳定性会对所接入的电网造成冲击,因此需要对光伏发电功率进行预测以保证电网的合理调度。但是光伏发电功率在天气、云层、湿度、季节等多种因素的影响下,表现出十分复杂的非线性特性,难以精确预测。而且预测时段越长,预测误差也就越大。由于在光伏发电中,未来几个小时内的发电功率对电网调度具有非常重要和直接的影响,而且其预测精度一般比长时预测精度要高,因此,本课题主要针对光伏发电系统输出功率短期预测技术进行了研究。首先,本论文对目前常用的短期预测方法进行了介绍,分析了不同预测方法的理论基础及其特点；从分析中看出,神经网络预测技术作为其中一种具有智能化特点的预测手段,可以很好的处理非线性问题,能较好的适用于光伏发电功率的预测。然后,本论文介绍了BP神经网络的结构、学习规则和BP神经网络的建模过程,探讨了BP神经网络在光伏发电功率预测中的建模方法；并对两种典型的BP神经网络预测模型进行了实验分析。实例一表明,天气因素对预测模型精度的提高有着重要影响；实例二中加入了常见的天气分类这一因素,能在一定程度上提高预测精度,但是依然不能很好地解决预测过程中因天气因素而导致的误差突然加大,甚至模型失效的不足。本论文针对传统的BP神经网络预测算法存在天气分类过于简单、不能很好地符合待预测日天气类型的问题,通过分析光伏发电的影响因素,基于历史数据提出合理的假设,提出了基于太阳辐射功率曲线匹配的预测模型。该模型以太阳辐射功率曲线作为匹配标准,在历史数据库中查找与预测日太阳辐射曲线匹配的历史数据,然后利用匹配得到的相似日、相似时段的历史数据构建并训练神经网络进行预测,实验表明,该模型可以达到较好的预测效果。最后,将本论文所提出的预测模型与传统预测模型进行对比表明,本论文所建立的模型具有较好的短期预测能力,能达到较高的预测精度,对光伏发电短期预测能起到很好的指导作用。
[Abstract]:The instability of photovoltaic output power will impact the connected power grid, so it is necessary to predict the photovoltaic power generation power to ensure the reasonable dispatch of the grid. However, under the influence of weather, cloud, humidity, season and other factors, photovoltaic power has a very complex nonlinear characteristics, which is difficult to predict accurately. And the longer the prediction period, the greater the prediction error. In photovoltaic power generation, the power generation in the next few hours has a very important and direct impact on the power grid dispatching, and its prediction accuracy is generally higher than that in the long term. This paper mainly focuses on the short-term prediction technology of output power of photovoltaic power generation system. First of all, this paper introduces the current commonly used short-term forecasting methods, analyzes the theoretical basis and characteristics of different forecasting methods, and finds out from the analysis that the neural network prediction technology is one of the intelligent forecasting methods. It can deal with nonlinear problems well and can be applied to the prediction of photovoltaic power generation. Then, this paper introduces the structure of BP neural network, learning rules and BP neural network modeling process, and discusses the modeling method of BP neural network in photovoltaic generation power prediction. Two typical BP neural network prediction models are analyzed experimentally. Example 1 shows that the weather factors have an important effect on the accuracy of the prediction model, and the common weather classification is added to the second example, which can improve the prediction accuracy to a certain extent. However, the error caused by weather factors in the prediction process can not be solved well, even the deficiency of model failure. This paper aims at the problem that the traditional BP neural network forecasting algorithm is too simple to classify weather, which can not accord with the forecast weather type well. By analyzing the influencing factors of photovoltaic power generation, the reasonable assumptions are put forward based on the historical data. A prediction model based on solar radiation power curve matching is proposed. In this model, the solar radiation power curve is used as the matching standard, and the historical data matching with the predicted solar radiation curve is found in the historical database, and then the similar day is obtained by using the matching data. The historical data of similar periods are constructed and trained to predict. The experimental results show that the model can achieve a good prediction effect. Finally, the comparison between the proposed prediction model and the traditional prediction model shows that the model established in this paper has a better short-term forecasting ability and can achieve a higher prediction accuracy. It can play a good guiding role in short-term prediction of photovoltaic power generation.
【学位授予单位】：西华大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM615

【参考文献】