基于神经网络的光伏发电功率预测研究

发布时间：2018-04-12 15:56

  本文选题：光伏发电功率 + 粒子群优化算法　； 参考：《沈阳工程学院》2017年硕士论文

【摘要】：随着人类对能源需求的增加,太阳能的利用受到了越来越广泛的关注,光伏发电因其无污染、可再生的特点,作为一种可持续的能源,近年来受到了国内外学者们的广泛关注。但光伏发电具有间歇性、随机性,光伏发电功率的波动会影响电网的稳定性及电能质量,因此光伏发电预测技术是一项亟需深入研究的工作,准确预测光伏发电功率对电网的优化调度、电能质量均有较好影响。本文首先通过华能营口热电有限责任公司光伏观测站数据分析了光伏发电功率的波动特性及其影响因素,通过分析表明,季节、天气类型对光伏发电功率的波动特性具有重要影响。基于传统粒子群算法存在粒子“早熟”的问题,本文通过对粒子群算法改进,删除了惯性权重并增加了随机因子,提高了粒子群算法的全局收敛性。用粒子群算法优化了神经网络光伏发电预测模型,通过预测数据与实测数据的比较,验证了本文所提出方法的有效性。基于以上理论研究,本文设计了光伏发电功率预测系统,可以将本文所提出的算法应用到实际中。本文设计的光伏发电功率预测系统,按照电力二次系统安全防护规定,配备了反向物理隔离装置以保障数据传输的安全。此外,还设计了该系统的软件结构与硬件结构,并对该系统的功能加以展示。通过本文的分析可以得出结论:季节、天气类型对光伏发电功率具有较大影响,在预测光伏发电功率时应将同季节的天气类型相似日功率作为神经网络模型的输入层;删除了惯性权重并增加了随机搜索因子可以提高粒子群算法的全局搜索能力,从而得到更好的神经网络模型以准确预测光伏发电功率;晴天的预测效果要明显优于云天和雨雪天,因此云天和雨雪天气的光伏发电功率波动规律还有待更深入研究。
[Abstract]:With the increasing demand for energy, the use of solar energy has attracted more and more attention. Photovoltaic power generation, as a kind of sustainable energy, has been widely concerned by scholars at home and abroad in recent years because of its pollution-free and renewable characteristics.However, photovoltaic generation is intermittent, random, and the fluctuation of photovoltaic power will affect the stability and power quality of power grid. Therefore, photovoltaic generation prediction technology is a work that needs to be studied deeply.Accurate prediction of photovoltaic power generation has a good impact on power quality.In this paper, firstly, the fluctuation characteristics of photovoltaic power and its influencing factors are analyzed through the data of photovoltaic observation station of Huaneng Yingkou Thermal Power Co., Ltd.The weather type has an important influence on the fluctuation characteristics of photovoltaic power generation.Based on the problem of precocity of particles in the traditional particle swarm optimization algorithm, the inertia weight is removed and the random factor is added to improve the global convergence of the particle swarm optimization algorithm.The prediction model of photovoltaic generation based on neural network is optimized by particle swarm optimization. The validity of the proposed method is verified by comparing the predicted data with the measured data.Based on the above theoretical research, a photovoltaic power prediction system is designed, which can be applied to practice.The photovoltaic power prediction system designed in this paper is equipped with reverse physical isolation device to ensure the safety of data transmission according to the safety protection regulations of the secondary power system.In addition, the software structure and hardware structure of the system are designed, and the functions of the system are demonstrated.Through the analysis of this paper, we can draw a conclusion: season, weather type has a great influence on photovoltaic power generation, in the prediction of photovoltaic power generation, we should take the similar daily power of the same season weather type as the input layer of the neural network model;Removing inertial weight and adding random search factor can improve the global search ability of PSO and obtain a better neural network model to accurately predict photovoltaic power generation.The forecasting effect of sunny weather is obviously better than that of cloud and rain, so the fluctuation of photovoltaic power in cloudy and rainy weather still needs to be further studied.
【学位授予单位】：沈阳工程学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM615

【相似文献】

中国期刊全文数据库 前10条

1 刘士荣;李松峰;宁康红;周啸波;荣延泽;;基于极端学习机的光伏发电功率短期预测[J];控制工程;2013年03期

2 ;科技文摘[J];中学物理教学参考;1994年08期

3 王丽婕;廖晓钟;高阳;高爽;;风电场发电功率的建模和预测研究综述[J];电力系统保护与控制;2009年13期

4 田丽;邓阅;;联合条件下风力发电风速预测[J];安徽工程大学学报;2012年03期

5 卢静;翟海青;刘纯;王晓蓉;;光伏发电功率预测统计方法研究[J];华东电力;2010年04期

6 徐星;张虹;乐海洪;徐敏;;采用气象信息的神经网络应用于短期风力发电功率预测[J];南昌大学学报(工科版);2011年01期

7 阿碧;;身背核电站移民外星[J];发明与创新(综合科技);2011年11期

8 许昌;李e，

本文编号：1740414