并网型光伏电站的发电功率短期预测研究与实现

发布时间：2018-07-17 20:34

【摘要】：光伏发电因其清洁无污染受到了各国的青睐。但同时因其输出功率的随机性和波动性,使得当光伏电站并网后会对公用电网造成一定的冲击。这对电网的正常运行极其不利。因此对并网型光伏电站的发电功率进行准确的短期预测,有利于电力调度部门合理分配和规划光伏能源和常规能源所占的比重,有利于及时调整调度计划,使得电力系统以安全、稳定、经济的方式运行。本文在大量阅读国内外相关文献的基础上,采用甘肃某光伏电站的历史数据作为研究对象,并利用HS-ESN模型建立了预测模型。然后对该电站进行了光伏发电功率的短期预测分析。最后,通过C#与MATLAB的混编,介绍了光伏发电功率短期预测系统的设计。基于上文描述,本文的研究内容主要包括以下几个方面:首先,利用MATLAB建立的光伏电池仿真模型梳理了光伏电池的输出特性,分析了太阳辐照强度和温度对光伏发电功率的影响,进一步确定了影响光伏发电功率的主要因素。然后将各影响因素组成特征向量,利用相似日选择算法提取相似日和训练样本。其次,在深入研究了回声状态网络(Echo State Network,ESN)算法的前提下,提出了和声搜索(Harmony Search,HS)算法优化回声状态网络算法的混合算法。本文利用HS算法优化ESN算法的储备池参数,有效提高了ESN算法的精度,并将HS-ESN算法应用到光伏发电功率短期预测问题中,通过不同预测模型分别预测不同天气类型下的光伏发电功率,验证了引入相似日选择算法的有效性,同时表明了HS-ESN模型的性能更优于单一的ESN模型和目前常用的其他几种模型。最后,根据本文提出的预测模型和方法,设计出了一个光伏发电功率短期预测系统,并对该系统进行了需求分析和结构设计。该预测系统具备光伏发电功率短期预测的基本模块,包括登录界面及数据导入模块、发电功率短期预测模块、报表统计模块、数据查询及输出模块,因此具有一定的实用价值。
[Abstract]:Photovoltaic power generation is favored by many countries because of its clean and pollution-free. However, because of the randomness and fluctuation of the output power, the photovoltaic power station will have a certain impact on the public power grid when it is connected to the grid. This is extremely detrimental to the normal operation of the power grid. Therefore, the accurate short-term prediction of the generation power of grid-connected photovoltaic power stations is beneficial to the rational distribution and planning of the proportion of photovoltaic and conventional energy in the power dispatching department, and to the timely adjustment of the dispatching plan. Make the power system run in a safe, stable and economical way. Based on a large number of domestic and foreign literatures, this paper takes the historical data of a photovoltaic power plant in Gansu as the research object, and establishes a prediction model by using the HS-ESN model. Then the short-term prediction analysis of PV power generation is carried out. Finally, by mixing C # and MATLAB, the design of PV power short-term prediction system is introduced. Based on the above description, the research contents of this paper mainly include the following aspects: firstly, the output characteristics of photovoltaic cells are combed by using the simulation model established by MATLAB. The influence of solar radiation intensity and temperature on photovoltaic power generation is analyzed, and the main factors affecting photovoltaic power generation are determined. Then each factor is composed of feature vector and the similar day selection algorithm is used to extract the similar day and training sample. Secondly, based on the in-depth study of echo state network (Echo State Network) algorithm, a hybrid algorithm is proposed to optimize echo state network algorithm based on Harmony search (HS) algorithm. In this paper, HS algorithm is used to optimize the reserve cell parameters of ESN algorithm, and the precision of ESN algorithm is improved effectively. The HS-ESN algorithm is applied to the short-term prediction of photovoltaic power generation. The effectiveness of introducing the similar day selection algorithm is verified by forecasting PV power under different weather types by different prediction models. The performance of HS-ESN model is better than that of single ESN model and other models commonly used at present at the same time the performance of HS-ESN model is better than that of single ESN model and other commonly used models. Finally, according to the prediction model and method proposed in this paper, a PV power short-term forecasting system is designed, and the demand analysis and structure design of the system are carried out. The prediction system has the basic modules of short-term prediction of photovoltaic power generation, including login interface and data import module, short-term power generation prediction module, report statistics module, data query and output module, so it has certain practical value.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM615

【参考文献】