光伏发电系统发电能力预测研究

发布时间：2018-07-29 20:53

【摘要】：太阳能以其清洁、无污染、可再生等特点越来越受到关注，光伏发电已成为当今世界可再生能源领域的研究热点。然而，光伏功率具有不确定性和间歇性等特点，大规模光伏并网运行会增加电网调度的难度，影响电力系统的安全、稳定及经济运行。精确预测光伏功率是有效减缓大规模光伏并网对电网不利影响的前提，对电网调度计划、常规能源规划和光伏发电规划等具有重要的指导意义。本文以光伏发电功率为研究对象，通过分析光伏发电功率的影响因素，对光伏发电系统发电能力预测问题展开研究。首先，分析了季节类型、天气类型及辐照强度、环境温度、云量等气象因子对光伏功率的影响，确定了预测模型的输入变量，并提出利用相似日理论确定训练样本的方法；其次，分析了传统BP算法的优缺点，提出了基于动量项-陡度因子-自适应学习率的改进BP算法，并建立了相应的光伏功率预测模型；然后，针对改进BP算法和PSO算法的不足，将混沌搜索和自适应变异思想引入到粒子群算法中，以提高算法的全局收敛概率和速度，，建立了基于混沌搜索的AMPSO-BPNN的光伏功率预测模型，并提出利用相似日功率修正模型预测结果的方法；最后，依托实际光伏电站和气象观测站对预测模型进行了实例验证，并通过分析光伏出力与负荷用电间的相关性，进一步明确了光伏电站调度运行的研究方向。在Microsoft Visual C++6.0环境下编制了光伏发电系统功率预测软件，对比分析了不同光伏功率预测模型的优化性能，预测结果表明所提模型及算法具有较高的预测精度和收敛速度，且基于相似日功率的修正方法具有一定的可行性。
[Abstract]:Solar energy has attracted more and more attention because of its clean, pollution-free and renewable characteristics. Photovoltaic power generation has become a research hotspot in the field of renewable energy in the world. However, photovoltaic power has the characteristics of uncertainty and intermittency. Large-scale photovoltaic grid-connected operation will increase the difficulty of power grid dispatching and affect the security, stability and economic operation of power system. Accurate prediction of photovoltaic power is the premise to effectively mitigate the adverse effects of large-scale photovoltaic grid connection, and has important guiding significance for grid scheduling, conventional energy planning and photovoltaic generation planning. In this paper, the photovoltaic power generation as the research object, through the analysis of the factors affecting photovoltaic generation power, PV power generation capacity prediction problem is studied. Firstly, the effects of seasonal type, weather type and radiation intensity, ambient temperature, cloud amount on photovoltaic power are analyzed, and the input variables of the prediction model are determined, and a method to determine the training samples by using the similarity day theory is proposed. Secondly, the advantages and disadvantages of the traditional BP algorithm are analyzed, and an improved BP algorithm based on momentum term, steepness factor and adaptive learning rate is proposed, and the corresponding photovoltaic power prediction model is established. Chaotic search and adaptive mutation are introduced into particle swarm optimization to improve the global convergence probability and speed of the algorithm. A photovoltaic power prediction model based on chaotic search for AMPSO-BPNN is established. The method of using the similar daily power correction model to forecast the results is put forward. Finally, the prediction model is verified by the actual photovoltaic power station and the meteorological observation station, and the correlation between the photovoltaic output and the load consumption is analyzed. The research direction of photovoltaic power plant dispatching and operation is further clarified. The power prediction software of photovoltaic power generation system is developed under Microsoft Visual C 6.0. The optimized performance of different photovoltaic power prediction models is compared and analyzed. The prediction results show that the proposed model and algorithm have higher prediction accuracy and convergence speed. And the correction method based on similar day power has certain feasibility.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM615

【参考文献】