光伏功率预测在风光储系统中的应用

发布时间：2018-05-31 04:37

本文选题：光伏发电功率预测 + 组合预测　；参考：《华北电力大学》2014年硕士论文

【摘要】：光伏功率预测可以有效的避免并网光伏发电系统输出功率间歇性和不可控性等缺点对电网的冲击,因此对光伏发电系统进行发电量预测具有十分重要的意义。本文综述了光伏功率预测研究现状和预测方法,针对光伏功率预测精度问题,提出了基于熵权法的光伏输出功率组合预测模型和基于组合权重相似日选取方法的光伏输出功率预测模型,并提出基于光伏功率预测结果的风光储系统平滑输出控制策略。首先,本文提出了基于熵权法的光伏输出功率组合预测模型,该方法组合基于待预测日前一天功率的持续法预测模型、支持向量机预测模型和相似数据预测模型,采用熵权法确定三种模型的组合预测权重系数,建立了基于熵权法的光伏输出功率组合预测模型。Matlab仿真结果表明基于熵权法的光伏输出功率组合预测模型提高了预测精度,对比三种单一预测模型,预测结果最大相对误差和均方根误差都有所减小,并且基于熵权法的光伏输出功率组合预测模型能够适应天气类型变化,在不同的天气类型下的预测效果都较好,适合工程应用。其次,针对气象条件相似天光伏输出功率曲线具有很高的关联度,本文提出了基于组合权重法选取相似日的光伏输出功率预测方法。基于组合权重法的相似日选取方法,首先选择太阳辐照度为相似变量,采用组合权重相似日选取方法确定各历史天与待预测天相似误差,选出相似误差最小的3个历史天确定为待预测日的相似天。将相似天光伏输出功率的平均值作为预测日光伏输出功率预测值。该预测方法的关键是相似天选取时各基值点组合权重系数的恰当确定,本文先确定各基值点的主观权重系数和客观熵权,再采用最小鉴别信息原理融合上述两种权重系数,得到组合权重系数。Matlab仿真对比表明,基于组合权重法选取相似日的光伏输出功率预测方法能够选出相似程度很高的相似天,提高了光伏输出功率的预测精度。最后,本文提出基于光伏功率预测的风光储系统平滑输出控制策略。该方法以光伏输出功率预测为基础,利用下一时刻功率预测值,估算下一时刻电池荷电状态的变化趋势,从而调整储能电池充放电量,该方法能够实现系统输出功率平滑控制并保持储能电池系统SOC稳定在正常范围。采用Matlab仿真,验证了该控制策略的有效性。
[Abstract]:Photovoltaic power prediction can effectively avoid the impact of intermittent and uncontrollable output power of grid-connected photovoltaic system, so it is of great significance to predict the power generation of photovoltaic system. In this paper, the current situation and prediction methods of photovoltaic power prediction are reviewed, aiming at the problem of photovoltaic power prediction accuracy. A combination prediction model of photovoltaic output power based on entropy weight method and a photovoltaic output power prediction model based on the method of combination weight similarity day selection are proposed, and a smooth output control strategy for wind-storage system based on PV power prediction results is proposed. First of all, this paper presents a photovoltaic output power combination prediction model based on entropy weight method, which is based on the continuous prediction model, support vector machine prediction model and similar data prediction model, which is based on the power prediction model one day before the day to be forecasted. The combined prediction weight coefficient of three models is determined by entropy weight method. The combined prediction model of photovoltaic output power based on entropy weight method is established. The simulation results of Matlab show that the combined prediction model of photovoltaic output power based on entropy weight method improves the prediction accuracy. Compared with the three single prediction models, the maximum relative error and root mean square error of the prediction results are all reduced, and the photovoltaic output power combination prediction model based on entropy weight method can adapt to the change of weather type. The prediction results of different weather types are good and suitable for engineering application. Secondly, in view of the high correlation degree of photovoltaic output power curve with similar weather conditions, this paper proposes a photovoltaic output power prediction method based on combination weight method to select similar days. Based on the combination weight method, the similar day selection method is used to select solar irradiance as a similar variable, and the similarity error between each historical day and the day to be predicted is determined by the combination weight similarity day selection method. Three historical days with the smallest similarity error are selected as the similar days to be predicted. The average value of photovoltaic output power of similar days is taken as the prediction value of photovoltaic output power. The key of this prediction method is to determine the combination weight coefficient of each base point when selecting the similar day. The subjective weight coefficient and objective entropy weight of each base point are determined first, and then the least discriminant information principle is used to fuse the above two weight coefficients. The simulation results of combined weight coefficient and Matlab show that the photovoltaic output power prediction method based on the combination weight method can select the similar days with high similarity degree and improve the precision of photovoltaic output power prediction. Finally, a smooth output control strategy based on PV power prediction is proposed. Based on the prediction of photovoltaic output power, the change trend of the charge state of the battery at the next moment is estimated by using the predicted value of the power at the next moment, and the charge and discharge quantity of the energy storage cell is adjusted. This method can realize the smooth control of the output power of the system and keep the SOC of the energy storage battery system stable in the normal range. The effectiveness of the control strategy is verified by Matlab simulation.
【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】：TM61

【参考文献】