论文部分内容阅读
目前我国正处于大力发展可再生能源的阶段,尤以光伏发电系统为例,其建设规模及安装容量不断扩大。然而光伏发电系统输出功率的不确定性会对电网造成一定的影响,为减轻其对电网系统的损害,需要进行准确的光伏出力预测,这就对光伏发电的历史数据的完整性与准确性提出了更高的要求。本文结合历史发电量数据,通过对以往数据中的不良数据进行识别及剔除,并利用大量的数据样本Elman神经网络进行训练并对伪数据剔除进行数据重构。实验结果表明,本文建立的Elman反馈递归模型能够很好的剔除光伏发电功率预测的不良数据并进行详细补全,对减小功率预测的误差起到重要作用。
At present, China is in the stage of energetically developing renewable energy. Especially in the case of photovoltaic power generation system, its construction scale and installation capacity are constantly expanding. However, the uncertainty of the output power of photovoltaic power generation system will affect the power grid to a certain extent. In order to reduce the damage to the power grid system, the accurate prediction of PV output power needs to be made. This puts forward the completeness and accuracy of the historical data of photovoltaic power generation A higher requirement. Combining historical generation data, we identify and remove the bad data in the past data, and use a large number of data samples Elman neural network to train and reconstruct the data by eliminating the false data. The experimental results show that the Elman feedback recursive model established in this paper can eliminate bad data of PV power prediction well and make up in detail, which plays an important role in reducing the error of power prediction.