摘要
光伏发电受天气因素的影响,造成供电的稳定性降低,使得光伏功率预测准确性不高。为保证供电服务质量,提出一种基于改进LSTM-TCN的预测模型。利用灰色关联度算法选取光伏功率影响因素,通过相似度计算,获取分布式光伏功率值,组成训练样本数据。将LSTM与TCN结合,构建改进LSTM-TCN预测模型,并通过模型训练,得到成熟的预测模型。实验结果表明,所研究模型预测结果的平均平方差和平均绝对误差率分别为0.562 6 kW和2.653 2%,远低于另外两种对比方法,表明所提方法的功率预测准确性高。
Due to the impact of weather factors on photovoltaic power generation,the stability of power supply is reduced,making the accuracy of photovoltaic power prediction not high.To ensure the quality of power supply service,a prediction model based on improved LSTM-TCN is proposed.The gray correlation degree algorithm is used to select the influencing factors of photovoltaic power,and the distributed photovoltaic power value is obtained through similarity calculation to form the training sample data.Combining LSTM with TCN,an improved LSTM-TCN prediction model is constructed,and a mature prediction model is obtained through model training.The experimental results show that the average square error and average absolute error rates of the prediction results of the studied model are 0.5626 kW and 2.6532%,which are significantly lower than the other two comparison methods,indicating that the proposed method has high accuracy in power prediction.
作者
陈建军
李良杰
卢泽汉
CHEN Jianjun;LI Liangjie;LU Zehan(Tangshan Power Supply Company,State Grid Jibei Electric Power Co.,Ltd.,Tangshan 432500,China)
出处
《电子设计工程》
2024年第16期187-190,195,共5页
Electronic Design Engineering
基金
唐山供电公司科学技术项目(SGTYHT/21-JS-223)。
