摘要
基于光伏阵列正常组件的稳定运行状态中的数据,引入深度学习中的长短期记忆网络LSTM,以电流测点作为特征,建立光伏阵列电流测点的时序预测模型。建模过程中,调整各个参数优化模型来保证预测的准确性。基于预测结果建立报警规则,实现光伏阵列有效及时的异常预警。最后以某光伏电站的历史数据为例,进行了算法有效性与可行性的验证。实验表明,本文的异常检测与预警方案可以实现对光伏阵列电流数据异常的准确判断与预警,对实际工程应用中相关设备的运维检修具有很大的参考价值。
Based on the data in the stable operation state of the normal modules of photovoltaic array,the long and short-term memory network LSTM(Long Short-Term Memory)in deep learning was introduced,and the timing pre⁃diction model of photovoltaic array current measurement points was established with the characteristics of current measurement points.During the modeling process,the individual parameter optimization model was adjusted to en⁃sure the accuracy of the prediction.The alarm rules were established based on the prediction results to realize the effective and timely abnormal early warning of the photovoltaic array.Finally,taking the historical data of a photo⁃voltaic power station as an example,the validity and feasibility of the algorithm were verified.The experiments showed that the anomaly detection and early warning scheme in this paper could realize the accurate judgment and early warning of the abnormal photovoltaic array current data,and had great reference value for the operation and maintenance of related equipment in practical engineering applications.
作者
姜汉国
柴东元
戴恩哲
胡玉
JIANG Hanguo;CHAI Dongyuan;DAI Enzhe;HU Yu(National Energy Group Ningxia Electric Power Co.,Ltd.,Yinchuan 750061,China;Shangtejie Electric Power Technology Co.,Ltd.,Hefei 230088,China)
出处
《粘接》
CAS
2024年第8期146-149,共4页
Adhesion
关键词
变电站
光伏阵列
风险预控
深度学习
substation
photovoltaic array
risk early warning
deep learning