基于深度学习的配变停电电量损失预测

停电电量损失预测可为电网调度及规划提供参考,有利于为用户提供可靠供电服务。针对当前配变停电过程中的电量损失问题,先基于模糊C均值聚类算法实现对配变负荷曲线的分类处理及精细化分析,挖掘配变负荷数据规律;在此基础上,运用皮尔逊相关系数算法提取选择输入特征,构建基于门控循环单元神经网络的预测模型,从而得到停电时间负荷值,进而分析预测负荷曲线得到损失电量;最后,基于停电管理工作分析,实现基于粒子群优化的台区用电行为停电优化问题求解。算例测试验证了所提方法的正确性和有效性。

Power Outage Loss Prediction of Distribution Transformer Based on Deep Learning

Power outage loss prediction can provide reference for power grid dispatching and planning as well as reliable power supply services for users. Based on the fuzzy C-means clustering algorithm, the load curve of distribution transformer can be classified and analyzed finely. The rules of load data of distribution transformer are mined. Then, the Pearson correlation coefficient algorithm is used to extract the selected input characteristics, and a prediction model based gated cyclic unit neural network is constructed to obtain the load value during power outage. The loss of electricity is obtained by analyzing the load curve. Finally, based on the analysis of power outage management, the power outage optimization problem is solved based on particle swarm optimization. The validity of the proposed method and model are verified by example tests.