摘要
为解决通信电源故障导致的供电中断问题,提升电力调度灵活性,提出融合深度学习的故障预测与自适应调度方法。采集通信电源运行数据,经预处理后构建卷积神经网络-长短时记忆(Convolutional Neural Network-Long Short-Term Memory,CNN-LSTM)故障预测模型,结合AdamW与Dropout算法优化模型。基于预测结果设计多目标自适应调度机制,以强化学习生成动态策略,实现故障前调度调整。实验表明,所提方法的日均电力损耗、每天调度操作成本及故障中断时长均低于传统方法,供电可靠性高于传统方法。
To address power supply interruptions caused by communication power failures and enhance the flexibility of power dispatching,a fault prediction and adaptive dispatching method integrating deep learning is proposed.Communication power operation data is collected and preprocessed to construct a Convolutional Neural Network-Long Short-Term Memory(CNN-LSTM)fault prediction model,which is optimized using the AdamW and Dropout algorithms.Based on the prediction results,a multi-objective adaptive dispatching mechanism is designed,employing reinforcement learning to generate dynamic strategies for pre-fault dispatching adjustments.Experiments demonstrate that the proposed method achieves lower daily power losses,daily dispatching operation costs,and fault interruption durations compared to traditional approaches,while delivering higher power supply reliability.
作者
闫超
YAN Chao(Inner Mongolia Power(Group),Inner Mongolia EHV Power Supply Branch Company,Hohhot 010080,China)
出处
《通信电源技术》
2025年第21期95-97,共3页
Telecom Power Technology
关键词
深度学习
通信电源
故障预测
自适应电力调度
deep learning
communication power supply
fault prediction
adaptive power dispatching
