摘要
本文设计并实现了一种基于长短期记忆(LSTM)神经网络的长周期负载预测系统,通过自动数据采集、模型训练优化、负载预测和可视化展示,实现了多时间跨度的高精度预测。系统采用高性能GPU、企业级存储和高带宽网络设备,支持高效训练和实时预测。通过功能和性能测试验证了系统的稳定性和可靠性,表明其在高并发环境下具有良好的扩展性和应用潜力。
This paper presents a long-term load prediction system using an LSTM neural network.Through automated data collection,model training,load prediction,and visualization,the system achieves high-precision forecasts across multiple time spans.It employs high-performance GPUs,enterprise storage,and high-bandwidth networking to enable efficient training and real-time prediction.The stability and reliability of the system have been verified through functional and performance testing,indicating its good scalability and application potential in high concurrency environments.
作者
吕家欣
LV Jiaxin(Xiamen University of Technology,Xiamen Fujian 361000)
出处
《软件》
2025年第1期94-96,共3页
Software
