摘要
为了有效地提高天基信息网中的资源利用效率,提出了一种基于改进GRU(gated recurrent unit)算法的天基信息网资源预测模型.首先,提出递阶式三级架构的资源预测框架来解决天基环境长时延的问题;然后,采用Adam优化器优化GRU网络的学习速率;最后,引入Dropout技术解决网络中存在的过拟合问题.实验仿真了不同预测模型下对各种天基资源的预测,同时对比不同优化器作用下GRU模型的预测准确率,结果表明,基于改进GRU网络的资源预测模型具有更好的性能.
In orde to improve the resource utilization of space-based information metwork efficiency,a resource prediction model of space-based information network was presented based on the improved GRU(gated recurrent unit)algorithm.Firstly,a hierarchical three-level resource prediction framework was proposed to solve the problem of long delay in space-based environment.Then,Adam optimizer was used to optimize the learning rate of GRU network.Finally,Dropout technology was introduced to solve the over-fitting problem in the network.The experiments simulated the prediction of various space-based resources under different prediction models,and compared the prediction accuracy of GRU model under different optimizers.The results show that the resource prediction model based on improved GRU network has better performance.
作者
耿蓉
吴亚倩
肖倩倩
徐赛
GENG Rong;WU Ya-qian;XIAO Qian-qian;XU Sai(School of Computer Science&Engineering,Northeastern University,Shenyang 110169,China)
出处
《东北大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2023年第3期305-314,共10页
Journal of Northeastern University(Natural Science)
基金
中央高校基本科研业务费专项资金资助项目(N2116015,N2116020)
辽宁省医工交叉基金资助项目(2021-YGJC-24)
国家自然科学基金资助项目(62071134)。
关键词
天基信息网
资源预测
GRU网络
Adam优化器
Dropout技术
space-based information network
resource prediction
GRU(gated recurrent unit)network
Adam optimizer
Dropout technology