陈卫军, 金显华. 一种最大共轭梯度连续泛函的网络峰值预测[J]. 信阳师范学院学报(自然科学版), 2015, 28(2): 275-278. DOI: 10.3969/j.issn.1003-0972.2015.02.029
引用本文: 陈卫军, 金显华. 一种最大共轭梯度连续泛函的网络峰值预测[J]. 信阳师范学院学报(自然科学版), 2015, 28(2): 275-278. DOI: 10.3969/j.issn.1003-0972.2015.02.029
Chen Weijun , Jin Xianhua . Network Peak Prediction Based on Maximum Conjugate Gradient Continuous Functional[J]. Journal of Xinyang Normal University (Natural Science Edition), 2015, 28(2): 275-278. DOI: 10.3969/j.issn.1003-0972.2015.02.029
Citation: Chen Weijun , Jin Xianhua . Network Peak Prediction Based on Maximum Conjugate Gradient Continuous Functional[J]. Journal of Xinyang Normal University (Natural Science Edition), 2015, 28(2): 275-278. DOI: 10.3969/j.issn.1003-0972.2015.02.029

一种最大共轭梯度连续泛函的网络峰值预测

Network Peak Prediction Based on Maximum Conjugate Gradient Continuous Functional

  • 摘要: 提出一种基于最大共轭梯度连续泛函的网络峰值预测算法和模型, 分析网络峰值预测影响因素,建立一个包含网络流量、 网络峰值范围和信号强度的 SVM 模型. 采用 SVM 模型的主成分分析方法实现对网络峰值的PCA 估计系统设计. 通过最大共轭梯度连续泛函, 在奇异半正定性双周期性复分析下, 实现对网络峰值的预测, 考察网络流量的波动以及网络信号对网络峰值影响贡献程度, 对网络峰值特征进行状态信息融合处理, 减少预测误差. 实验结果表明, 该算法对网络峰值的预测精度较高, 预测误差控制在 1. 5%以内, 性能优越.

     

    Abstract: A conjugate gradient based on continuous network peak prediction algorithm and model of functional analysis were presented. The main factors affecting the network peak forecast were analyzed. The SVM model including  network traffic, network peak range and network signal strength was formulated. The principal component analysis method was adopted to realize the network peak PCA to estimate the system design. Through the largest continuous functional conjugate gradient, the peak of the internet  was predicted under the singular positive semidefinite double periodic complex analysis. The degree of contribution of the network traffic and network signal peak to the network was investigated, the network characteristic of peak was considered by using state information fusion processing to reduce the prediction error. The experimental results showed that the algorithm of network peak prediction accuracy was higher, the prediction error was controlled within 1.5%,  and the performance of the algorithm was superior

     

/

返回文章
返回