计算机工程与应用
計算機工程與應用
계산궤공정여응용
COMPUTER ENGINEERING AND APPLICATIONS
2013年
8期
100-104
,共5页
映射矩阵%预测模型%网络流量%用户行为
映射矩陣%預測模型%網絡流量%用戶行為
영사구진%예측모형%망락류량%용호행위
mapping matrix%prediction model%network traffic%user behavior
通过对用户行为分析,发现IP数据流具有平稳性、自相关性等特点,提出基于映射矩阵流量预测模型,并与线性模型AR、ARIMA和非线性基于反馈神经网络BP模型、Elman神经网络作对比,试验结果证明,映射矩阵模型,比现有模型具有预测精度高、收敛快等特点.
通過對用戶行為分析,髮現IP數據流具有平穩性、自相關性等特點,提齣基于映射矩陣流量預測模型,併與線性模型AR、ARIMA和非線性基于反饋神經網絡BP模型、Elman神經網絡作對比,試驗結果證明,映射矩陣模型,比現有模型具有預測精度高、收斂快等特點.
통과대용호행위분석,발현IP수거류구유평은성、자상관성등특점,제출기우영사구진류량예측모형,병여선성모형AR、ARIMA화비선성기우반궤신경망락BP모형、Elman신경망락작대비,시험결과증명,영사구진모형,비현유모형구유예측정도고、수렴쾌등특점.
By analyzing user behavior, this paper finds that the IP data stream has smooth and self-correlation characteristics, pro-poses a prediction model based on the mapping matrix flow in comparison with the linear AR, RIMA models and nonlinear BP model based on feedback neural network, Elman neural network. The test results demonstrate that the mapping matrix model has high prediction accuracy and fast convergence characteristics compared with existing models.
