论文部分内容阅读
传统的全光信息网络业务量预测方法以特定业务量反推全光信息网络预测结果,需花费较多的时间与资源去平衡变化中的业务量,性能不佳。根据上述描述,提出一种全光信息网络业务量预测新方法。方法使用马尔科夫链预测技术,将全光信息网络的传输通道分为开启和关闭两种状态,在卜氏分配给出的实时业务量分配模型中寻找到两种状态之间的通讯空白区域,利用概率论推断出该区域的存在时间,实现对全光信息网络实时业务量的合理分配。分配后的实时业务量经由套袋算法对BP神经网络的训练次序进行调整,给出全光信息网络业务量的单相预测结果并进行集成,获取高精度预测结果。实验结果表明,所提方法预测用时短、资源耗费量少。
The traditional all-optical information network traffic prediction method uses the specific traffic volume to predict all-optical information network prediction results. It takes more time and resources to balance the changing traffic and has poor performance. According to the above description, a new method for all-optical information network traffic prediction is proposed. The method uses Markov chain prediction technology to divide the transmission channel of OPN into two states, ON and OFF. In the real-time traffic distribution model given by Buh distribution, a blank area of communication between two states is found , The use of probability theory to infer the existence of the region to achieve real-time all-optical information network traffic distribution. The distribution of real-time traffic through bagging algorithm to adjust the training sequence of BP neural network, all-optical information network traffic single-phase prediction results are given and integrated to obtain high-precision prediction results. Experimental results show that the proposed method has short time-consuming and less resource consumption.