神经网络在无线网络流量预测中的应用作者：雷晓明来源：《现代电子技术》2017年第02期 摘 要： 无线互联网的流量数据非常分离且极其不稳定，混沌理论在其身上体现得特别明显，因此对无线网络流量进行预测具有一定难度。该文使用BP神经网络建立预测模型，在常规神经网络系统进行训练之前，需要对系统内部各个层次之间的连接权值以及阈值范围实行初始化操作，但是此操作将会影响神经网络最终收敛速度，有可能造成最终结果为非最优解，使得流量预测结果不是很理想。因此这里使用布谷鸟搜索优化方式对神经网络系统内各层之间链接值与阈值进行初始化操作，提高系统预测精度。该文使用遗传优化神经网络算法和粒子优化神经网络算法建立同样的预测模型，并与该文研究的预测模型进行对比。实例分析结果表明，初期预测结果精度较高，与实际值比较吻合，但测试数据越靠后，预测值越不稳定，这主要是累计误差造成的。但总的来说，该文使用的布谷鸟优化BP神经网络预测模型的预测性能要优于由遗传算法和粒子算法优化的BP总线上的音频设备神经网络。 关键词：针式吸盘 无线网络； 流量预测； BP神经网络； 布谷鸟算法
中图分类号： TN915⁃34地震云与地震预测； TP393 文献标识码： A 超高压软管文章编号： 1004⁃373X（2017）02⁃0111⁃03
Abstract： The wireless Internet traffic data is dispersive and extremely unstable， on which the chaos theory is reflected obviously， so it is difficult to predict the wireless network traffic. The BP neural network is used to establish the prediction model. Before training the conventional neural network system， the initialization operation needs to be carried out for the connection weights and threshold range among each layer inside the system， but this operation will affect the final convergence speed of the neural network，灰板纸 which may cause the final result not to be the optimal solution， and make the traffic prediction result unsatisfied. Therefore， the cuckoo search optimization method is used to conduct the initialization operation for the link value and threshold among each layer inside the neural network system to improve the prediction accuracy of the system. The identical prediction models were established with the genetic neural network optimization algorithm and particle swarm optimization neural network algorithm， and compared with the prediction model established with the method figured out in this paper.
The results of instance analysis indicate that the accuracy of the initial prediction result is high and identical with the actual value， but the more the test data leans tack， the less stable the prediction value becomes due to the cumulative error. In general氧气止回阀， the prediction performance of the BP neural network prediction model optimized with cuckoo algorithm is better than that of the BP neural network prediction model optimized with the genetic algorithm and particle swarm algorithm.
