为了提高话务量的预测精度,针对传统人工神经网络存在的参数优化问题,提出一种蚁群算法优化神经网络的话务量预测模型(ACO-BPNN)。首先考虑话务量数据之间的时间相关性,重构话务量的学习样本,然后将样本输入到神经网络进行训练,并通过蚁群算法不断优化神经网络连接权值和阈值,使实际输出与期望输出之间的误差最小,建立话务量预测模型,最后将预测模型用于某市移动通信网络的忙时话务量预测中。实验结果表明,本文模型提高了话务量的预测精度,尤其是多步预测效果明显优于对比模型,具有良好的实用性。 In order to improve the prediction accuracy of traffic volume, aiming at the problem of parameter optimization existing in traditional artificial neural network, an ACO-BPNN model based on ant colony optimization is proposed. First consider the time correlation between traffic data, reconstruct the learning samples of traffic, then input the samples to the neural network for training, and optimize the neural network connection weights and thresholds continuously by the ant colony algorithm so that the actual output And the expected output of the smallest error between the establishment of traffic prediction model, and finally the prediction model is used in a city’s mobile communications network busy traffic prediction. Experimental results show that this model improves the prediction accuracy of traffic volume, especially the multi-step prediction is better than the contrast model, which has good practicability.