Abstract: Aiming at the problems of low accuracy and unstable data of existing prediction models in the process of intelligent traffic vehicle flow prediction, this paper proposes a short-period traffic flow prediction method based on the combination of convolutional neural network (CNN) and time series prediction model auto regressive integrated moving average (ARIMA). In this paper, the historical traffic flow data of Wangxi road intersection in Wuhu city within a week is taken as the sample to predict the traffic vehicle flow in the next time period, and the obtained data are normalized and simulated. The simulation results show that the scheme can be applied to traffic control, and the scheme can effectively maintain a high accuracy without being affected by the size of the data set, and the prediction results can provide a better solution to the problem of vehicle congestion in advance.

Key words: Intelligent transportation; Traffic intersections; Short cycle; Traffic vehicle flow predic- tion; CNN-ARIMA; Prediction method

摘要： 针对现有预测模型在智慧交通车流量预测过程中准确度较低、数据不平稳的问题，文 章提出了一种基于卷积神经网络（CNN）和时间序列预测模型自回归差分移动平均（ARIMA） 相结合的短周期交通车流量预测方法。文章以芜湖市汪溪路路口一周内的历史车流量数据为 样本去预测下一时间周期的交通车流量，将得到的数据经过归一化处理后模拟仿真实验，仿 真结果表明，该方案能够运用于交通管控中，且能有效保持较高精确率而不受数据集大小的 影响，其预测结果可以为车辆拥堵问题提供较好的提前应对方案。

关键词: 智慧交通；交通路口；短周期；交通车流量；CNN-ARIMA；预测方法