交通流量预测是城市智能交通系统的重要研究内容之一,是缓解城市拥堵、实现智能交通管理和建设智慧城市的前提,基于短时交通流量的复杂性及非线性等特点,提出耦合AF-SVR的短时交通流量预测模型.模型结合了鱼群算法较好的并行搜索性能和支持向量回归机较好的非线性拟合能力,利用该模型对短时交通流量数据进行仿真实验,结果表明:模型较BP神经网络预测模型具有较高的预测精度,是短时交通流预测的一种有效方法. Traffic flow prediction is one of the important research contents of urban intelligent transportation system. It is the premise of alleviating urban congestion, realizing intelligent traffic management and building a smart city. Based on the characteristics of complexity and nonlinearity of short-term traffic flow, The model combines the better parallel search performance of fish swarm algorithm with the better nonlinear fitting ability of support vector regression machine.The simulation results of the short-term traffic flow data show that: The model has higher prediction accuracy than the BP neural network prediction model and is an effective method for short-term traffic flow prediction.