基于IFA优化RBF神经网络的短时交通流预测模型

针对短时交通流不确定性极强引起的预测结果精度低的问题,提出一种改进萤火虫算法(IFA)优化RBF神经网络的短时交通流预测模型(IFA-RBF).该模型通过引入线性递减惯性权重和混沌机制,来改进FA后期存在的易陷入局部极值和种群多样性匮乏的不足,利用IFA优化RBF神经网络的连接权重和基函数宽度,以提升RBF神经网络的短时交通流预测精度.实验结果表明,与Elman、BP、RBF和FA-RBF模型相比,构建的短时交通流预测模型(IFA-RBF)具有更高的预测精度,预测值与实际值拟合度较高.

Short-term traffic flow prediction model based on IFA optimized RBF neural network

Aiming at the problem of low accuracy of prediction results caused by the extremely strong uncertainty of short-term traffic flow data, a short-term traffic flow prediction model(IFA-RBF) based on the improved firefly algorithm (IFA) optimized RBF neural network is proposed. By introducing linear decreasing inertial weight and chaos mechanism, the model improves the shortcoming of being trapped into local extremes and lack of population diversity in the later stage of FA. IFA is used to optimize the connection weight and basis function width of the RBF neural network is used to improve the prediction accuracy of the RBF network. Experimental results show that compared with Elman, BP, RBF and FA-RBF models, the constructed model has higher prediction accuracy, and the predicted value has higher fitting degree with the real value.