论文部分内容阅读
以实际采集的交通流量序列作为研究对象,分别应用互信息法和虚假邻点法确定其延迟时间和最佳嵌入维数,完成交通流量序列的相空间重构.通过计算交通流量序列的饱和关联维数和最大Lyapunov指数判定其混沌特性.以最小均方(LMS)算法为基础,构建了一种基于Davidon-Fletcher-Powell方法的二阶Volterra模型(DFPSOVF),其应用了一种可随输入信号变化而实时变化的基于后验误差假设的可变收敛因子技术.DFPSOVF模型避免了在Volterra模型中采用LMS自适应算法调整系数时参数选择不当引起的问题.将DFPSOVF模型应用于具有混沌特性的短时交通流量预测,结果表明:当模型记忆长度与交通流量序列的嵌入维数选择一致时,模型的预测精度较高,可以满足交通诱导和交通控制的需要,为智能交通控制提供了新方法、新思路及工程应用参考.
Taking the actual traffic flow sequence as the research object, the delay time and the optimal embedding dimension are respectively determined by the mutual information method and the pseudo-neighbors method, and the phase space reconstruction of the traffic flow sequence is completed. By calculating the saturation correlation of the traffic flow sequence Dimension and maximum Lyapunov exponent.A second order Volterra model based on Davidon-Fletcher-Powell method (DFPSOVF) is constructed based on Least Mean Square (LMS) algorithm, Signal changes and changes in real time based on the posterior error assumptions of the variable convergence factor technology.DFPSOVF model avoids the Volterra model using LMS adaptive algorithm to adjust the coefficient of the problem caused by inappropriate choice of parameters.The DFPSOVF model is applied to a chaotic The results show that when the model memory length is consistent with the embedding dimension of traffic flow, the prediction accuracy of the model is high, which can meet the needs of traffic guidance and traffic control, and provides a new method for intelligent traffic control , New ideas and engineering application reference.