摘要
交通拥堵汽车流量的准确预测,能够有效改善现有的道路网络的通行能力。对车流量的预测,需要利用模糊推理系统将交通流分为不同的模糊集,给出模糊隶属度函数,完成交通拥堵汽车流量的预测。传统方法先选出相关性较高的道路断面,得到路网空间关系归一化约束矩阵,但忽略了给出交通隶属度函数,导致预测精度偏低。提出基于模糊神经网络的拥堵汽车流量预测方法。将神经网络和自适应卡尔曼滤波模型相结合,组建拥堵汽车流量预测模型,将任意时刻的交通流预测期望值作为模型的输入,将实际预测误差作为模型的输出,利用模糊推理系统将交通流分为不同的模糊集,给出模糊隶属度函数,将小波函数作为模糊推理系统的输入,利用神经网络实现模糊推理,借鉴遗传理论优化模糊神经网络,完成对拥堵汽车流量的预测。实验结果表明,所提方法预测精度较高,具有较高的预测准确性。
A prediction method of vehicle flow in traffic jam based on fuzzy neural network is proposed. The neural network is combined with adaptive Kalman filtering model to build prediction model of vehicle flow in traffic jam and the value of predictive expectation at any time is taken as input of the model, then, the actual predictive encoding is regarded as output of the model and fuzzy inference system is used to divide traffic flow into different fuzzy sets. Moreover, a fuzzy membership function is given and the wavelet function is taken as input of fuzzy inference system. Thus, the neural network is used to realize fuzzy inference. Finally, the genetic optimization theory of fuzzy neural net- work is used to complete vehicle flow in traffic jam. Simulation result shows that the proposed method has high pre- diction accuracy.
作者
程山英
CHENG Shan- ying(College of Math and Computer of the Jiangxi Science ＆ Technology Normal University, Nanchang, Jiangxi 330038, Chin)
出处
《计算机仿真》
北大核心
2017年第10期123-126,共4页
Computer Simulation
基金
江西省科技计划指导性项目(2015ZBAB201007)
江西科技师范大学校级科研重点项目(2016XJZD006)
江西省高校人文社会科学研究项目(TQ1505)
关键词
模糊神经网络
短项交通流量
交通流量预测
Fuzzy neural network
Short term traffic flow
Traffic flow forecasting
