论文部分内容阅读
为了实现城市快速路的入口匝道智能动态控制,通过建立入口匝道数学模型,并应用具有递归环节的动态模糊神经网络于匝道控制系统中。在模糊神经网络第二层中加入内部反馈连接,使控制系统更好地响应复杂多变的交通状况,解决了以往静态网络无法处理的暂态问题。控制入口匝道的动态模糊神经网络使用遗传算法与反向传播BP算法相结合来训练,遗传算法的宏观搜索能力及鲁棒性强等优点有效地避免了神经网络算法易陷入局部极小及震荡效应等缺点。通过仿真结果,验证了基于动态模糊神经网络的控制算法相对于经典的ALINEA入口匝道控制算法具有改善,能够更好地保证城市快速路的通行效率。
In order to realize intelligent control of entrance ramp of urban expressway, a mathematical model of entrance ramp is established, and a dynamic fuzzy neural network with recursion is applied to the ramp control system. The internal feedback connection is added in the second layer of the fuzzy neural network to make the control system respond to the complex and changeable traffic conditions better and solve the transient problem that the static network can not handle in the past. The dynamic fuzzy neural network which controls the entrance ramp uses the combination of genetic algorithm and backpropagation BP algorithm to train, the macro search ability of genetic algorithm and strong robustness effectively avoid the neural network algorithm easy to fall into local minima and shock effect Other shortcomings. The simulation results show that the control algorithm based on the dynamic fuzzy neural network has better performance than the classic ALINEA entry ramp control algorithm and can better ensure the traffic efficiency of the urban expressway.