Research And Application Of Multisensor Information Fusion Method In Intelligent Traffic Control

Multisensor information fusion technology is used to comprehensive analysis information derived from several different uncertain and multi-level sensor.It is a processing procedure in order to get more comprehensive,more accurate and more complete information than a single information source.Information fusion is a kind of automation integrated information processing technology,using each sensor data complementary information and computer,the operation,the different temporal and spatial data to get a more accurate description of the real environment,to achieve more accurate recognition and decision function,improve the information quality.Intelligent traffic control system using intelligent information processing method,research the vehicle flow rate detected by the sensor,calculate the traffic signal lamp signal control strategy,which on each phase intersection traffic flow control.How to obtain the traffic information of the traffic information in real time by using the information fusion technology is the key research object of this paper.This paper according to the actual needs of the analysis of the information fusion application in intelligent traffic control,aiming at real-time,distribution and the complex traffic integration environment,established in the data layer using weighted least squares algorithm to process the sensor data,decision-making layer using evolutionary rules improved BP neural network for traffic to make prediction information fusion model.In order to overcome the shortcoming of BP neural network for initial weights random selection and easy to fall into local minimal solution of the problem,this paper studies the method of evolutionary programming to optimize the initial weights of BP neural network to solve the problem.Evolutionary rule and algorithm of neural network with random initial weights are encoded and the mutation operator with appropriate choice of strategy to find the optimal adaptation of individual and reduces the error produced in the process of neural network training experience,in order to improve the efficiency of neural network training and accelerate the convergence rate of the network.At last,a signal lamp control system is designed based on the predicted traffic flow.The system of sensor data for processing and analysis,forecast next intersection period traffic and thus the distribution of traffic signal lamp,administrators can understand the real-time signal lamp,sensors and traffic information and on the signal model and the sensor control,using the VISSIM software at the junction of three-dimensional modeling to intuitive way to display intersection traffic state.