| With the development of science and technology and the growth of the national economy,city modernization and automobile industry have progressed rapidly.Besides,the number of cars increases sharply,and the per capita quantity of cars grows year by year,resulting in frequently-occurred road traffic accidents.Most of the accidents are due to exceeding speed limits.Overspeed is the most common traffic violations,and the main factor to cause traffic accidents.In order to ensure that vehicle drivers are not overspeed,speed limits need to be clearly marked at the corresponding sections and vehicle monitors are also required.Therefore,a complete set of radar velocity measurement system is needed to monitor the vehicle speed.Traditional traffic radar systems can't change the speed limit standard according to real-time environmental modification.In addition,speed measuring radars are not accurate enough.They are difficult to achieve wide-range coverage,and are slow on data processing.They could also be affected by other signals,and are difficult to recognize vehicle license plate after video capture.Sometimes they even can't finish the identification,and require manual interventions,which not only increases the workload of law enforcement personnel,but also delays serious violations reporting.Therefore,it is necessary to develop a set of intelligent,high precision,high speed,economical and practical traffic radar speed measurement system,which can provide effective methods for traffic management control to reduce the traffic accident rate,improve road use efficiency and increase social benefits.Based on digital image processing technology and narrow beam radar working principle,this dissertation combines two neural network models to achieve the design of framework,hardware and software of the intelligent radar system.This dissertation also provides further study on the license plate recognition algorithm,control model of the intelligent safety speed limiting,and narrow wave radar core model,which could help to solve the license plate recognition rate problem,velocity measurement accuracy problem,and adaptive speed limiting problem in the radar system.Firstly,this dissertation introduces license plate information processing methods,and explains the reason of the low recognition rate of traditional recognition algorithms,according to the characteristic of Chinese license plate,i.e.mixed characters with digital,Chinese and character.Based on traditional ART1(Adaptive resonance theory)neural network,this dissertation presents a three channel parallel ART1 network classification algorithm.This algorithm adopts regional management and could realize rapid and accurate character recognition after the basic license plate acquisition,location,and image preprocessing.Secondly,this dissertation discusses the accuracy requirement of existing radar,and the difference between measurement results and real speed due to the effect of actual measurements,and influence of radar itself.Then system engineering analytical methods are adopted to analyze the major reason of the police vehicle velocity radar speed measurement error,find out the main cause.After that,this dissertation applies automatic gain control unit to achieve the gain control of the radar signal in radar signal processing,which solves the measurement error problem caused by "multiplier",and the angle difference between vehicles and radar in practical works,resulting in the reduction measurement error.Based on the optimization design,this dissertation also analyzes the key components theory of flat plate narrow beam radar,and designs a set of intelligent flat narrow beam radar according to the working environment and installation guidance.The radar utilizes a single oscillator transmitter,which is based on VCO piezoelectric oscillator,to modulate triangular wave into radio-frequency signal.The signals received by antenna are handled to become two signals corresponding to I and Q channel,which are used to record information frequency and signal phase,respectively.After adaptive filtering,the returned signal are sent to special signal processing module to achieve the analysis of vehicle speed,distance,and size.This radar has been recognized as a new key product by Ministry of Science and Technology,Commerce Department,Ministry of Environmental Protection,National Technical Supervision and Inspection Bureau in 2013.At last,this dissertation realizes the real time rainfall monitoring by utilizing the R-T model-based pluviometer,which has the dynamic change technology of rainfall collection period.BP neural network is chosen to build the intelligent speed control model,which uses environment parameters,road condition evaluations and the number of vehicles as the input,and real-time traffic safety speed as the output.By combining narrow flat wave radar hardware,and intelligent license plate recognition technology,a set of intelligent traffic radar speed measurement system is established,and a high-speed illegal evidence collection system is developed based on computer platform.The intelligent radar speed measurement system has been applied to actual traffic management.After long-time application,the system has been proved to be stable,convenient,and high performance.It has contributed to reduce the death rate of traffic accident,and is widely recognized. |
PDF Full Text Request