Development Of Remote Sensing Detection And Control System For Automobile Exhaust Emission

Automobile exhaust pollutant emission has become the main source of urban air pollution.The traditional automobile environmental protection testing method is full of pressure in the face of the increasing number of cars.In the future,automobile exhaust emission testing will develop towards the direction of real-time online testing with high efficiency,high accuracy and big data collection.Remote sensing detection technology can meet the development needs and has become the main research direction of automobile exhaust emission detection.At present,there are still bottlenecks in the analysis of spectral influencing factors and vehicles screening model algorithm of telemetry.This paper aims to design remote sensing test system through existing spectral absorption technology,conduct in-depth research on pollutant concentration collection algorithm and high-emission vehicles screening algorithm,and optimize the existing vehicle exhaust remote sensing detection and control system.Firstly,based on the existing spectral absorption technology and remote sensing detection application technology,technical research was carried out to determine the scheme of using TDLAS technology to detect CO₂ and CO gas concentration in the telemetry system,using DOAS technology to detect HC and NO gas concentration,and using direct absorption spectrum technology to detect impermeable light.Based on the lambert-beer law,the principle of laser detection is systematically analyzed to determine the relationship between laser signal and target concentration.Inquire the spectral database,determine the gas detection wavelength according to the wavelength measurement conditions,and customize the spectral sensor.Infineon single chip microcomputer?SAK-XE164FN-40F80L?was selected to design the telemetry system control board,and the AD amplifier circuit to complete the hardware design of the control system.Secondly,on the basis of the hardware system design of TDLAS,DOAS,and opacity test system,in TDLAS test,determined by standards of different concentration of CO₂ and CO gas concentrations and second harmonic relations,fitting linear function,test temperature,air chamber length and the laser echo energy influence on measuring result,determines the temperature compensation algorithm and second harmonic normalization algorithm.In DOAS test,differential absorption cross sections were measured by using different concentrations of NO and 1,3-butadiene standard gas,and the differential absorption algorithm was improved by background spectrum.The relationship between temperature and spectral signal was tested to determine the temperature compensation algorithm.In the impermeable luminosity test,AD signals of filters with different N values were measured and fitted by cubic spline interpolation function.The engine combustion equation is constructed,and the true values of pollutant concentrations in exhaust emissions are deduced by the relative volume ratio.Finally,high emission vehicles screening model was constructed based on vehicle driving cycle,through the simple transient gasoline-fuelled vehicles experiment data,the analysis of the pollutant concentration and the change of the relationship between vehicle driving cycle,determine the condition of impact on the results of the telemetry,self-organizing map neural network is adopted to improve the condition of online matching,with the speed,acceleration,than power,concentration of CO concentration and the concentration of HC and NO six characteristic parameters of high emission vehicles screening model was constructed.In simulation tests found that neural network filter model error within 10%,the real vehicle experiment,using simple transient gasoline-fuelled vehicles to verify the exactness of the telemetering system of pollutants concentration collection,CO₂,CO,NO concentration detection of maximum error is 7.49%,through neural network filter model with fixed threshold model contrast test,show that neural network filter model accuracy can be increased by 7.5%.