CN113505160A - Motor vehicle exhaust emission estimation method based on convolution deep neural network - Google Patents

Motor vehicle exhaust emission estimation method based on convolution deep neural network Download PDF

Info

Publication number
CN113505160A
CN113505160A CN202110822850.XA CN202110822850A CN113505160A CN 113505160 A CN113505160 A CN 113505160A CN 202110822850 A CN202110822850 A CN 202110822850A CN 113505160 A CN113505160 A CN 113505160A
Authority
CN
China
Prior art keywords
motor vehicle
data
tail gas
neural network
emission
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110822850.XA
Other languages
Chinese (zh)
Inventor
陆星家
王志
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ningbo University of Technology
Original Assignee
Ningbo University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ningbo University of Technology filed Critical Ningbo University of Technology
Priority to CN202110822850.XA priority Critical patent/CN113505160A/en
Publication of CN113505160A publication Critical patent/CN113505160A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2465—Query processing support for facilitating data mining operations in structured databases
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/0004—Gaseous mixtures, e.g. polluted air
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Development Economics (AREA)
  • Databases & Information Systems (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Educational Administration (AREA)
  • General Health & Medical Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Biochemistry (AREA)
  • Artificial Intelligence (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Food Science & Technology (AREA)
  • Fuzzy Systems (AREA)
  • Probability & Statistics with Applications (AREA)
  • Evolutionary Computation (AREA)
  • Biomedical Technology (AREA)
  • Combined Controls Of Internal Combustion Engines (AREA)
  • Analytical Chemistry (AREA)
  • Biophysics (AREA)
  • Computing Systems (AREA)
  • Game Theory and Decision Science (AREA)
  • Molecular Biology (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)

Abstract

The invention discloses a motor vehicle exhaust emission estimation method based on a convolution deep neural network, which realizes real-time online estimation of motor vehicle exhaust emission factors in urban areas or the whole situation by utilizing macroscopic exhaust data acquired by an urban air monitoring station and exhaust emission data on an actual road acquired by motor vehicle exhaust remote sensing detection equipment. The invention overcomes the difference between the traditional motor vehicle tail gas emission factor and the actual road running condition obtained by adopting a rack testing machine, can carry out local and global emission estimation on the urban motor vehicle tail gas, simultaneously combines the current meteorological conditions to further estimate the emission of the motor vehicle tail gas in different areas of a city, and carries out grading and online estimation on the urban motor vehicle tail gas emission on the basis.

Description

Motor vehicle exhaust emission estimation method based on convolution deep neural network
Technical Field
The invention relates to an exhaust emission estimation method, in particular to a motor vehicle exhaust emission estimation method based on a convolution deep neural network.
Background
Since the industrial revolution, air pollution has been accompanied by the progress of human economy, and large-scale environmental disasters have occurred occasionally. Coal is used as a main industrial fuel, and a large amount of dust, sulfur dioxide and sulfur dioxide are generated in the combustion process,Nitrogen oxide and other air pollutants, which bring power and warmth to people and also bring serious environmental pollution problems. PM (particulate matter)2.5Refers to particles in the atmosphere having a diameter of less than or equal to 2.5 microns, also known as accessible lung particles. PM (particulate matter)2.5Small particle size, large specific surface area, and easy enrichment of various toxic metals, acidic oxides, organic pollutants and other chemical substances, bacteria, viruses and other microorganisms in the air, compared with PM10Dust, pollen and mould, PM of pollutants2.5Can penetrate into lung tissue, and is PM for the elderly, children and patients with cardiopulmonary diseases2.5The sensitive population of the contamination.
PM pair according to NASA2.5Estimation of concentration, PM2.5The emission of pollutants presents an ascending situation. PM (particulate matter)2.5The pollutant has a complex composition of constituents including fine particles (primary PM) directly emitted from the combustion process2.5Particles) and secondary particulates generated by multi-phase chemical reactions of atmospheric pollutants (i.e., gases undergo chemical reactions to convert to solids, most typically sulfur dioxide and nitrogen oxides, which react to form sulfates and nitrates). Early PM2.5Research emphasizes passing urban PM2.5The control strategy was studied by source analysis, and multiple studies showed that primary PM2.5The proportion gradually decreases and the proportion of secondary particles becomes more severe.
The tail gas emission of motor vehicles is a main emission source of urban air pollution, and the tail gas emission of the motor vehicles is influenced by indexes such as the number of the motor vehicles in the area, the running speed of the motor vehicles and the like. With the rapid development of network appointment vehicles, the reservation quantity of the network appointment vehicles in many cities is increased year by year, the network appointment vehicles have the characteristics of large data transmission quantity and timely data transmission compared with the traditional taxis, the traffic flow density in a specific area and the traffic flow speeds of different road sections can be estimated by estimating the trip data of the network appointment vehicles, and the tail gas emission quantity of motor vehicles in the area is estimated by combining the vehicle information of the motor vehicles. There are many factors that affect the exhaust emission of motor vehicles, such as the technical level of vehicle emission control, driving conditions, traffic environment, fuel quality, etc. An IVE Model (International Vehicle Emission Model) is developed by the university of California, and the IVE adopts a method based on Vehicle technology and local driving modes, so that the problem of using a motor Vehicle exhaust Emission Model for reference can be well solved.
The invention patent of application number 201510745166.0 discloses a method for correcting vehicle emission factor based on speed of vehicle specific power, which calculates the vehicle specific power according to the vehicle running speed, obtains the specific power distribution condition of different speed intervals, and corrects the specific power distribution condition by using the correction coefficient of average speed calculation. The invention patent needs the operation data of the vehicle, and also needs the data of basic emission factors, emission rates in the MOVES database and the like, has more calculation variables, and does not consider the influence of meteorological conditions on the exhaust emission of the motor vehicle.
The invention patent of application No. 201611267901.2 discloses a motor vehicle exhaust emission factor estimation method based on an MLP neural network, which utilizes motor vehicle exhaust remote sensing monitoring equipment to acquire motor vehicle exhaust emission data on an actual road, and combines with meteorological conditions such as local temperature, humidity, pressure, wind direction, wind speed and the like of a city. However, laboratory data is difficult to represent actual road conditions, making estimated vehicle exhaust data less accurate.
At present, environment monitoring equipment is arranged in most cities, and environment data of all areas of the cities can be monitored in real time, wherein the air quality fraction calculation method comprises the following steps, and the air quality index of a pollutant item p is calculated according to a formula 1:
Figure BDA0003172444450000021
in equation 1, IAQIpAir mass fraction index representing the pollutant item p:
cp represents the contaminant concentration, i.e. the input value;
BPHiclose to Cp in Table 1The high value of the concentration limit;
BPLolower values of the contaminant concentration limit in table 1, close to Cp;
IAQIHiwith BP in Table 1HiA corresponding air mass fraction index;
IAQILowith BP in Table 1LoA corresponding air mass fraction index;
for air quality index rating: the air quality index classes are divided according to the specifications of table 1; the method for determining the air quality index and the primary pollutants comprises the following steps: and adopting an air quality index calculation method. The air quality index is calculated according to equation 2:
AQI＝max{IAQI1，IAQI2，IAQI3，...，IAQInequation 2
In equation 2, IAQI represents an air mass fraction index, and n represents a pollutant item.
TABLE 1 air quality index and corresponding pollutant item concentration limit
Figure BDA0003172444450000031
The air quality obtained by the air monitoring station mainly comprises AQI, CO and NO2、O3、SO2、PM2.5、PM10And 7 kinds of air pollutant detection data.
In summary, it is highly desirable to provide a motor vehicle exhaust emission estimation method based on a convolutional deep neural network, in which real exhaust emission levels of motor vehicles running are obtained by motor vehicle exhaust monitoring devices beside roads in some cities, and the air pollution data of an air quality monitoring station and the exhaust emission data obtained by road motor vehicle exhaust remote sensing monitoring devices are used to evaluate the motor vehicle exhaust emission in local and global areas of the city.
Disclosure of Invention
In order to solve the defects of the technology, the invention provides a motor vehicle exhaust emission estimation method based on a convolution deep neural network.
In order to solve the technical problems, the invention adopts the technical scheme that: a motor vehicle exhaust emission estimation method based on a convolution deep neural network comprises the following processing steps:
estimating the exhaust emission of motor vehicles in an urban area, and acquiring local and global air quality data of the city;
step two, preprocessing the motor vehicle exhaust emission data and the global air quality data obtained in the step one, and establishing a motor vehicle exhaust emission factor database; combining the urban meteorological condition data and the air quality database to establish an urban local area and global motor vehicle exhaust emission database;
and step three, establishing a deep convolutional neural network model based on the motor vehicle exhaust emission factor database obtained in the step two and the data acquired in the step one, and performing real-time online estimation on the urban area or global motor vehicle exhaust emission factors according to the deep convolutional neural network model.
Further, in the first step, interpolation operation is carried out on the air quality of different areas of the city by utilizing real-time air quality data of the city environment monitoring station, and the motor vehicle tail gas emission data of the actual road, which is acquired by the motor vehicle tail gas remote sensing monitoring equipment, is utilized to estimate the motor vehicle tail gas emission of the city areas by combining meteorological condition data of each area of the city.
Further, the data acquisition process of the remote sensing monitoring equipment for the tail gas of the motor vehicle comprises the following steps: the motor vehicle tail gas remote sensing monitoring equipment emits infrared light and ultraviolet light with specific wavelength, the infrared light and the ultraviolet light pass through the motor vehicle tail gas and can be absorbed by gas in the tail gas, the intensity of the light is weakened, the receiving end receives the weakened light, and CO and NO are calculated2、O3、SO2、PM2.5、PM10The emission concentration of (a);
the speed and acceleration detector of the motor vehicle tail gas remote sensing monitoring equipment measures the speed and acceleration of the motor vehicle by utilizing the time interval of the wheels passing through the two correlation light paths;
the image acquisition system of the motor vehicle tail gas remote sensing monitoring equipment can acquire the vehicle type of the motor vehicle, and the motor vehicle type is artificially divided into four types: light gasoline vehicles, heavy gasoline vehicles, light diesel vehicles, heavy diesel vehicles;
the environment data obtained by the auxiliary equipment comprises current time, weather, temperature, humidity, pressure, wind direction and wind speed.
Further, motor vehicle exhaust remote sensing monitoring equipment collects motor vehicle exhaust data of a local area, and the local area data and air quality data of an urban air quality monitoring station are fused to obtain local and global air quality data of an city.
Further, in the second step, an emission factor database of the motor vehicle exhaust gases CO, HC and NO is established, and the CO emitted when the motor vehicle runs and collected by motor vehicle exhaust remote sensing monitoring equipment arranged beside the urban road is established2And the volume concentration data of CO, HC and NO calculate the emission factors of the tail gas CO, HC and NO of the motor vehicle, as shown in the formula 3-5:
Figure BDA0003172444450000051
Figure BDA0003172444450000052
Figure BDA0003172444450000053
in formulas 3 to 5, CO (gL)-1)、HC(gL-1)、NO(gL-1) Respectively refer to the emission factors of CO, HC and NO in the tail gas of the motor vehicle, and the unit is gL-1(ii) a Q is CO and CO collected by remote sensing monitoring equipment for tail gas of motor vehicle2Volume concentration ratio; q' is HC and CO collected by remote sensing monitoring equipment for motor vehicle exhaust2Volume concentration ratio; q' NO and CO collected by motor vehicle tail gas remote sensing monitoring equipment2Volume concentration ratio; mfuelIs motorizedThe molar mass of the vehicle fuel oil; dfuelIs the density of the motor vehicle fuel.
Further, in the third step, the emission factor data of CO, HC and NO and the speed, acceleration, temperature, humidity, pressure, wind direction and wind speed data are normalized by the following formulas 6 and 7:
Figure BDA0003172444450000054
wherein, the data is marked as x, and the maximum value and the minimum value in the reorganized data are respectively marked as xmaxAnd xminAnd x' is normalized data;
Figure BDA0003172444450000055
wherein the content of the first and second substances,
Figure BDA0003172444450000056
is the overall mean of the data, μ the sample mean of the data set, and σ is the variance of the sample data;
after data are normalized, a data set is divided into a training set, a verification set and a test set according to a vehicle type, the training set trains the convolution deep neural network, the verification set checks the performance of the convolution deep neural network in the training process, and the test set can be used for evaluating the performance of the convolution deep neural network trained; the proportion of the training set data, the verification set data and the test set data is 60 percent, 20 percent and 20 percent respectively.
Further, the verification set sets a training suspension condition through a training effect, the training suspension condition is terminated when the performance condition of the trained convolutional deep neural network reaches a maximum value or begins to be reduced, the performance detection standard of the convolutional deep neural network is the occurrence rate of type I errors and type II errors, and the performance of the convolutional deep neural network is improved by reducing the proportion of false negatives and false positives, as shown in formulas 8-12;
Figure BDA0003172444450000061
Figure BDA0003172444450000062
efficiency-sensitivity-1-beta equation 10
Figure BDA0003172444450000063
Figure BDA0003172444450000064
In the formulas 8 to 12, FP represents false positive, TN represents true negative, TP represents true positive, FN represents false negative, α represents false positive rate, and β represents false negative rate.
Further, the structure of the convolution deep neural network model used is: the input layer, the convolution layer, the pooling layer, the full-connection layer and the output layer.
Further, the input of the convolution depth network model is the motion data of the motor vehicle, including the speed and the acceleration of the motor vehicle and the traffic flow data of a road area, the temperature, the humidity, the pressure, the wind direction and the wind speed of the fusion meteorological data, and the output of the convolution depth network model is CO, HC, NO, CO and NO2、O3、SO2、PM2.5、PM10The exhaust emission of (1).
The invention discloses a motor vehicle exhaust emission estimation method based on a convolution deep neural network, which has the advantages compared with the prior art that:
1) the motor vehicle tail gas emission data is adopted to collect data on an actual road and global air quality data of an air quality monitoring station, which are acquired by motor vehicle tail gas remote sensing monitoring equipment, so that local distribution of tail gas emission can be obtained, global distribution of motor vehicle tail gas is considered, the local data can truly reflect the emission level of a motor vehicle under an actual working condition, and speed and acceleration data with a large range can be obtained on the basis of an actual road structure by combining global data of an urban road network, and motor vehicle emission data under various temperature, humidity, pressure, wind direction and wind speed conditions can be obtained.
2) The deep convolution network is used for establishing the relation between the driving condition and the meteorological condition of the motor vehicle and the exhaust emission factor of the motor vehicle, the influence of the driving condition and the meteorological condition on the emission factor is complex, the deep convolution network can continuously receive input and output data in the training process even if the complex nonlinear relation between the input and the output is not known to a great extent, and the connection weight between neurons and the connection weight between the hierarchies are improved by adjusting the number of different hierarchies and nodes, so that the internal relation between the input and the output is established.
3) The deep convolutional network used by the invention comprises an input layer, a plurality of convolutional layers and a pooling layer, and the structure can accurately acquire the nonlinear relation between an input variable and an output variable so as to estimate the motor vehicle exhaust emission in local and global ranges of a city.
Drawings
FIG. 1 is a schematic structural diagram of a deep convolutional neural network according to the present invention.
FIG. 2 is a schematic diagram of the estimation of the exhaust emission of the motor vehicle based on the deep convolutional neural network.
Fig. 3 is a schematic diagram of tail gas detection sections of guangxi roads in the sea area of Ningbo town in the embodiment of the present invention.
FIG. 4 is a schematic diagram of the estimation result of the tail gas space of Ningbo city motor vehicles in the embodiment of the invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
The invention discloses a motor vehicle exhaust emission estimation method based on a convolution deep neural network, which is an estimation method for motor vehicle exhaust emission factors in urban road areas and overcomes the defect that the traditional motor vehicle exhaust emission factors adopt a rack testing machine to obtain motor vehicle exhaust emission data and actual motor vehicle exhaust emission dataThe invention utilizes the macroscopic tail gas data acquired by an urban air monitoring station and the tail gas emission data on the actual road acquired by a motor vehicle tail gas remote sensing detection device to realize the following steps: including CO and NO emitted during running of motor vehicle2、O3、SO2、PM2.5、PM10And the like, tail gas concentration data, automobile model, speed, acceleration, environmental data and the like. The invention mainly comprises the following processing steps:
estimating the exhaust emission of motor vehicles in an urban area, and acquiring local and global air quality data of the city;
step two, preprocessing the motor vehicle exhaust emission data and the global air quality data obtained in the step one, and establishing a motor vehicle exhaust emission factor database; combining the urban meteorological condition data and the air quality database to establish an urban local area and global motor vehicle exhaust emission database;
and step three, establishing a deep convolutional neural network model based on the motor vehicle exhaust emission factor database obtained in the step two and the data acquired in the step one, and performing real-time online estimation on the urban area or global motor vehicle exhaust emission factors according to the deep convolutional neural network model.
In the first step, real-time air quality data of an urban environment monitoring site are utilized to perform interpolation operation on air quality of different areas of an urban area, and motor vehicle tail gas emission data of an actual road, which is acquired by motor vehicle tail gas remote sensing monitoring equipment, is utilized to estimate motor vehicle tail gas emission of the urban area in combination with meteorological condition data of each area of the urban area;
the data acquisition process of the motor vehicle tail gas remote sensing monitoring equipment comprises the following steps: the motor vehicle tail gas remote sensing monitoring equipment emits infrared light and ultraviolet light with specific wavelength, the infrared light and the ultraviolet light pass through the motor vehicle tail gas and can be absorbed by gas in the tail gas, the intensity of the light is weakened, the receiving end receives the weakened light, and CO and NO are calculated2、O3、SO2、PM2.5、PM10The emission concentration of (a);
the speed and acceleration detector of the motor vehicle tail gas remote sensing monitoring equipment measures the speed and acceleration of the motor vehicle by utilizing the time interval of the wheels passing through the two correlation light paths;
the image acquisition system of the motor vehicle tail gas remote sensing monitoring equipment can acquire the vehicle type of the motor vehicle, and the motor vehicle type is artificially divided into four types: light gasoline vehicles, heavy gasoline vehicles, light diesel vehicles, heavy diesel vehicles;
the environment data obtained by the auxiliary equipment comprises current time, weather, temperature, humidity, pressure, wind direction and wind speed.
The motor vehicle tail gas remote sensing monitoring equipment acquires motor vehicle tail gas data of a local area, and the local area data and the air quality data of the urban air quality monitoring station are fused to obtain local and global air quality data of the city.
In the second step, an emission factor database of the motor vehicle exhaust CO, HC and NO is established, and the CO emitted when the motor vehicle runs and collected by motor vehicle exhaust remote sensing monitoring equipment arranged beside a city road is collected2And the volume concentration data of CO, HC and NO calculate the emission factors of the tail gas CO, HC and NO of the motor vehicle, as shown in the formula 3-5:
Figure BDA0003172444450000091
Figure BDA0003172444450000092
Figure BDA0003172444450000093
in formulas 3 to 5, CO (gL)-1)、HC(gL-1)、NO(gL-1) Respectively refer to the emission factors of CO, HC and NO in the tail gas of the motor vehicle, and the unit is gL-1(ii) a Q is CO and CO collected by remote sensing monitoring equipment for tail gas of motor vehicle2Volume concentration ratio; q' is the remote sensing monitoring equipment of motor vehicle exhaust gasCollected HC and CO2Volume concentration ratio; q' NO and CO collected by motor vehicle tail gas remote sensing monitoring equipment2Volume concentration ratio; mfuelIs the molar mass of the motor vehicle fuel; dfuelIs the density of the motor vehicle fuel.
In the third step, the emission factor data of CO, HC and NO and the speed, acceleration, temperature, humidity, pressure, wind direction and wind speed data are normalized by the following formulas 6 and 7:
Figure BDA0003172444450000094
wherein, the data is marked as x, and the maximum value and the minimum value in the reorganized data are respectively marked as xmaxAnd xminAnd x' is normalized data;
Figure BDA0003172444450000101
where x is the overall mean of the data, μ is the sample mean of the data set, and σ is the variance of the sample data.
After the data are normalized, the data are classified according to vehicle types, wherein the data mainly comprise four types of data such as light gasoline vehicles, heavy gasoline vehicles, light diesel vehicles and heavy diesel vehicles. Dividing a data set into a training set, a verification set and a test set, training the convolution deep neural network by the training set, checking the performance of the convolution deep neural network by the verification set in the training process, setting a training suspension condition according to the training effect, terminating when the performance condition of the trained convolution deep neural network reaches a maximum value or begins to be reduced, and determining the performance detection standard of the convolution deep neural network as the occurrence rate of I-type errors and II-type errors, as shown in a formula 8 and a formula 9; the performance of the network is improved by reducing the proportion of false negative and false positive, as shown in formulas 11 and 12; the type I error, namely the false positive rate (alpha), occurs in the deep convolutional network, and indicates that negative sample data is misjudged as positive samples. Class II errors indicate a false negative rate (β), and class II errors occur, indicating that positive sample data is missed as a negative sample. The test set can be used to evaluate the performance of the trained convolutional deep neural network. The proportion of the training set data, the verification set data and the test set data is 60 percent, 20 percent and 20 percent respectively.
Figure BDA0003172444450000102
Figure BDA0003172444450000103
Efficiency-sensitivity-1-beta equation 10
Figure BDA0003172444450000104
Figure BDA0003172444450000105
In the formulas 8-12, FP represents false positive, TN represents true negative, TP represents true positive, FN represents false negative, α represents false positive rate, and β represents false negative rate; equation 10 represents the sensitivity of the deep convolutional neural network, the higher the sensitivity, the lower the undetected rate, the likelihood ratio is used to evaluate the value of the deep convolutional network to the model estimation, the positive likelihood ratio of equation 11 represents the result of the true negative rate to the false positive rate, and the negative likelihood ratio of equation 12 represents the result of the false negative rate to the true positive rate.
The structure of the convolutional deep neural network model used is shown in fig. 1, and has: the input layer, the convolution layer, the pooling layer, the full-connection layer and the output layer.
The input of the convolution depth network model is the motion data of the motor vehicle, including the speed and the acceleration of the motor vehicle and the traffic flow data of a road area, the temperature, the humidity, the pressure, the wind direction and the wind speed of the fusion meteorological data, and the output is CO, HC, NO, CO and NO2、O3、SO2、PM2.5、PM10And the like. The number of the input layer neural units is 483, the number of the output layer neural units is 9, and the hierarchy of the convolution depth neural network is 7 layers.
The technical solution of the present invention is further described in detail with reference to fig. 1 and 2.
The method comprises the steps of firstly, utilizing real-time air quality data of an urban environment monitoring station to carry out interpolation operation on air quality of different areas of an urban area, combining meteorological condition data of each area of the urban area, and estimating motor vehicle tail gas emission of the urban area by utilizing motor vehicle tail gas emission data of an actual road collected by motor vehicle tail gas remote sensing monitoring equipment.
The motor vehicle tail gas remote sensing monitoring equipment can detect the polluted gas in the motor vehicle tail gas detected by the motor vehicle tail gas remote sensing monitoring equipment by utilizing the tail gas, the motor vehicle tail gas remote sensing monitoring equipment emits infrared light and ultraviolet light with specific wavelength, the light passes through the motor vehicle tail gas and can be absorbed by the gas in the tail gas, the intensity of the light is weakened, the receiving end receives the weakened light, and the CO and NO are calculated2、O3、SO2、PM2.5、PM10The emission concentration of (a); meanwhile, a speed acceleration detector of the motor vehicle tail gas remote sensing monitoring equipment measures the speed and the acceleration of the motor vehicle by utilizing the time interval of the wheels passing through two correlation light paths; the image acquisition system of the motor vehicle tail gas remote sensing monitoring equipment can acquire the vehicle type of the motor vehicle, and the vehicle type of the motor vehicle is divided into four types: light gasoline vehicles, heavy gasoline vehicles, light diesel vehicles, heavy diesel vehicles; the environmental data obtained by other auxiliary equipment comprises current time, weather, temperature, humidity, pressure, wind direction and wind speed.
The motor vehicle tail gas remote sensing monitoring equipment acquires motor vehicle tail gas data of a local area, and the local area data and the air quality data of the urban air quality monitoring station are fused to obtain local and global air quality data of the city.
And step two, preprocessing the motor vehicle exhaust emission data and the global air quality data acquired in the step one, establishing an emission factor database of motor vehicle exhaust CO, HC and NO, and establishing a city local area and global motor vehicle exhaust emission database by combining the city meteorological condition data and the air quality database.
CO (carbon monoxide) emitted by motor vehicles during running and collected by motor vehicle tail gas remote sensing monitoring equipment distributed beside urban roads2And the volume concentration data of CO, HC and NO calculate the emission factors of the tail gas CO, HC and NO of the motor vehicle, as shown in the formula 3-5:
Figure BDA0003172444450000121
Figure BDA0003172444450000122
Figure BDA0003172444450000123
in formulas 3 to 5, CO (gL)-1)、HC(gL-1)、NO(gL-1) Respectively refer to the emission factors of CO, HC and NO in the tail gas of the motor vehicle, and the unit is gL-1(ii) a Q is CO and CO collected by remote sensing monitoring equipment for tail gas of motor vehicle2Volume concentration ratio; q' is HC and CO collected by remote sensing monitoring equipment for motor vehicle exhaust2Volume concentration ratio; q' NO and CO collected by motor vehicle tail gas remote sensing monitoring equipment2Volume concentration ratio; mfuelIs the molar mass of the motor vehicle fuel; dfuelIs the density of the motor vehicle fuel.
Taking the corresponding data of the molar mass and density of the gasoline taken in the above formula, the following emission factor calculation formula for gasoline cars is obtained:
Figure BDA0003172444450000124
Figure BDA0003172444450000131
Figure BDA0003172444450000132
and step three, respectively establishing a deep convolutional neural network model aiming at the CO, HC and NO based on the motor vehicle exhaust CO, HC and NO emission factor database obtained in the step two and other relevant data collected in the step one, and realizing real-time online estimation of the urban area or global motor vehicle exhaust emission factor according to the deep learning model.
The emission factor data and the speed, acceleration, temperature, humidity, pressure, wind direction and wind speed data of CO, HC and NO are normalized by equations 6 and 7:
Figure BDA0003172444450000133
wherein, the data is marked as x, and the maximum value and the minimum value in the reorganized data are respectively marked as xmaxAnd xminAnd x' is normalized data;
Figure BDA0003172444450000134
wherein the content of the first and second substances,
Figure BDA0003172444450000135
is the overall mean of the data, mu is the sample mean of the data set, and sigma is the variance of the sample data.
After the data are normalized, the data are classified according to vehicle types, wherein the data mainly comprise four types of data such as light gasoline vehicles, heavy gasoline vehicles, light diesel vehicles and heavy diesel vehicles. Dividing a data set into a training set, a verification set and a test set, training the convolution deep neural network by the training set, checking the performance of the convolution deep neural network by the verification set in the training process, setting a training suspension condition according to the training effect, terminating when the performance condition of the trained convolution deep neural network reaches a maximum value or begins to be reduced, and determining the performance detection standard of the convolution deep neural network as the occurrence rate of I-type errors and II-type errors, as shown in a formula 8 and a formula 9; the performance of the neural network is improved by reducing the proportion of false negative and false positive, as shown in formulas 11 and 12; the type I error, namely the false positive rate (alpha), occurs in the deep convolutional network, and indicates that negative sample data is misjudged as positive samples. Class II errors indicate a false negative rate (β), and class II errors occur, indicating that positive sample data is missed as a negative sample. The test set can be used to evaluate the performance of the trained convolutional deep neural network. The proportion of the training set data, the verification set data and the test set data is 60 percent, 20 percent and 20 percent respectively.
Figure BDA0003172444450000141
Figure BDA0003172444450000142
Efficiency-sensitivity-1-beta equation 10
Figure BDA0003172444450000143
Figure BDA0003172444450000144
In the formulas 8-12, FP represents false positive, TN represents true negative, TP represents true positive, FN represents false negative, α represents false positive rate, and β represents false negative rate; equation 10 represents the sensitivity of the deep convolutional neural network, the higher the sensitivity, the lower the undetected rate, the likelihood ratio is used to evaluate the value of the deep convolutional network to the model estimation, the positive likelihood ratio of equation 11 represents the result of the true negative rate to the false positive rate, and the negative likelihood ratio of equation 12 represents the result of the false negative rate to the true positive rate.
The structure of the convolutional deep neural network model used is shown in fig. 1, and has: the input layer, the convolution layer, the pooling layer, the full-connection layer and the output layer.
The input of the convolution depth network model is the motion data of the motor vehicle, including the speed and the acceleration of the motor vehicle and the traffic flow data of a road area, the temperature, the humidity, the pressure, the wind direction and the wind speed of the fusion meteorological data, and the output is CO, HC, NO, CO and NO2、O3、SO2、PM2.5、PM10Waiting for the estimation result of the exhaust emission; the traffic data can be acquired by connecting a database of an electronic police detection system of a traffic police branch. The number of the input layer neural units is 483, the number of the output layer neural units is 9, and the hierarchy of the convolution depth neural network is 7 layers.
Wherein the neurons of the convolutional deep neural network have the following form, as shown in equation 13:
Figure BDA0003172444450000151
Figure BDA0003172444450000152
wherein the content of the first and second substances,
Figure BDA0003172444450000153
is the output of the 0 th neuron of the input layer, N is the number of neurons of the input layer, wk, i is the connection weight between the kth neuron of the input layer and the ith neuron of the convolutional layer, and k is 0,1,2, … N; biThe deviation constant is the ith deviation constant, sigma represents an activation function, and the learning process of the input data of the convolution deep neural network is carried out in a feedback learning mode; in the learning process of feedback, the weight coefficient is corrected by the sigma function, and the next stage a is obtained by the delivery of the inner product in the formula 14l+1：
al+1＝σ(Wlal+bl) Formula (II)14
In the formula, al+1Representing the output of the convolutional neural network training step l +1, WlalInner product of weight and output parameter representing l training times, blBias representing the l steps;
the sigma function is a Sigmold function, which has excellent normalization characteristic, when the input variable tends to minus infinity, the Sigmold function value tends to 0, and when the input variable tends to plus infinity, the Sigmold function value tends to 1. Equation 15 transforms the values of different ranges into [0,1] range uniformly, i.e. data normalization is performed.
Figure BDA0003172444450000154
In the formula, σ (x) represents a Sigmold function, e-xRepresenting a natural number exponential function;
the pooling layer adopts a Max function, the matrix output by the Sigmold function is subjected to maximization processing, the maximum value of each row is selected, in the pooling process, the dimension of a data input matrix can be reduced, and according to the core selected by the pooling layer, if a pooling core of 2 x 2 is selected, the input matrix is a square matrix of (n, n), and the output matrix is changed into a matrix of (n/2 ); equation 16 extracts the most significant features of the pooling layer by maximizing.
Figure BDA0003172444450000161
Wherein n represents the dimension of the output value, n/2 represents half of the dimension of the output value, annThe output variable is represented by a number of variables,
Figure BDA0003172444450000162
representing a bias variable;
training a model based on a training set, respectively carrying out contrastive analysis on a series of models obtained by training based on a verification set and a test set, wherein the number of input layers of the trained model is 784 neurons, the number of layers of a deep convolutional network is selected to be 7, after multiple experiments, the number of the neurons of a convolutional layer and a pooling layer behind the input layers can be consistent with the number of the input layers, the number of the neurons of a second convolutional layer and a pooling layer can be reduced to 392, and the number of the convolutional layers and the full-connection layers on the third layer is consistent with the number of the full-connection layers.
Therefore, the invention can carry out local and global emission estimation on the urban motor vehicle exhaust by combining the urban air quality monitoring data aiming at the exhaust emitted by the motor vehicles in different time periods and different areas, and simultaneously estimates the emission of the motor vehicle exhaust such as CO, HC and NO in different areas of the city by combining the current meteorological conditions, and carries out grading and online estimation on the urban motor vehicle exhaust emission on the basis.
[ examples ] A method for producing a compound
The Guangxi road in the Ningbo town sea area is used as an application example of the motor vehicle exhaust emission estimation method based on the convolution depth neural network, the Guangxi road is in the Ningbo town sea area, the road length is 4.3KM, the road section is in the east-west trend, the road section has 9 traffic light intersections in total, and the road section is provided with green wave bands, so that the traffic flow condition of the road section and the real-time exhaust emission condition of the vehicle are conveniently recorded (figure 3).
And (3) carrying out interpolation estimation on the motor vehicle exhaust of the road in the designated area by adopting a space interpolation method at the designated road section, wherein the interpolation formula adopts linear interpolation, which is a simpler interpolation method, and the interpolation function is a first-order polynomial. And linear interpolation, wherein the interpolation error is 0 at each interpolation node.
Let the function y be f (x) at two points x0，x1Are each y0，y1Solving a polynomial equation:
Figure BDA0003172444450000163
equation 17 is a first order linear equation estimate of the interpolation function such that equation 18 (sample point) is satisfied:
Figure BDA0003172444450000171
the first order linear interpolation of equation 19 estimates the unknown y variable:
Figure BDA0003172444450000172
wherein
Figure BDA0003172444450000173
Is at x0The first order difference of (d), equation 20 can be obtained,
Figure BDA0003172444450000174
according to y0，y1It can be seen that in the equation 21,
Figure BDA0003172444450000175
by linear interpolation, an estimate of the concentration of vehicle emissions in the city region of Ningbo city can be obtained, as shown in FIG. 4. Namely, the measured space motor vehicle exhaust emission concentration can be calculated out by first-order linear interpolation, and the motor vehicle exhaust concentration in the neighborhood of the detection point can be calculated.
The above embodiments are not intended to limit the present invention, and the present invention is not limited to the above examples, and those skilled in the art may make variations, modifications, additions or substitutions within the technical scope of the present invention.

Claims (9)

1. A motor vehicle exhaust emission estimation method based on a convolution deep neural network is characterized in that: the method comprises the following processing steps:
estimating the exhaust emission of motor vehicles in an urban area, and acquiring local and global air quality data of the city;
step two, preprocessing the motor vehicle exhaust emission data and the global air quality data obtained in the step one, and establishing a motor vehicle exhaust emission factor database; combining the urban meteorological condition data and the air quality database to establish an urban local area and global motor vehicle exhaust emission database;
and step three, establishing a deep convolutional neural network model based on the motor vehicle exhaust emission factor database obtained in the step two and the data acquired in the step one, and performing real-time online estimation on the urban area or global motor vehicle exhaust emission factors according to the deep convolutional neural network model.
2. The method of estimating the exhaust emission of a motor vehicle based on a convolutional deep neural network as claimed in claim 1, wherein: in the first step, real-time air quality data of an urban environment monitoring site are utilized to perform interpolation operation on air quality of different areas of an urban area, meteorological condition data of each area of the urban area are combined, and meanwhile motor vehicle tail gas emission data of an actual road, which are acquired by motor vehicle tail gas remote sensing monitoring equipment, are utilized to estimate motor vehicle tail gas emission of the urban area.
3. The method of estimating the exhaust emission of a motor vehicle based on a convolutional deep neural network as claimed in claim 2, wherein: the data acquisition process of the motor vehicle tail gas remote sensing monitoring equipment comprises the following steps: the motor vehicle tail gas remote sensing monitoring equipment emits infrared light and ultraviolet light with specific wavelength, the infrared light and the ultraviolet light pass through the motor vehicle tail gas and can be absorbed by gas in the tail gas, the intensity of the light is weakened, the receiving end receives the weakened light, and CO and NO are calculated2、O3、SO2、PM2.5、PM10The emission concentration of (a);
the speed and acceleration detector of the motor vehicle tail gas remote sensing monitoring equipment measures the speed and acceleration of the motor vehicle by utilizing the time interval of the wheels passing through the two correlation light paths;
the image acquisition system of the motor vehicle tail gas remote sensing monitoring equipment can acquire the vehicle type of the motor vehicle, and the motor vehicle type is artificially divided into four types: light gasoline vehicles, heavy gasoline vehicles, light diesel vehicles, heavy diesel vehicles;
the environment data obtained by the auxiliary equipment comprises current time, weather, temperature, humidity, pressure, wind direction and wind speed.
4. The method of claim 3 for estimating the exhaust emission of a motor vehicle based on a convolutional deep neural network, wherein: the motor vehicle tail gas remote sensing monitoring equipment acquires motor vehicle tail gas data of a local area, and the local area data and the air quality data of the urban air quality monitoring station are fused to obtain local and global air quality data of the city.
5. The method for estimating the exhaust emission of a motor vehicle based on a convolutional deep neural network as claimed in claim 1 or 4, wherein: in the second step, an emission factor database of the motor vehicle exhaust CO, HC and NO is established, and the CO emitted when the motor vehicle runs and collected by motor vehicle exhaust remote sensing monitoring equipment arranged beside a city road is collected2And the volume concentration data of CO, HC and NO calculate the emission factors of the tail gas CO, HC and NO of the motor vehicle, as shown in the formula 3-5:
Figure FDA0003172444440000021
Figure FDA0003172444440000022
Figure FDA0003172444440000023
in formulas 3 to 5, CO (gL)-1)、HC(gL-1)、NO(gL-1) Respectively refer to the emission factors of CO, HC and NO in the tail gas of the motor vehicleIn the unit of gL-1(ii) a Q is CO and CO collected by remote sensing monitoring equipment for tail gas of motor vehicle2Volume concentration ratio; q' is HC and CO collected by remote sensing monitoring equipment for motor vehicle exhaust2Volume concentration ratio; q' NO and CO collected by motor vehicle tail gas remote sensing monitoring equipment2Volume concentration ratio; mfuelIs the molar mass of the motor vehicle fuel; dfuelIs the density of the motor vehicle fuel.
6. The method of claim 5 for estimating the exhaust emission of a motor vehicle based on a convolutional deep neural network, wherein: in the third step, the emission factor data of CO, HC and NO and the speed, acceleration, temperature, humidity, pressure, wind direction and wind speed data are normalized by the following formulas 6 and 7:
Figure FDA0003172444440000024
wherein, the data is marked as x, and the maximum value and the minimum value in the reorganized data are respectively marked as xmaxAnd xminAnd x' is normalized data;
Figure FDA0003172444440000031
wherein the content of the first and second substances,
Figure FDA0003172444440000032
is the overall mean of the data, μ the sample mean of the data set, and σ is the variance of the sample data;
after data are normalized, a data set is divided into a training set, a verification set and a test set according to a vehicle type, the training set trains the convolution deep neural network, the verification set checks the performance of the convolution deep neural network in the training process, and the test set can be used for evaluating the performance of the convolution deep neural network trained; the proportion of the training set data, the verification set data and the test set data is 60 percent, 20 percent and 20 percent respectively.
7. The method of claim 6, wherein the method comprises: the verification set sets a training suspension condition through a training effect, the training suspension condition is terminated when the performance condition of the trained convolutional deep neural network reaches a maximum value or begins to be reduced, the performance detection standard of the convolutional deep neural network is the incidence rate of I-type errors and II-type errors, and the performance of the convolutional deep neural network is improved by reducing the proportion of false negative and false positive, as shown in formulas 8-12;
Figure FDA0003172444440000033
Figure FDA0003172444440000034
efficiency-sensitivity-1-beta equation 10
Figure FDA0003172444440000035
Figure FDA0003172444440000036
In the formulas 8 to 12, FP represents false positive, TN represents true negative, TP represents true positive, FN represents false negative, α represents false positive rate, and β represents false negative rate.
8. The method of claim 7 for estimating the exhaust emission of a motor vehicle based on a convolutional deep neural network, wherein: the structure of the convolutional deep neural network model used is: the input layer, the convolution layer, the pooling layer, the full-connection layer and the output layer.
9. The method of estimating the exhaust emission of a motor vehicle based on a convolutional deep neural network as claimed in claim 8, wherein: the convolution depth network model has the input of the motion data of the motor vehicle, including the speed and the acceleration of the motor vehicle and the traffic flow data of the road area, the temperature, the humidity, the pressure, the wind direction and the wind speed of the fusion meteorological data and the output of the motion data including CO, HC, NO, CO and NO2、O3、SO2、PM2.5、PM10The exhaust emission of (1).
CN202110822850.XA 2021-07-21 2021-07-21 Motor vehicle exhaust emission estimation method based on convolution deep neural network Pending CN113505160A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110822850.XA CN113505160A (en) 2021-07-21 2021-07-21 Motor vehicle exhaust emission estimation method based on convolution deep neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110822850.XA CN113505160A (en) 2021-07-21 2021-07-21 Motor vehicle exhaust emission estimation method based on convolution deep neural network

Publications (1)

Publication Number Publication Date
CN113505160A true CN113505160A (en) 2021-10-15

Family

ID=78013327

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110822850.XA Pending CN113505160A (en) 2021-07-21 2021-07-21 Motor vehicle exhaust emission estimation method based on convolution deep neural network

Country Status (1)

Country Link
CN (1) CN113505160A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117093829A (en) * 2023-10-19 2023-11-21 四川国蓝中天环境科技集团有限公司 Inferior oil filling point identification method and system based on vehicle-mounted diagnostic data

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106650287A (en) * 2016-12-31 2017-05-10 中国科学技术大学 Motor vehicle exhaust emission factor estimation method based on MLP neural network
CN106650158A (en) * 2016-12-31 2017-05-10 中国科学技术大学 Computational fluid dynamics (CFD) and multi-data sources-based urban real-time global environment estimation method
CN108593844A (en) * 2018-04-19 2018-09-28 沈夕尧 A kind of motor-vehicle tail-gas remote sensing monitoring method
CN110133191A (en) * 2019-05-16 2019-08-16 北京英视睿达科技有限公司 A method of road automobile discharge is calculated based on air quality monitoring equipment
CN112132264A (en) * 2020-09-11 2020-12-25 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) Regional exhaust emission prediction method and system based on space-time residual perception network
CN112712007A (en) * 2020-12-25 2021-04-27 天津城建大学 Algorithm for identifying automobile exhaust remote sensing data based on DCNN-inclusion network

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106650287A (en) * 2016-12-31 2017-05-10 中国科学技术大学 Motor vehicle exhaust emission factor estimation method based on MLP neural network
CN106650158A (en) * 2016-12-31 2017-05-10 中国科学技术大学 Computational fluid dynamics (CFD) and multi-data sources-based urban real-time global environment estimation method
CN108593844A (en) * 2018-04-19 2018-09-28 沈夕尧 A kind of motor-vehicle tail-gas remote sensing monitoring method
CN110133191A (en) * 2019-05-16 2019-08-16 北京英视睿达科技有限公司 A method of road automobile discharge is calculated based on air quality monitoring equipment
CN112132264A (en) * 2020-09-11 2020-12-25 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) Regional exhaust emission prediction method and system based on space-time residual perception network
CN112712007A (en) * 2020-12-25 2021-04-27 天津城建大学 Algorithm for identifying automobile exhaust remote sensing data based on DCNN-inclusion network

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117093829A (en) * 2023-10-19 2023-11-21 四川国蓝中天环境科技集团有限公司 Inferior oil filling point identification method and system based on vehicle-mounted diagnostic data
CN117093829B (en) * 2023-10-19 2024-01-23 四川国蓝中天环境科技集团有限公司 Inferior oil filling point identification method and system based on vehicle-mounted diagnostic data

Similar Documents

Publication Publication Date Title
CN112113912B (en) Remote sensing big data monitoring system and method for diesel vehicle emission
CN106650825B (en) Motor vehicle exhaust emission data fusion system
US11226289B2 (en) Remote sensing-based detection system and method for gaseous pollutant from diesel vehicle exhaust
CN107577910B (en) Vehicle exhaust concentration inversion method based on deep neural network
CN106920007B (en) PM based on second-order self-organizing fuzzy neural network2.5Intelligent prediction method
US11938784B2 (en) Health protection system for passengers on a train in a polluted indoor environment and method thereof
CN109086946B (en) Method for predicting emission of polluted gas of conventional energy and new energy public transport vehicle
CN110135466B (en) Pollutant emission standard exceeding vehicle judgment method and system
CN108197731B (en) Motor vehicle exhaust remote measurement and vehicle inspection result consistency method based on co-training
CN109489978B (en) Multi-source data correlation analysis method of diesel locomotive multi-emission detection method based on V-a working condition
CN106650287A (en) Motor vehicle exhaust emission factor estimation method based on MLP neural network
CN106202533A (en) Street canyon pollutant of vehicle exhaust concentration estimating system
CN107330300A (en) A kind of Vehicular exhaust retrieving concentration method based on support vector regression
CN210221806U (en) Pollution determination calibration vehicle and vehicle-mounted tail gas detection equipment thereof
CN102706782B (en) Method for detecting weight of micro-particles exhausted by gasoline car
CN108763643B (en) Regional motor vehicle emission factor calculation method
CN111024898A (en) Vehicle exhaust concentration standard exceeding judging method based on Catboost model
CN111125636A (en) Motor vehicle emission factor calculation method based on urban tunnel
CN111077048A (en) Opportunistic group intelligent air quality monitoring and evaluating method based on mobile equipment
Yusuf et al. Spatial patterns of urban air pollution in an industrial estate, Lagos, Nigeria
CN113505160A (en) Motor vehicle exhaust emission estimation method based on convolution deep neural network
Ragatz et al. Aerodynamic drag reduction technologies testing of heavy-duty vocational vehicles and a dry van trailer
Alattar et al. Evaluating particulate matter (PM2. 5 and PM10) impact on human health in Oman based on a hybrid artificial neural network and mathematical models
CN112394040A (en) Motor vehicle pollution discharge supervision system and method
Alattar et al. Neural and Mathematical Predicting Models for Particulate Matter Impact on Human Health in Oman

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination