CN113435471A - Deep feature clustering high-emission mobile source pollution identification method and system - Google Patents
Deep feature clustering high-emission mobile source pollution identification method and system Download PDFInfo
- Publication number
- CN113435471A CN113435471A CN202110534816.2A CN202110534816A CN113435471A CN 113435471 A CN113435471 A CN 113435471A CN 202110534816 A CN202110534816 A CN 202110534816A CN 113435471 A CN113435471 A CN 113435471A
- Authority
- CN
- China
- Prior art keywords
- data
- emission
- clustering
- forest
- source
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000007637 random forest analysis Methods 0.000 claims abstract description 35
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 28
- 238000007689 inspection Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 238000005259 measurement Methods 0.000 claims abstract description 7
- 239000013598 vector Substances 0.000 claims description 31
- 239000003344 environmental pollutant Substances 0.000 claims description 29
- 231100000719 pollutant Toxicity 0.000 claims description 29
- 238000003066 decision tree Methods 0.000 claims description 25
- 230000002159 abnormal effect Effects 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 11
- 230000001133 acceleration Effects 0.000 claims description 9
- 239000000779 smoke Substances 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 235000019580 granularity Nutrition 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000003064 k means clustering Methods 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 claims description 4
- 230000003197 catalytic effect Effects 0.000 claims description 4
- 239000000356 contaminant Substances 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 4
- 239000000446 fuel Substances 0.000 claims description 4
- 230000003068 static effect Effects 0.000 claims description 4
- 238000013480 data collection Methods 0.000 claims description 3
- 239000007789 gas Substances 0.000 abstract description 46
- 238000001514 detection method Methods 0.000 abstract description 7
- 238000012544 monitoring process Methods 0.000 abstract description 6
- 238000011156 evaluation Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 10
- 238000004590 computer program Methods 0.000 description 8
- 238000013145 classification model Methods 0.000 description 5
- 238000003860 storage Methods 0.000 description 4
- 238000007635 classification algorithm Methods 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000005484 gravity Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
- G01D21/02—Measuring two or more variables by means not covered by a single other subclass
-
- 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
- G01N33/0009—General constructional details of gas analysers, e.g. portable test equipment
- G01N33/0027—General constructional details of gas analysers, e.g. portable test equipment concerning the detector
- G01N33/0036—General constructional details of gas analysers, e.g. portable test equipment concerning the detector specially adapted to detect a particular component
-
- 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
- G01N33/0009—General constructional details of gas analysers, e.g. portable test equipment
- G01N33/0027—General constructional details of gas analysers, e.g. portable test equipment concerning the detector
- G01N33/0036—General constructional details of gas analysers, e.g. portable test equipment concerning the detector specially adapted to detect a particular component
- G01N33/0037—NOx
-
- 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
- G01N33/0009—General constructional details of gas analysers, e.g. portable test equipment
- G01N33/0027—General constructional details of gas analysers, e.g. portable test equipment concerning the detector
- G01N33/0036—General constructional details of gas analysers, e.g. portable test equipment concerning the detector specially adapted to detect a particular component
- G01N33/004—CO or CO2
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A50/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE in human health protection, e.g. against extreme weather
- Y02A50/20—Air quality improvement or preservation, e.g. vehicle emission control or emission reduction by using catalytic converters
Landscapes
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Medicinal Chemistry (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Food Science & Technology (AREA)
- Combustion & Propulsion (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Probability & Statistics with Applications (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a high-emission mobile source pollution identification method and system based on deep feature clustering, which comprises the steps of preprocessing collected annual inspection data of vehicles and remote measurement data of tail gas; the method comprises the following steps of utilizing a random forest feature selection algorithm to carry out importance evaluation on external attributes influenced by the concentrations of main components CO, HC and NO in the exhaust gas of a mobile source, and selecting main influence feature factors of various polluted gases; clustering the data after feature selection by using various clustering algorithms to obtain a high-emission motor vehicle label; and training the high-emission class label data by utilizing the deep forest to obtain an automatic classification recognition model. The method comprehensively considers the influence of external actual factors on pollution detection, screens out main influence factors on different tail gas components, and then respectively models and identifies, thereby effectively improving the prediction precision and providing an effective technical method for monitoring and controlling the high-emission mobile pollution source by related departments.
Description
Technical Field
The invention relates to the technical field of environmental monitoring, in particular to a high-emission mobile source pollution identification method and system based on deep feature clustering.
Background
With the development of economy in China, the number of urban motor vehicles is continuously increased, and harmful gases emitted by mobile sources cause damage to the health of the masses and environmental protection while the national living standard is improved, particularly the high-emission mobile sources. Therefore, there is a need for fast identification of high-emission mobile sources with excessive emission of harmful gases, so that the relevant departments can efficiently manage and control the exhaust emission of the mobile sources.
In the actual remote sensing detection work of the mobile source exhaust, factors such as the service time of a motor vehicle, the vehicle weight, the vehicle speed, the vehicle acceleration, the vehicle length, the VSP (vehicle specific power), the wind speed, the air temperature, the air density, the pressure intensity, the wind direction and the like all have certain influence on the emission, but the existing pollutant concentration measurement method only starts from pollutants, does not consider that the detection of the mobile source pollution emission is influenced by actual factors, so that the estimation of the mobile source pollution emission is not accurate and reliable, and the method causes difficulty in the targeted control of each component of the mobile source pollution emission by related departments.
Disclosure of Invention
The invention provides a high-emission mobile source pollution identification method and system based on deep feature clustering, which can solve the technical problems.
In order to achieve the purpose, the invention adopts the following technical scheme:
a high-emission mobile source pollution identification method based on depth feature clustering comprises the following steps,
s10, collecting remote measuring data and vehicle inspection data of the tail gas of the motor vehicle;
s20, preprocessing the collected tail gas data;
s30, evaluating components-CO, HC and NO in the exhaust emission and actual influence factors by adopting a random forest to the preprocessed data, and selecting influence characteristic factors of each polluted gas;
s40, according to the factors influencing the emission concentration of each pollutant gas obtained in the step S30, clustering CO, HC and NO respectively by adopting a clustering algorithm to obtain a class label of the high-emission mobile pollution source;
s50, updating a data set according to the class label of the high-emission mobile pollution source obtained in the step S40, and training through a deep forest algorithm to obtain a classification recognition model of the high-emission mobile pollution source;
and S60, carrying out pollutant identification on the tail gas data by using the trained classification identification model of the high-emission mobile pollution source.
Further, the step S10 is collecting the vehicle exhaust telemetry data and the vehicle inspection data, specifically including,
(11) the data collected from the telemetry system includes: the device number is used for detecting vehicle passing time, license plate number, vehicle color, recognition confidence, vehicle running speed, acceleration, vehicle body length, CO measured concentration, HC measured concentration, NO measured concentration, smoke value, dynamic/static measurement, effective data, passing data, specific power, smoke value, wind speed, wind direction, air temperature, humidity and atmospheric pressure;
(12) the data collected from the vehicle inspection system includes: the number plate number, the maximum quality, the form of a transmission, the number of gears, the fuel specification, the type of a vehicle, the use property, the reference quality, the driving mode, the driving tire air pressure, the type of an engine, an engine manufacturer, the engine discharge capacity, whether a catalytic converter exists or not and an exhaust aftertreatment device.
Further, the step S20 of preprocessing the collected exhaust data specifically includes:
(21) combining different characteristic attributes in the telemetering data and the vehicle inspection data into comprehensive tail gas data information through the license plate number;
(22) finding out the data segment with missing value for discarding, finding out abnormal value for discarding by means of boxplot principle, and deleting irrelevant attribute; after the invalid attribute is deleted, the reference mass, the running speed, the running acceleration, the specific power, the wind speed, the wind direction, the air temperature, the humidity, the atmospheric pressure, the vehicle body length and the service life are remained as relevant external attributes.
Further, the step S30 of evaluating the components of CO, HC and NO in the exhaust emission and the actual influence factors by using the random forest to the preprocessed data, and selecting the influence characteristic factors of each pollutant gas, specifically including,
the input exhaust pollutants CO, HC and NO respectively and the related external attribute characteristics form a characteristic selection set Ai＝{ai0,ai1,ai2,…,ai11I is more than or equal to 1 and less than or equal to 3), wherein ai0Representing a characteristic value of a contaminant, aij(j is more than or equal to 1 and less than or equal to 11) represents the characteristic value of the influence attribute, and the method for selecting the characteristics by using the random forest comprises the following steps:
(31) determining an input sample N and a feature dimension M;
(32) sampling input samples in a put-back mode, randomly sampling the characteristic M, and constructing a decision tree by utilizing a GINI index and adopting a complete splitting mode;
(33) repeating the step (32) to construct N decision trees to form a random forest, and calculating the error of the data outside the bag by using the data outside the bag (OOB) of the nth decision tree (N is more than or equal to 1 and less than or equal to N), and recording the error as En1；
(34) Feature x of all OOB samples of random out-of-bag dataiAdding noise interference value, calculating error of data outside bag again, and recording as En2；
(35) Characteristic xiVIM (importance score)iThe calculation method is as follows:
(36) and calculating the importance scores and the average value alpha (1/M) of the M characteristics, and selecting the characteristics of VIM & gtalpha as specific influence characteristics.
Further, in step S40, according to the factors affecting the emission concentration of each pollutant gas obtained in step S30, clustering CO, HC, and NO respectively by using a clustering algorithm to obtain a category label of the high-emission mobile pollution source, where the clustering algorithm uses a K-means clustering algorithm, and the specific process is as follows:
(41) input sample data set X ═ X1,x2,…,xp,…,xPP.ltoreq.1) where the sample points
Representing a real number set, and d is a dimension of a sample point;
(42) randomly selecting k cluster centers u1,u2,…,uk；
(43) Calculating a sample point xpEuclidean distance d to the center of each clusterp1,dp2,…,dpi,…,dpk(1≤i≤k)；
(44) If d ispiThe smallest value, the sample point xpDivision into cluster centers uiIn range, co-forming k clusters C in the sample1,C2,…,Ck；
(45) In cluster class CiIn the method, the mean value of the sample points is calculated as a new cluster center ui；
(46) Iterations (43) - (45) are repeated until all cluster centers are unchanged.
Further, in the step (44), DBI is used as a measure of clustering effect, wherein DBI is calculated as follows:
where k represents the total number of clusters, avg (C) represents the average of the distances from the sample point to the cluster center point in cluster class C, dcen(ui,uj) Indicating a cluster class uiAnd ujThe distance between the class center points.
Further, in the step S50, the data set is updated according to the class label of the high-emission mobile pollution source obtained in the step S40, and a deep forest algorithm is trained to obtain a classification recognition model of the high-emission mobile pollution source, where the process of the deep forest classification process is as follows:
(51) inputting a data set which comprises an attribute data set and a corresponding category label set;
(52) multi-granularity scanning, namely setting a plurality of sliding windows with different dimensions, scanning the input attribute data set characteristic vectors, and splicing to obtain characteristic vectors with different granularities;
(53) constructing a cascade forest, wherein the cascade forest comprises a plurality of levels of decision tree forests, each level of decision tree forest is composed of a plurality of random forests and a plurality of completely random forests, the random forests and the completely random forests are constructed based on the decision trees, and the classification judgment results of each decision tree in the forests are averaged to serve as the judgment results of the forest;
(54) each forest on the upper layer of the cascading forest can output and generate a d-dimensional identification vector, and the d-dimensional identification vector is connected with the feature vector after the granularity scanning to form the input of the lower layer of the cascading forest;
(55) and (4) repeating the step (54), and stopping the increase of the level along with the continuous deepening of the training level until the accuracy is not improved any more, and outputting a final classification result.
Further, the step (22) further comprises an abnormal value processing step: judging whether the data is abnormal by using a boxplot principle, and discarding an abnormal value; the concrete implementation is as follows:
[1] respectively calculating 25% quantiles (Q1), 50% quantiles (Q2) and 75% quantiles (Q3) of each attribute;
[2] calculating the quartering distance IQR (Q3-Q1);
[3] and calculating the maximum value max which is Q3+1.5 XIQR, the minimum value min which is Q1-1.5 XIQR, and the value which is greater than max or less than min is the abnormal value of the attribute sample.
Further, the step S50 further includes the following formula for calculating the model classification accuracy Acc:
wherein, yiA category label indicating the truth of the ith data,
and (3) representing a model prediction label of the ith data, wherein the function f (a, b) is used for judging whether a and b are equal, if so, the function is 1, and otherwise, the function is 0.
On the other hand, the invention also discloses a high-emission mobile source pollution identification system with depth feature clustering, which comprises the following units,
the data collection unit is used for collecting the motor vehicle tail gas remote measuring data and the vehicle inspection data;
the data processing unit is used for preprocessing the collected tail gas data;
the influence factor determining unit is used for evaluating components-CO, HC and NO in the tail gas emission and actual influence factors by adopting random forests for the preprocessed data and selecting influence characteristic factors of each polluted gas;
the category label determining unit is used for clustering CO, HC and NO respectively by adopting a clustering algorithm according to the obtained factors influencing the emission concentration of each pollutant gas to obtain a category label of the high-emission mobile pollution source;
the classification recognition model training unit is used for updating the data set according to the obtained class label of the high-emission mobile pollution source and obtaining a classification recognition model of the high-emission mobile pollution source through deep forest algorithm training;
and the pollutant identification unit is used for identifying pollutants in the tail gas data by utilizing the trained classification identification model of the high-emission mobile pollution source.
According to the technical scheme, the method for identifying the high-emission mobile-source pollution by deep feature clustering preprocesses collected vehicle annual inspection data and tail gas telemetering data; the method comprises the following steps of utilizing a random forest feature selection algorithm to carry out importance evaluation on external attributes influenced by the concentrations of main components CO, HC and NO in the exhaust gas of a mobile source, and selecting main influence feature factors of various polluted gases; clustering the data after feature selection by using various clustering algorithms to obtain a high-emission motor vehicle label; and training the high-emission class label data by utilizing the deep forest to obtain an automatic classification recognition model.
Compared with other methods, the method comprehensively considers the influence of external actual factors on pollution detection, screens out main influence factors on different tail gas components, and then respectively models and identifies, thereby effectively improving the prediction precision and providing an effective technical method for monitoring and controlling the high-emission mobile pollution source by related departments.
Specifically, conventional motor vehicle exhaust detection classifies the vehicle into high and normal emissions based on a defined threshold of the relevant standards, and classification is not accurate enough. The method fully utilizes a large amount of data accumulated by the telemetering equipment, considers influence factors in the actual environment, further classifies the detected vehicles, constructs a high-emission classification recognition model, and helps related departments to promote monitoring and control of high-emission mobile pollution sources.
Drawings
FIG. 1 is a schematic diagram of the method of the present invention;
FIG. 2 is a histogram of the impact ratios of various pollutant characteristic factors;
fig. 3 is a diagram of a deep forest model structure.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention.
As shown in fig. 1, the method for identifying pollution of a high-emission mobile source by depth feature clustering according to the embodiment includes the steps of random forest feature selection, feature clustering and depth forest classification;
the method comprises the following steps:
(1) collecting motor vehicle exhaust telemetering data and vehicle inspection data;
(2) preprocessing the collected tail gas data;
(3) for the preprocessed data, importance evaluation is carried out on main components, namely CO, HC and NO, in the tail gas emission and actual influence factors by adopting a random forest, and main influence characteristic factors of each polluted gas are selected;
(4) according to the main factors influencing the emission concentration of each pollutant gas obtained in the step (3), clustering CO, HC and NO respectively by adopting a clustering algorithm to obtain class labels of the high-emission mobile pollution sources;
(5) and (4) updating the data set according to the class label of the high-emission mobile pollution source obtained in the step (4), and training through a deep forest algorithm to obtain a classification recognition model of the high-emission mobile pollution source.
Further description is as follows:
in the step (1), the process of extracting the telemetering data of the tail gas of the motor vehicle is as follows:
(11) the data collected from the telemetry system includes: the device number is used for detecting vehicle passing time, license plate number, vehicle color, recognition confidence, vehicle running speed, acceleration, vehicle body length, CO measured concentration, HC measured concentration, NO measured concentration, smoke value, dynamic/static measurement, effective data, passing data, specific power, smoke value, wind speed, wind direction, air temperature, humidity and atmospheric pressure;
(12) the data collected from the vehicle inspection system includes: the number plate number, the maximum quality, the form of a transmission, the number of gears, the fuel specification, the type of a vehicle, the use property, the reference quality, the driving mode, the driving tire air pressure, the type of an engine, an engine manufacturer, the engine discharge capacity, whether a catalytic converter exists or not and an exhaust aftertreatment device.
In the step (2), the motor vehicle tail gas telemetering data and the vehicle inspection data are preprocessed as follows:
(21) combining different characteristic attributes in the telemetering data and the vehicle inspection data into more comprehensive tail gas data information through the license plate number;
(22) and finding out the data segment with the missing value for discarding, finding out the abnormal value by virtue of the boxplot principle for discarding, and deleting the irrelevant attribute. After the invalid attribute is deleted, 11 attributes such as residual reference mass, running speed, running acceleration, specific power, wind speed, wind direction, air temperature, humidity, atmospheric pressure, vehicle body length, service life and the like are related external attributes researched by the invention;
in the step (3), the input exhaust pollutants CO, HC and NO and the relevant external attribute characteristics form a characteristic selection set Ai＝{ai0,ai1,ai2,…,ai11I is more than or equal to 1 and less than or equal to 3), wherein ai0Representing a characteristic value of a contaminant, aij(j is more than or equal to 1 and less than or equal to 11) represents the characteristic value of the influence attribute, and the method for selecting the characteristics by using the random forest comprises the following steps:
(31) determining an input sample N and a feature dimension M;
(32) sampling input samples in a put-back mode, randomly sampling the characteristic M, and constructing a decision tree by utilizing a GINI index and adopting a complete splitting mode;
(33) repeating the step (32) to construct N decision trees to form a random forest, and calculating the error of the data outside the bag by using the data outside the bag (OOB) of the nth decision tree (N is more than or equal to 1 and less than or equal to N), and recording the error as En1；
(34) Feature x of all OOB samples of random out-of-bag dataiAdding noise interference value, calculating error of data outside bag again, and recording as En2；
(35) Characteristic xiVIM (importance score)iThe calculation method is as follows:
(36) and calculating the importance scores and the average value alpha (1/M) of the M characteristics, and selecting the characteristics of VIM & gtalpha as main influence characteristics.
In the step (4), performing cluster analysis on the sample data after the feature selection in the step (3) by using a K-Means clustering algorithm;
the K-means clustering algorithm process is as follows:
(41) input sample data set X ═ X1,x2,…,xp,…,xPP.ltoreq.1) where the sample points
Representing a real number set, and d is a dimension of a sample point;
(42) randomly selecting k cluster centers u1,u2,…,uk；
(43) Calculating a sample point xpEuclidean distance d to the center of each clusterp1,dp2,…,dpi,…,dpk(1≤i≤k)；
(44) If d ispiThe smallest value, the sample point xpDivision into cluster centers uiIn range, co-forming k clusters C in the sample1,C2,…,Ck；
(45) In cluster class CiIn the method, the mean value of the sample points is calculated as a new cluster center ui；
(46) Iterations (43) - (45) are repeated until all cluster centers are unchanged.
In the step (4), the DBI is used as a measurement index of the clustering effect. The DBI is calculated as follows:
where k represents the total number of clusters, avg (C) represents the average of the distances from the sample point to the cluster center point in cluster class C, dcen(ui,uj) Indicating a cluster class uiAnd ujThe distance between class center points of (1);
in the step (5), a classification model of the high-emission mobile source is constructed as follows: and (4) training the class labels obtained in the step (4) on the feature data of the exhaust emission of the detected vehicle by adopting a deep forest classification algorithm to obtain a classification model, and calculating the identification accuracy of the model.
The process of the deep forest classification process is as follows:
(51) inputting a data set which comprises an attribute data set and a corresponding category label set;
(52) multi-granularity scanning. Setting a plurality of sliding windows with different dimensions, scanning the input attribute data set characteristic vectors, and splicing to obtain characteristic vectors with different granularities;
(53) and constructing a cascade forest. The cascaded forests comprise multilevel decision tree forests, each level of decision tree forest is composed of a plurality of random forests and completely random forests, and the random forests and the completely random forests are constructed based on decision trees. Calculating the mean value of the classification judgment result of each decision tree in the forest as the judgment result of the forest;
(54) each forest on the upper layer of the cascading forest can output and generate a d-dimensional identification vector, and the d-dimensional identification vector is connected with the feature vector after the granularity scanning to form the input of the lower layer of the cascading forest;
(55) and (4) repeating the step (54), and stopping the increase of the level along with the continuous deepening of the training level until the accuracy is not improved any more, and outputting a final classification result.
The following are exemplified:
as shown in fig. 1, the present invention is specifically implemented as follows:
(1) collecting motor vehicle tail gas remote measuring data and corresponding vehicle annual inspection data of an urban target road section;
(11) data collected from the telemetry system includes: the device number is used for detecting vehicle passing time, license plate number, vehicle color, recognition confidence, vehicle running speed, acceleration, vehicle body length, CO measured concentration, HC measured concentration, NO measured concentration, smoke value, dynamic/static measurement, effective data, passing data, specific power, smoke value, wind speed, wind direction, air temperature, humidity and atmospheric pressure;
(12) the data obtained from the vehicle inspection includes: the number plate number, the maximum quality transmission form, the gear number, the fuel specification, the vehicle type, the use property, the reference quality, the driving mode, the driving tire pressure, the engine model, the engine manufacturing enterprise, the engine discharge capacity, whether a catalytic converter exists or not and an exhaust aftertreatment device.
(2) And preprocessing the collected data, which mainly comprises the fusion of tail gas remote measurement data and vehicle inspection data, tail gas data cleaning and abnormal value processing.
(21) Data fusion: combining different characteristic attributes of remote sensing monitoring tail gas data records with the same license plate number and vehicle annual inspection data records to form a tail gas data record with more comprehensive information;
(22) cleaning tail gas data: when the confidence coefficient of the license plate recognition is lower than 85%, deleting the invalid tail gas data record;
(23) abnormal value processing: and (4) judging whether the data is abnormal or not by using a boxplot principle, and discarding the abnormal value. The concrete implementation is as follows:
[1] respectively calculating 25% quantiles (Q1), 50% quantiles (Q2) and 75% quantiles (Q3) of each attribute;
[2] calculating the quartering distance IQR (Q3-Q1);
[3] and calculating the maximum value max which is Q3+1.5 XIQR, the minimum value min which is Q1-1.5 XIQR, and the value which is greater than max or less than min is the abnormal value of the attribute sample.
(3) And (4) evaluating the importance of the random forest of the tail gas emission components and the influencing factors.
Based on step 2, the remaining factors relevant to the detection of motor vehicle exhaust: driving conditions (speed, acceleration), service life of the vehicle, wind speed, air temperature, reference mass, specific power, wind direction, humidity, atmospheric pressure, vehicle body length and the like. Merging the attribute data sets into an attribute data set, and regularizing the attribute data set;
calculating importance scores for different characteristic factors by adopting a random forest;
using formulas
Calculating to obtain the influence proportion of each factor on each emission component;
wherein, VIMiIs a characteristic factor xiThe influence on the pollutants, En1Initial out-of-bag data error for the nth decision tree, En2For the nth decision tree random pair feature xiAdding the error of the data outside the bag after noise interference, wherein the specific gravity histogram is influenced by the characteristic factors as shown in figure 2;
the results of the calculation of the ratios of the importance of the influencing factors of the respective pollutants are shown in the following table:
wherein, the attribute that the specific gravity exceeds the average value alpha (namely 1/11, 1/11 ≈ 0.0909) is influenced on a certain discharge pollutant as the main characteristic factor;
(4) performing clustering analysis by adopting a K-Means algorithm according to the main factors influencing the concentration of each discharged pollutant obtained in the step (3);
the K-Means clustering algorithm process is as follows:
(1) input sample data set X ═ X1,x2,…,xp,…,xPP.ltoreq.1) where the sample points
Representing a real number set, and d is a dimension of a sample point;
(2) randomly selecting k cluster centers u1,u2,…,uk；
(3) Calculating a sample point xpEuclidean distance d to the center of each clusterp1,dp2,…,dpi,…,dpk(1≤i≤k)；
(4) If d ispiThe smallest value, the sample point xpDivision into cluster centers uiIn range, co-forming k clusters C in the sample1,C2,…,Ck；
(5) In cluster class CiIn the method, the mean value of the sample points is calculated as a new cluster center ui；
(6) Iterations (3) - (5) are repeated until all cluster centers are unchanged.
DBI was used as a measure of clustering effect. The DBI is calculated as follows:
where k represents the total number of clusters, avg (C) represents the average of the distances from the sample point to the cluster center point in cluster class C, dcen(ui,uj) Indicating a cluster class uiAnd ujThe distance between class center points of (1);
(41) setting different initial cluster numbers k, clustering and calculating corresponding DBI values;
(42) selecting the initial cluster number with the minimum DBI as the optimal cluster number;
analyzing and determining the serial number of the category label of the high-emission mobile source by combining the specific data;
(5) and (4) training the class labels obtained in the step (4) on the data of the exhaust emission characteristics of the detected vehicles by adopting a deep forest classification algorithm to obtain a classification model, and calculating the recognition accuracy of the model, wherein the structure diagram of the deep forest model is shown in fig. 3.
The construction process of the deep forest classification model is as follows:
(1) an input data set comprising an attribute data set S ═ S1X,S2X,…,SnXAnd the corresponding category label set K ═ y1,y2,…,ynWhere n represents the longitudinal dimension of the data set and X represents the transverse dimension of the data set. Setting a division ratio of 7:3, and dividing a data set into a training set and a test set;
(2) setting the number of multi-granularity scanning windows to be 3, and setting the dimensionality of each scanning window to be Wj(j is more than or equal to 1 and less than or equal to 3), and 2X (X-W) is obtained through calculation and splicing of a layer of random forest and a layer of completely random forestj+1) dimensional feature vectors;
(3) setting the number of random forests and the number of complete random forests in the cascade forests to be 2, setting the number of decision trees in each forest to be 500, solving the mean value of the classification judgment results of each decision tree in the 4 forests as the output class vector of each forest, and setting the dimensionality of the output class vector to be d (equal to the clustering cluster number k), namely obtaining 4 d-dimensional class vectors in each level of forest;
(4) 2 (X-W) obtained in the step (2)1+1) dimensional feature vectors are fed into the cascade forests as initial vectors;
(5) level 1A4 d-dimensional class vectors are obtained, which are then compared with 2 (X-W)1Splicing the +1 dimension characteristic vectors to obtain (2 (X-W)1+1) +4d) dimensional feature vector as the next level 1BThe input of (1);
(6) level 1B4 d-dimensional class vectors are obtained, which are then compared with 2 (X-W)2Splicing the +1 dimension characteristic vectors to obtain (2 (X-W)2+1) +4d) dimensional feature vector as the next level 1CThe input of (1);
(7) level 1C4 d-dimensional class vectors are obtained, which are then compared with 2 (X-W)3Splicing the +1 dimension characteristic vectors to obtain (2 (X-W)3+1) +4d) dimensional feature vector as the next level 2AThe input of (1);
(8) and (5) repeating the steps (5) to (7) until the accuracy of the Nth-level forest (N is set as a super-large value) is not improved any more, calculating the average value of the output feature vectors (k dimensions) of the Nth-level forest, and then calculating the maximum value as the final prediction result.
The formula for calculating the model classification accuracy Acc is as follows:
wherein, yiA category label indicating the truth of the ith data,
and (3) representing a model prediction label of the ith data, wherein the function f (a, b) is used for judging whether a and b are equal, if so, the function is 1, and otherwise, the function is 0.
Through calculation, the classification model of the invention has the following results of identifying the high-emission labels of various pollutants with the correct rate:
| contaminants | Accuracy (%) |
| HC | 97.378 |
| CO | 98.719 |
| NO | 98.625 |
。
According to the technical scheme, compared with other methods, the method for identifying the high-emission mobile pollution source by deep feature clustering comprehensively considers the influence of external actual factors on pollution detection, screens out main influence factors on different tail gas components, and then carries out modeling and identification respectively, so that the prediction precision is effectively improved, and an effective technical method is provided for monitoring and controlling the high-emission mobile pollution source by related departments.
On the other hand, the invention also discloses a high-emission mobile source pollution identification system with depth feature clustering, which comprises the following units,
the data collection unit is used for collecting the motor vehicle tail gas remote measuring data and the vehicle inspection data;
the data processing unit is used for preprocessing the collected tail gas data;
the influence factor determining unit is used for evaluating components-CO, HC and NO in the tail gas emission and actual influence factors by adopting random forests for the preprocessed data and selecting influence characteristic factors of each polluted gas;
the category label determining unit is used for clustering CO, HC and NO respectively by adopting a clustering algorithm according to the obtained factors influencing the emission concentration of each pollutant gas to obtain a category label of the high-emission mobile pollution source;
the classification recognition model training unit is used for updating the data set according to the obtained class label of the high-emission mobile pollution source and obtaining a classification recognition model of the high-emission mobile pollution source through deep forest algorithm training;
and the pollutant identification unit is used for identifying pollutants in the tail gas data by utilizing the trained classification identification model of the high-emission mobile pollution source.
In a third aspect, the present invention also discloses a computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the method as described above.
It is understood that the system provided by the embodiment of the present invention corresponds to the method provided by the embodiment of the present invention, and the explanation, the example and the beneficial effects of the related contents can refer to the corresponding parts in the method.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A high-emission mobile source pollution identification method based on depth feature clustering is characterized by comprising the following steps,
s10, collecting remote measuring data and vehicle inspection data of the tail gas of the motor vehicle;
s20, preprocessing the collected tail gas data;
s30, evaluating components-CO, HC and NO in the exhaust emission and actual influence factors by adopting a random forest to the preprocessed data, and selecting influence characteristic factors of each polluted gas;
s40, according to the factors influencing the emission concentration of each pollutant gas obtained in the step S30, clustering CO, HC and NO respectively by adopting a clustering algorithm to obtain a class label of the high-emission mobile pollution source;
s50, updating a data set according to the class label of the high-emission mobile pollution source obtained in the step S40, and training through a deep forest algorithm to obtain a classification recognition model of the high-emission mobile pollution source;
and S60, carrying out pollutant identification on the tail gas data by using the trained classification identification model of the high-emission mobile pollution source.
2. The method for identifying high-emission mobile-source pollution by depth feature clustering according to claim 1, wherein: the S10 collects vehicle exhaust telemetry data and vehicle inspection data, including,
(11) the data collected from the telemetry system includes: the device number is used for detecting vehicle passing time, license plate number, vehicle color, recognition confidence, vehicle running speed, acceleration, vehicle body length, CO measured concentration, HC measured concentration, NO measured concentration, smoke value, dynamic/static measurement, effective data, passing data, specific power, smoke value, wind speed, wind direction, air temperature, humidity and atmospheric pressure;
(12) the data collected from the vehicle inspection system includes: the number plate number, the maximum quality, the form of a transmission, the number of gears, the fuel specification, the type of a vehicle, the use property, the reference quality, the driving mode, the driving tire air pressure, the type of an engine, an engine manufacturer, the engine discharge capacity, whether a catalytic converter exists or not and an exhaust aftertreatment device.
3. The method for identifying high-emission mobile-source pollution by depth feature clustering according to claim 2, wherein: the S20 preprocessing the collected exhaust data, specifically including:
(21) combining different characteristic attributes in the telemetering data and the vehicle inspection data into comprehensive tail gas data information through the license plate number;
(22) finding out the data segment with missing value for discarding, finding out abnormal value for discarding by means of boxplot principle, and deleting irrelevant attribute; after the invalid attribute is deleted, the reference mass, the running speed, the running acceleration, the specific power, the wind speed, the wind direction, the air temperature, the humidity, the atmospheric pressure, the vehicle body length and the service life are remained as relevant external attributes.
4. The method for identifying high-emission mobile-source pollution by depth feature clustering according to claim 3, wherein: and S30, evaluating the components of CO, HC and NO in the tail gas emission and the actual influence factors by using random forests to the preprocessed data, selecting the influence characteristic factors of each pollutant gas, specifically comprising,
the input exhaust pollutants CO, HC and NO respectively and the related external attribute characteristics form a characteristic selection set Ai＝{ai0,ai1,ai2,…,ai11I is more than or equal to 1 and less than or equal to 3), wherein ai0Representing a characteristic value of a contaminant, aij(j is more than or equal to 1 and less than or equal to 11) represents the characteristic value of the influence attribute, and the method for selecting the characteristics by using the random forest comprises the following steps:
(31) determining an input sample N and a feature dimension M;
(32) sampling input samples in a put-back mode, randomly sampling the characteristic M, and constructing a decision tree by utilizing a GINI index and adopting a complete splitting mode;
(33) repeating the step (32) to construct N decision trees to form a random forest, and calculating the error of the data outside the bag by using the data outside the bag (OOB) of the nth decision tree (N is more than or equal to 1 and less than or equal to N), and recording the error as En1；
(34) Feature x of all OOB samples of random out-of-bag dataiAdding noise interference value, calculating error of data outside bag again, and recording as En2；
(35) Characteristic xiVIM (importance score)iThe calculation method is as follows:
(36) and calculating the importance scores and the average value alpha (1/M) of the M characteristics, and selecting the characteristics of VIM & gtalpha as specific influence characteristics.
5. The method of claim 4, wherein the depth feature clustering high-emission mobile-source pollution identification method comprises: and S40, clustering CO, HC and NO respectively by adopting a clustering algorithm according to the factors influencing the emission concentration of each pollutant gas obtained in the step S30 to obtain a class label of the high-emission mobile pollution source, wherein the clustering algorithm adopts a K-means clustering algorithm, and the specific process is as follows:
(41) input sample data set X ═ X1,x2,…,xp,…,xPP.ltoreq.1) where the sample points
Representing a real number set, and d is a dimension of a sample point;
(42) randomly selecting k cluster centers u1,u2,…,uk；
(43) Calculating a sample point xpEuclidean distance d to the center of each clusterp1,dp2,…,dpi,…,dpk(1≤i≤k)；
(44) If d ispiThe smallest value, the sample point xpDivision into cluster centers uiIn range, co-forming k clusters C in the sample1,C2,…,Ck；
(45) In cluster class CiIn the method, the mean value of the sample points is calculated as a new cluster center ui；
(46) Iterations (43) - (45) are repeated until all cluster centers are unchanged.
6. The method of claim 5, wherein the depth feature clustering high-emission moving-source pollution identification method comprises: in the step (44), DBI is used as a measure of clustering effect, wherein DBI is calculated as follows:
where k represents the total number of clusters, avg (C) represents the average of the distances from the sample point to the cluster center point in cluster class C, dcen(ui,uj) Indicating a cluster class uiAnd ujThe distance between the class center points.
7. The method of claim 5, wherein the depth feature clustering high-emission moving-source pollution identification method comprises: and in the step S50, updating the data set according to the class label of the high-emission mobile pollution source obtained in the step S40, and obtaining a classification recognition model of the high-emission mobile pollution source through deep forest algorithm training, wherein the process of the deep forest classification process is as follows:
(51) inputting a data set which comprises an attribute data set and a corresponding category label set;
(52) multi-granularity scanning, namely setting a plurality of sliding windows with different dimensions, scanning the input attribute data set characteristic vectors, and splicing to obtain characteristic vectors with different granularities;
(53) constructing a cascade forest, wherein the cascade forest comprises a plurality of levels of decision tree forests, each level of decision tree forest is composed of a plurality of random forests and a plurality of completely random forests, the random forests and the completely random forests are constructed based on the decision trees, and the classification judgment results of each decision tree in the forests are averaged to serve as the judgment results of the forest;
(54) each forest on the upper layer of the cascading forest can output and generate a d-dimensional identification vector, and the d-dimensional identification vector is connected with the feature vector after the granularity scanning to form the input of the lower layer of the cascading forest;
(55) and (4) repeating the step (54), and stopping the increase of the level along with the continuous deepening of the training level until the accuracy is not improved any more, and outputting a final classification result.
8. The method for identifying high-emission mobile-source pollution by depth feature clustering according to claim 3, wherein: step (22) further comprises outlier processing: judging whether the data is abnormal by using a boxplot principle, and discarding an abnormal value; the concrete implementation is as follows:
[1] respectively calculating 25% quantiles (Q1), 50% quantiles (Q2) and 75% quantiles (Q3) of each attribute;
[2] calculating the quartering distance IQR (Q3-Q1);
[3] and calculating the maximum value max which is Q3+1.5 XIQR, the minimum value min which is Q1-1.5 XIQR, and the value which is greater than max or less than min is the abnormal value of the attribute sample.
9. The method of claim 7, wherein the depth feature clustering high-emission moving-source pollution identification method comprises: the step S50 further includes the following formula for calculating the model classification accuracy Acc:
wherein, yiA category label indicating the truth of the ith data,
and (3) representing a model prediction label of the ith data, wherein the function f (a, b) is used for judging whether a and b are equal, if so, the function is 1, and otherwise, the function is 0.
10. The utility model provides a high emission mobile source pollution identification system of degree of depth feature clustering which characterized in that: comprises the following units of a first unit, a second unit,
the data collection unit is used for collecting the motor vehicle tail gas remote measuring data and the vehicle inspection data;
the data processing unit is used for preprocessing the collected tail gas data;
the influence factor determining unit is used for evaluating components-CO, HC and NO in the tail gas emission and actual influence factors by adopting random forests for the preprocessed data and selecting influence characteristic factors of each polluted gas;
the category label determining unit is used for clustering CO, HC and NO respectively by adopting a clustering algorithm according to the obtained factors influencing the emission concentration of each pollutant gas to obtain a category label of the high-emission mobile pollution source;
the classification recognition model training unit is used for updating the data set according to the obtained class label of the high-emission mobile pollution source and obtaining a classification recognition model of the high-emission mobile pollution source through deep forest algorithm training;
and the pollutant identification unit is used for identifying pollutants in the tail gas data by utilizing the trained classification identification model of the high-emission mobile pollution source.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110534816.2A CN113435471A (en) | 2021-05-17 | 2021-05-17 | Deep feature clustering high-emission mobile source pollution identification method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110534816.2A CN113435471A (en) | 2021-05-17 | 2021-05-17 | Deep feature clustering high-emission mobile source pollution identification method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113435471A true CN113435471A (en) | 2021-09-24 |
Family
ID=77802299
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110534816.2A Pending CN113435471A (en) | 2021-05-17 | 2021-05-17 | Deep feature clustering high-emission mobile source pollution identification method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113435471A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113919235A (en) * | 2021-10-29 | 2022-01-11 | 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) | Method and medium for detecting abnormal emission of mobile source pollution based on LSTM evolution clustering |
| CN114219690A (en) * | 2021-12-17 | 2022-03-22 | 北京邮电大学 | Big data-based construction method of enterprise unorganized emission behavior model |
| CN115081526A (en) * | 2022-06-16 | 2022-09-20 | 中国汽车工程研究院股份有限公司 | Method for identifying and judging emission hazard of motor vehicle |
| CN115238807A (en) * | 2022-07-29 | 2022-10-25 | 中用科技有限公司 | AMC detection method based on artificial intelligence |
| CN115575570A (en) * | 2022-09-22 | 2023-01-06 | 吉林大学 | Method for detecting vehicle tail gas by using miniaturized artificial olfaction system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106682699A (en) * | 2016-12-31 | 2017-05-17 | 中国科学技术大学 | Vehicle exhaust emission characteristic analytic processing method based on clustering analysis |
| CN109858477A (en) * | 2019-02-01 | 2019-06-07 | 厦门大学 | The Raman spectrum analysis method of object is identified in complex environment with depth forest |
| CN109948726A (en) * | 2019-03-28 | 2019-06-28 | 湘潭大学 | A kind of Power Quality Disturbance Classification Method based on depth forest |
| CN111985796A (en) * | 2020-08-07 | 2020-11-24 | 华中科技大学 | Method for predicting concrete structure durability based on random forest and intelligent algorithm |
-
2021
- 2021-05-17 CN CN202110534816.2A patent/CN113435471A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106682699A (en) * | 2016-12-31 | 2017-05-17 | 中国科学技术大学 | Vehicle exhaust emission characteristic analytic processing method based on clustering analysis |
| CN109858477A (en) * | 2019-02-01 | 2019-06-07 | 厦门大学 | The Raman spectrum analysis method of object is identified in complex environment with depth forest |
| CN109948726A (en) * | 2019-03-28 | 2019-06-28 | 湘潭大学 | A kind of Power Quality Disturbance Classification Method based on depth forest |
| CN111985796A (en) * | 2020-08-07 | 2020-11-24 | 华中科技大学 | Method for predicting concrete structure durability based on random forest and intelligent algorithm |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113919235A (en) * | 2021-10-29 | 2022-01-11 | 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) | Method and medium for detecting abnormal emission of mobile source pollution based on LSTM evolution clustering |
| CN114219690A (en) * | 2021-12-17 | 2022-03-22 | 北京邮电大学 | Big data-based construction method of enterprise unorganized emission behavior model |
| CN115081526A (en) * | 2022-06-16 | 2022-09-20 | 中国汽车工程研究院股份有限公司 | Method for identifying and judging emission hazard of motor vehicle |
| CN115238807A (en) * | 2022-07-29 | 2022-10-25 | 中用科技有限公司 | AMC detection method based on artificial intelligence |
| CN115238807B (en) * | 2022-07-29 | 2024-02-27 | 中用科技有限公司 | AMC detection method based on artificial intelligence |
| CN115575570A (en) * | 2022-09-22 | 2023-01-06 | 吉林大学 | Method for detecting vehicle tail gas by using miniaturized artificial olfaction system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113435471A (en) | Deep feature clustering high-emission mobile source pollution identification method and system | |
| WO2020244288A1 (en) | Method and apparatus for evaluating truck driving behaviour based on gps trajectory data | |
| CN107577910B (en) | Vehicle exhaust concentration inversion method based on deep neural network | |
| CN111024898B (en) | Vehicle exhaust concentration standard exceeding judging method based on Catboost model | |
| CN109543874B (en) | Airport air quality prediction method combining meteorological condition influence | |
| CN114563150B (en) | Bridge health online detection module generation method, detection method, tool box and device | |
| CN113919235B (en) | Mobile source pollution abnormal emission detection method and medium based on LSTM evolution clustering | |
| CN112819217A (en) | Method, system and storage medium for predicting main influence factors of mobile source pollution emission | |
| CN116881749B (en) | Pollution site construction monitoring method and system | |
| CN110135466A (en) | A kind of exceeded vehicle judgment method of pollutant emission and system | |
| CN116485020B (en) | Supply chain risk identification early warning method, system and medium based on big data | |
| CN115422747A (en) | Method and device for calculating discharge amount of pollutants in tail gas of motor vehicle | |
| CN115659263A (en) | Ship control behavior risk assessment system and assessment method based on big data | |
| CN114332825B (en) | Road terrain distribution identification method and device based on deep learning and storage medium | |
| CN115565373A (en) | Real-time risk prediction method, device, equipment and medium for highway tunnel accident | |
| CN115659195A (en) | Online atmospheric pollution identification method | |
| Zhang et al. | Structural Damage Identification System Suitable for Old Arch Bridge in Rural Regions: Random Forest Approach. | |
| CN111210085B (en) | Coal mine gas concentration early warning method based on multi-view ensemble learning | |
| CN116952586A (en) | Axle fatigue crack identification method based on 1D-DCNN | |
| CN114814092A (en) | IP index measuring method based on BP neural network | |
| CN114692749B (en) | Processing and analyzing method for remote sensing detection data of motor vehicle tail gas | |
| CN117408520B (en) | Intelligent data service identification method and system | |
| CN117349777B (en) | Intelligent identification system and method for online monitoring data of water environment | |
| CN113222217B (en) | Diesel vehicle emission continuous value prediction method and system based on discrete emission label | |
| KR102523246B1 (en) | Environmental Space Exploration Method |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210924 |
|
| RJ01 | Rejection of invention patent application after publication |