CN114662060A - Vehicle-mounted nitrogen oxide sensor concentration measurement value correction method based on machine learning - Google Patents
Vehicle-mounted nitrogen oxide sensor concentration measurement value correction method based on machine learning Download PDFInfo
- Publication number
- CN114662060A CN114662060A CN202210578002.3A CN202210578002A CN114662060A CN 114662060 A CN114662060 A CN 114662060A CN 202210578002 A CN202210578002 A CN 202210578002A CN 114662060 A CN114662060 A CN 114662060A
- Authority
- CN
- China
- Prior art keywords
- corrected
- measurement data
- time correction
- sliding window
- correction value
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- 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
- G01N33/0006—Calibrating gas analysers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Analytical Chemistry (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Combustion & Propulsion (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Pure & Applied Mathematics (AREA)
- Evolutionary Computation (AREA)
- Mathematical Optimization (AREA)
- Artificial Intelligence (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Computing Systems (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Medical Informatics (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
- Testing Of Engines (AREA)
Abstract
The invention discloses a vehicle-mounted nitrogen oxide sensor concentration measurement value correction method based on machine learning, and relates to the technical field of machine learning. The method comprises the following steps: detecting vehicle exhaust through a vehicle-mounted nitrogen oxide sensor to obtain measurement data to be corrected output by the vehicle-mounted nitrogen oxide sensor; extracting a first statistical characteristic of the measured data to be corrected, and inputting the first statistical characteristic into a classification model to obtain a time correction value output by the classification model; integrally correcting the to-be-corrected measurement data according to the time correction value to obtain preliminarily corrected measurement data; extracting a second statistical characteristic of the preliminarily corrected measurement data; and inputting the preliminarily corrected measurement data and the second statistical characteristics into a regression model to obtain the finally corrected measurement data. The invention can solve the problem of inaccurate measured data of the vehicle-mounted nitrogen oxide sensor.
Description
Technical Field
The invention relates to a machine learning technology, in particular to a method for correcting a concentration measurement value of a vehicle-mounted nitrogen oxide sensor based on machine learning.
Background
Diesel vehicles are a major contributor to automotive NOx and PM emissions. Wherein vehicle NOx emissions are collected in real time by an on-board NOx sensor. With the continuous development of the car networking technology, the collection of large-scale running data of the vehicles is greatly promoted, including real-time vehicle speed, position, pollutant concentration emission and the like. Meanwhile, the continuous improvement of the networking coverage rate of the national six-wheeled motor vehicles makes the estimation and monitoring of the NOx emission possible during the actual running process of the motor vehicles based on the networking data of the motor vehicles of the national six-wheeled motor vehicles.
The NOx sensor is a sensor which is necessary for realizing closed-loop control and a vehicle-mounted OBD diagnosis system by SCR post-treatment of the national overweight diesel engine. The NOx sensor comprises 2 chambers and 3 pumps, and the measurement principle is as follows: part of tail gas passes through a first chamber and a second chamber of the NOx sensor in sequence for deoxidization treatment, and then NOx is subjected to reduction reaction under the catalytic action and is converted into N2And O2O generated in the second chamber is measured by zirconia or the like2Concentration to calculate the NOx concentration in the exhaust gas. The method is characterized in that the price is low, but because the concentration of NOx in the tail gas needs to be indirectly calculated by measuring the concentration of oxygen, the method is limited by the measurement precision of an oxygen sensor, the NOx sensor has dynamic measurement delay, insufficient measurement precision of a low-concentration region and higher stabilityPoor and the like. Therefore, improving the measurement accuracy of the vehicle-mounted NOx sensor is very key to pushing the intelligent network-linked emission monitoring of heavy vehicles.
The invention is provided in view of the above.
Disclosure of Invention
The embodiment of the invention provides a method for correcting a concentration measurement value of a vehicle-mounted nitrogen oxide sensor based on machine learning, which aims to solve the problem of inaccurate measurement data of the vehicle-mounted nitrogen oxide sensor.
The embodiment of the invention provides a vehicle-mounted nitrogen oxide sensor concentration measurement value correction method based on machine learning, which comprises the following steps:
detecting vehicle exhaust through a vehicle-mounted nitrogen oxide sensor to obtain measurement data to be corrected output by the vehicle-mounted nitrogen oxide sensor;
extracting a first statistical characteristic of the measured data to be corrected, and inputting the first statistical characteristic into a classification model to obtain a time correction value output by the classification model;
integrally correcting the to-be-corrected measurement data according to the time correction value to obtain preliminarily corrected measurement data; extracting a second statistical characteristic of the preliminarily corrected measurement data;
and inputting the preliminarily corrected measurement data and the second statistical characteristics into a regression model to obtain final corrected measurement data output by the regression model.
According to the embodiment of the invention, the first statistical characteristic of the measured data to be corrected output by the vehicle-mounted nitrogen oxide sensor is input into the classification model to obtain the time correction value, so that the problem of inaccurate time of the measured data is solved; and performing numerical correction on the measurement data by extracting the second statistical characteristic of the preliminarily corrected measurement data and inputting the second statistical characteristic into the regression model, thereby realizing double correction of time and error of the measurement data and solving the problem of inaccurate measurement data of the vehicle-mounted nitrogen oxide sensor.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flow chart of a method for correcting a concentration measurement of an on-board NOx sensor based on machine learning according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a shifted data sample and a sliding window provided by an embodiment of the invention;
FIG. 3 is a schematic diagram of a classification model and a regression model according to an embodiment of the present invention;
FIG. 4 is a graphical illustration of a result of a correction of on-board NOx sensor concentration measurement data provided by an embodiment of the present invention;
FIG. 5 (a) is a diagram illustrating pre-correction decision coefficients according to an embodiment of the present invention;
fig. 5 (b) is a schematic diagram of the modified decision coefficient according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present invention, it should also be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
The embodiment of the invention provides a vehicle-mounted nitrogen oxide sensor concentration measurement value correction method based on machine learning, and a flow chart of the method is shown in figure 1, so that the method can be suitable for the condition of correcting concentration measurement data acquired by a vehicle-mounted nitrogen oxide sensor. The present embodiment is performed by an electronic device. The method provided by the embodiment of the invention comprises the following operations:
and S110, detecting the vehicle tail gas through the vehicle-mounted nitrogen oxide sensor to obtain the measured data to be corrected output by the vehicle-mounted nitrogen oxide sensor.
Compared with the vehicle-mounted exhaust gas detection equipment (PEMS) adopting the non-dispersive ultraviolet gas analysis technology (NDUV) and the like, the vehicle-mounted nitrogen oxide (NOx) sensor has larger error in detection result and does not have the capability of truly and accurately evaluating the emission level of the vehicle in the actual driving process.
In the embodiment, the collected value of the vehicle-mounted exhaust gas detection device is taken as a reference, and the measured data output by the vehicle-mounted nitrogen oxide sensor has the problem of time advance or time lag and inaccuracy relative to the PEMS collected value, which is called as measured data to be corrected.
And S120, extracting a first statistical characteristic of the measured data to be corrected, and inputting the first statistical characteristic into a classification model to obtain a time correction value output by the classification model.
For convenience of description and distinction, the statistical feature of the measurement data to be corrected is referred to as a first statistical feature. The first statistical features include, but are not limited to, mean, standard deviation, variance, median, and the like. The NOx gas concentration is different, the concentration change rate is different, and the measured data to be corrected is advanced or lagged relative to the reference measured data of the PEMS; therefore, the first statistical feature has a strong correlation with the lead or lag condition and can be learned through the classification model.
Optionally, sliding on the measurement data to be corrected according to a sliding step length through at least one sliding window; and calculating the average value and the standard deviation of the measured data to be corrected in the sliding window after each sliding as a first statistical characteristic. In a specific embodiment, the size of the sliding window is 30s, and the sliding step length is 5 s; after each sliding, the mean and standard deviation of the data within the window are calculated. This results in two sets of data forming a matrix. When the number of the sliding windows is more than 2, and the sizes of the sliding windows are different, the average value and the standard deviation of each sliding window can be formed into a matrix as the first statistical characteristic of each sliding window. In the embodiment, the features are extracted layer by layer through the sliding window, so that the sufficient extraction of the features can be ensured.
And then, inputting the first statistical characteristic of each sliding window into a classification model to obtain a time correction value of each sliding window.
The classification model in this embodiment is a classification algorithm in machine learning, including but not limited to a multi-layered perceptron (MLP), a decision tree, a random forest, a Support Vector Machine (SVM), naive bayes, and the like. The classification model represents a mapping relation between the first statistical characteristic and the time correction value and can be obtained through training.
S130, correcting the measurement data to be corrected according to the time correction value to obtain preliminarily corrected measurement data; and extracting a second statistical characteristic of the preliminarily corrected measurement data.
The time correction value is advanced or retarded relative to the measured data to be corrected.
Optionally, the measurement data to be corrected in each sliding window is corrected in sequence according to the sliding direction; if the measurement data to be corrected in a sliding window is advanced, the time correction value is a negative value, and the measurement data to be corrected is integrally moved backwards by the absolute value of the time correction value to obtain the preliminarily corrected measurement data falling in the sliding window; and if the measured data to be corrected in a sliding window lags behind, the time correction value is a positive value, and the measured data to be corrected is integrally moved forwards by the time correction value to obtain the preliminarily corrected measured data falling in the sliding window.
It should be noted that, since the sliding step is generally smaller than the size of the sliding window, the sliding windows overlap, so that the data at a time has a plurality of time correction values. Through engineering practice, the data in the sliding window can be corrected according to the sliding direction and the time correction value of each sliding window. Thus, the preliminary corrected measurement data of the following sliding window may overwrite the partial preliminary corrected measurement data of the preceding sliding window.
Illustratively, the measured data to be corrected is 1 s-35 s, the size of the sliding window is 30s, and the sliding step length is 5 s. The first sliding window is 1 s-30 s, and the time correction value is 1; the second sliding window is 6 s-35 s, and the time correction value is-1. Then, firstly, moving the whole to-be-corrected measurement data forward for 1s, taking the to-be-corrected measurement data of 2 s-31 s as the preliminarily corrected measurement data in the first sliding window, and occupying the preliminarily corrected positions for 1 s-30 s. And then, moving the whole to-be-corrected measurement data backwards for 1s, taking the 5 s-34 s to-be-corrected measurement data as the preliminarily corrected measurement data in the second sliding window, and occupying the preliminarily corrected 6 s-35 s. The 25 preliminary modified data after the first sliding window are covered by the 25 preliminary modified data after the second sliding window.
For convenience of description and distinction, the statistical characteristic of the measurement data after the preliminary correction is referred to as a second statistical characteristic. The second statistical features include, but are not limited to, mean, standard deviation, variance, median, and the like. After correcting the time of the measurement data, the NOx gas concentration and the rate of change in the concentration also affect the error of the measurement data after the initial correction. Therefore, the second statistical characteristic has a strong correlation with the error.
Optionally, sliding on the preliminarily corrected measurement data according to a sliding step length through at least one sliding window; and calculating the average value and the standard deviation of the preliminarily corrected measurement data in the sliding window after each sliding as a second statistical characteristic. In a specific embodiment, the size of the sliding window is 10s and 5s, and the sliding step size is 1 s. After each sliding, the mean and standard deviation of the data within the window are calculated. Two sets of data are thus obtained to form a matrix as the second statistical feature. Illustratively, the average value and the standard deviation of a 5s sliding window of the initially corrected measurement data and the average value and the standard deviation of a 10s sliding window of the initially corrected measurement data are finally obtained. In the embodiment, the features are extracted layer by layer through the sliding window, so that the sufficient extraction of the features can be ensured.
And S140, inputting the preliminarily corrected measurement data and the second statistical characteristics into a regression model to obtain final corrected measurement data output by the regression model.
The present embodiment does not limit the type of regression model, and may be a random forest. In the embodiment, the measurement data needs to be corrected on the basis of the preliminarily corrected measurement data, the second statistical characteristic influences the magnitude of the correction error, and the two types of parameters have a strong association relation with the finally corrected measurement data and can be learned through a regression model. The regression model represents a mapping relationship between the preliminarily corrected measurement data and the second statistical characteristic and the finally corrected measurement data, and can be obtained through training.
According to the embodiment of the invention, the first statistical characteristic of the measured data to be corrected output by the vehicle-mounted nitrogen oxide sensor is input into the classification model to obtain the time correction value, so that the problem of inaccurate time of the measured data is solved; and performing numerical correction on the measurement data by extracting the second statistical characteristic of the preliminarily corrected measurement data and inputting the second statistical characteristic into the regression model, thereby realizing double correction of time and error of the measurement data and solving the problem of inaccurate measurement data of the vehicle-mounted nitrogen oxide sensor.
In this embodiment, the classification model is trained first, and then the regression model is trained. The following describes the training process of the classification model in detail, and before S120, the method further includes:
the first step is as follows: and constructing a training set, wherein the training set comprises a plurality of groups of first statistical characteristics of the measured data samples to be corrected and reference measured data samples, and the reference measured data samples are acquired by vehicle-mounted tail gas detection equipment PEMS.
In a specific embodiment, the concentration of the vehicle exhaust is synchronously detected by the vehicle-mounted nitrogen oxide sensor and the PEMS, the measurement data and the abnormal data before the dew point of the vehicle-mounted nitrogen oxide sensor is released are deleted, and the detection data of the vehicle-mounted nitrogen oxide sensor and the detection data of the PEMS are aligned in time. And carrying out time period segmentation on the alignment data to obtain a plurality of groups of measurement data samples to be corrected and reference measurement data samples. Furthermore, a plurality of groups of measurement data samples to be corrected and reference measurement data samples are divided into a training set and a verification set. The verification set is used for verifying the effect of the trained classification model.
Sliding on each set of measured data samples to be corrected by at least one sliding window in sliding steps, see description at S120; and calculating the average value and the standard deviation of the measured data sample to be corrected in the sliding window after each sliding as a first statistical characteristic.
The second step is that: and determining the time correction value label of each group of measurement data samples to be corrected.
Firstly, each group of measured data samples to be corrected is moved forward or backward by different time correction values, and a plurality of groups of moved data samples corresponding to each group of measured data samples to be corrected are obtained. According to experience, the different time correction values comprise integers between-5 and 5. Negative values are backward movement and positive values are forward movement. For a set of measured data samples to be corrected, 11 sets of shifted data samples can be obtained. The reference measurement data sample does not move.
And then, according to the similarity degree of the reference measurement data sample and each group of moved data samples, determining the time correction value label of each group of measurement data samples to be corrected.
In this embodiment, the calculation method of the similarity between two sets of data is not limited, and may be euclidean distance or cosine similarity. Preferably, the present embodiment uses a sliding window in combination with the cross-correlation coefficient, so that the similarity between two sets of data can be accurately calculated.
And respectively and synchronously sliding on the reference measurement data sample and each group of moved data samples according to the sliding step length through a sliding window. And calculating the cross-correlation coefficient of the reference measurement data sample in a sliding window and each group of moved data samples.
Illustratively, the size of the sliding window is 30s, the sliding step size is 5s, and fig. 2 is a schematic diagram of the shifted data sample and the sliding window provided by the embodiment of the present invention. In the figure: s201 is the original measurement data of the vehicle-mounted nitrogen oxide sensor, S202 is the integral forward movement for t seconds of the original measurement data of the vehicle-mounted nitrogen oxide sensor, S203 is the integral backward movement for t seconds of the original measurement data of the vehicle-mounted nitrogen oxide sensor, and S204 is the 30-second sliding window calculation for the processed data with different movement times.
After each sliding, calculating the sliding windowiCross correlation coefficient of inner reference measurement data sample and each group of moved data samples
：
s (x + t) is a measurement data sample after moving for t seconds; p is a reference measurement data sample; n is the data length of s and p, and x is the number of data samples. t being positive means moving forward and t being negative means moving backward. Next, the time correction value corresponding to the maximum value of the cross-correlation coefficient calculated by each sliding window on each set of shifted data samples (i.e. different t) is used as the time correction value label of the sliding window. Thus, each sliding window can be tagged with a time correction value.
The third step: and training the classification model by adopting the training set and the time correction value label.
Training a plurality of classification models by adopting the training set and the time correction value labels, and verifying the accuracy of each classification model (on a verification set); and selecting the classification model with the optimal accuracy from the multiple classification models. Preferably, the classification model is a decision tree model.
And after the classification model training is finished, correcting the corresponding measured data samples to be corrected according to the time correction value labels to obtain each group of preliminarily corrected measured data samples. For a specific correction method, reference is made to the description at S130, which is not described herein again. Then, second statistical characteristics of each group of preliminarily corrected measurement data samples are extracted, wherein the second statistical characteristics comprise the average value and the standard deviation of a 5s sliding window of each group of preliminarily corrected measurement data, and the average value and the standard deviation of a 10s sliding window of each group of preliminarily corrected measurement data.
Constructing a training set which comprises second statistical characteristics of a plurality of groups of preliminarily corrected measurement data samples and a plurality of groups of preliminarily corrected measurement data samples; determining a reference measurement data label of each group of the samples; and training a regression model by using the training set and the reference measurement data labels. Fig. 3 is a schematic training diagram of a classification model and a regression model according to an embodiment of the present invention. Firstly, a 30s sliding window average value and a 30s sliding window standard deviation are obtained according to a measurement data sample to be corrected, and a classification model is trained on the basis of the time correction value label obtained in the second step. And after the training of the classification model is finished, correcting the corresponding measured data sample to be corrected according to the time correction value label to obtain the preliminarily corrected measured data sample. And obtaining the average value and the standard deviation of a 10s sliding window and the average value and the standard deviation of a 5s sliding window on the basis of the preliminarily corrected measurement data sample, training a regression model according to a standard measurement data label obtained by PEMS, and outputting corrected measurement data.
Fig. 4 is a schematic diagram of a result of correcting concentration measurement data of an on-vehicle nox sensor according to an embodiment of the present invention. The difference between the original measured value of the NOx sensor and the measured value of the PEMS is large, and the corrected original measured value of the NOx sensor almost coincides with the measured value of the PEMS. Fig. 5 (a) is a schematic diagram of a pre-correction decision coefficient provided in an embodiment of the present invention, and fig. 5 (b) is a schematic diagram of a post-correction decision coefficient provided in an embodiment of the present invention. By comparison, the accuracy of the measurement result of the vehicle-mounted nitrogen oxide sensor is greatly improved after the intelligent correction method is adopted, the fitting degree of the corrected result and the PEMS measurement value is good, and the determining coefficient R2 of the corrected result and the PEMS measurement value is also obviously increased.
The correction method provided by the invention is based on a machine learning technology, and adopts a classification and regression fusion algorithm to correct the measured data of the vehicle-mounted nitrogen oxide sensor in real time, so that the prediction result is more accurate and stable, the correction effect can be further improved along with the increase of the training data amount, and the correction method has certain innovativeness and application prospect.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for correcting a concentration measurement value of a vehicle-mounted nitrogen oxide sensor based on machine learning is characterized by comprising the following steps:
detecting vehicle exhaust through a vehicle-mounted nitrogen oxide sensor to obtain measurement data to be corrected output by the vehicle-mounted nitrogen oxide sensor;
extracting a first statistical characteristic of the measured data to be corrected, and inputting the first statistical characteristic into a classification model to obtain a time correction value output by the classification model;
correcting the measurement data to be corrected according to the time correction value to obtain preliminarily corrected measurement data; extracting a second statistical characteristic of the preliminarily corrected measurement data;
and inputting the preliminarily corrected measurement data and the second statistical characteristics into a regression model to obtain final corrected measurement data output by the regression model.
2. The method according to claim 1, wherein the extracting the first statistical feature of the measurement data to be corrected comprises:
sliding on the measured data to be corrected according to the sliding step length through at least one sliding window;
and calculating the average value and the standard deviation of the measured data to be corrected in the sliding window after each sliding as the first statistical characteristic of each sliding window.
3. The method of claim 2, wherein inputting the first statistical feature into a classification model to obtain a time correction value output by the classification model comprises:
inputting the first statistical characteristics of each sliding window into a classification model to obtain a time correction value of each sliding window;
the correcting the measurement data to be corrected according to the time correction value to obtain the preliminarily corrected measurement data comprises the following steps:
correcting the measured data to be corrected in each sliding window in sequence according to the sliding direction;
if the time correction value of a sliding window is a positive value, moving the whole to-be-corrected measurement data forward by the time correction value to obtain preliminarily corrected measurement data falling in the sliding window;
and if the time correction value of a sliding window is a negative value, moving the whole measured data to be corrected backwards by the absolute value of the time correction value to obtain the preliminarily corrected measured data falling in the sliding window.
4. The method of claim 1, wherein extracting second statistical features of the preliminary corrected measurement data comprises:
sliding on the preliminarily corrected measurement data according to a sliding step length through at least one sliding window;
and calculating the average value and the standard deviation of the preliminarily corrected measurement data in the sliding window after each sliding as second statistical characteristics.
5. The method of claim 4, wherein the sliding window is sized for 10s and 5 s.
6. The method according to any one of claims 1 to 5, wherein before extracting the first statistical feature of the measurement data to be corrected and inputting the first statistical feature into the classification model, further comprising:
constructing a training set; the training set comprises a plurality of groups of first statistical characteristics of measured data samples to be corrected and reference measured data samples;
determining a time correction value label of each group of measurement data samples to be corrected;
and training the classification model by adopting the training set and the time correction value label.
7. The method of claim 6, wherein determining the time correction value label for each set of measured data samples to be corrected comprises:
moving each group of measured data samples to be corrected forward or backward by different time correction values to obtain a plurality of groups of moved data samples corresponding to each group of measured data samples to be corrected;
and determining the time correction value label of each group of the measurement data samples to be corrected according to the similarity degree of the reference measurement data samples and each group of the moved data samples.
8. The method of claim 7, wherein determining the time correction value label for each set of the to-be-corrected measurement data samples based on how similar the baseline measurement data samples are to each set of the shifted data samples comprises:
synchronously sliding on the reference measurement data sample and each group of moved data samples respectively according to the sliding step length through a sliding window;
calculating the cross-correlation coefficient between the reference measurement data sample in a sliding window and each group of moved data samples;
and taking the time correction value corresponding to the maximum value of the cross-correlation coefficient calculated on each group of the moved data samples by each sliding window as the time correction value label of the sliding window.
9. The method of claim 6, wherein training the classification model using the training set and temporal modifier labels comprises:
training various classification models by adopting the training set and the time correction value labels, and verifying the accuracy of each classification model;
and selecting the classification model with the optimal accuracy from the multiple classification models.
10. The method of claim 6, wherein the baseline measurement data sample is collected by a vehicle-mounted emission detection system (PEMS).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210578002.3A CN114662060B (en) | 2022-05-26 | 2022-05-26 | Vehicle-mounted nitrogen oxide sensor concentration measurement value correction method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210578002.3A CN114662060B (en) | 2022-05-26 | 2022-05-26 | Vehicle-mounted nitrogen oxide sensor concentration measurement value correction method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114662060A true CN114662060A (en) | 2022-06-24 |
| CN114662060B CN114662060B (en) | 2022-08-16 |
Family
ID=82038330
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210578002.3A Active CN114662060B (en) | 2022-05-26 | 2022-05-26 | Vehicle-mounted nitrogen oxide sensor concentration measurement value correction method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114662060B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117235453A (en) * | 2023-11-09 | 2023-12-15 | 天津大学 | Remote monitoring data restoration method and system for heavy diesel vehicle based on machine learning |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050192727A1 (en) * | 1994-05-09 | 2005-09-01 | Automotive Technologies International Inc. | Sensor Assemblies |
| CN104076295A (en) * | 2014-07-22 | 2014-10-01 | 重庆大学 | Method for real-time monitoring air pressure signals of new energy fuel cell electromobile |
| CN108197731A (en) * | 2017-12-26 | 2018-06-22 | 中国科学技术大学 | It is a kind of based on jointly trained telemetering motor vehicle tail and car test result coherence method |
| CN109901512A (en) * | 2019-01-08 | 2019-06-18 | 韶能集团韶关宏大齿轮有限公司 | One kind being based on the standardized turning hour norm method of machined parameters |
| US20200222010A1 (en) * | 2016-04-22 | 2020-07-16 | Newton Howard | System and method for deep mind analysis |
| US10783433B1 (en) * | 2019-04-22 | 2020-09-22 | Bell Integrator Inc. | Method for training and self-organization of a neural network |
| US20200405204A1 (en) * | 2016-04-22 | 2020-12-31 | Newton Howard | Detection of catecholamine levels inside the neurocranium |
| US20210138249A1 (en) * | 2016-04-22 | 2021-05-13 | Newton Howard | System and method for neural stimulation using spike frequency modulation |
| CN113538401A (en) * | 2021-07-29 | 2021-10-22 | 燕山大学 | Crowd counting method and system combining cross-modal information in complex scene |
-
2022
- 2022-05-26 CN CN202210578002.3A patent/CN114662060B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050192727A1 (en) * | 1994-05-09 | 2005-09-01 | Automotive Technologies International Inc. | Sensor Assemblies |
| CN104076295A (en) * | 2014-07-22 | 2014-10-01 | 重庆大学 | Method for real-time monitoring air pressure signals of new energy fuel cell electromobile |
| US20200222010A1 (en) * | 2016-04-22 | 2020-07-16 | Newton Howard | System and method for deep mind analysis |
| US20200405204A1 (en) * | 2016-04-22 | 2020-12-31 | Newton Howard | Detection of catecholamine levels inside the neurocranium |
| US20210138249A1 (en) * | 2016-04-22 | 2021-05-13 | Newton Howard | System and method for neural stimulation using spike frequency modulation |
| CN108197731A (en) * | 2017-12-26 | 2018-06-22 | 中国科学技术大学 | It is a kind of based on jointly trained telemetering motor vehicle tail and car test result coherence method |
| CN109901512A (en) * | 2019-01-08 | 2019-06-18 | 韶能集团韶关宏大齿轮有限公司 | One kind being based on the standardized turning hour norm method of machined parameters |
| US10783433B1 (en) * | 2019-04-22 | 2020-09-22 | Bell Integrator Inc. | Method for training and self-organization of a neural network |
| CN113538401A (en) * | 2021-07-29 | 2021-10-22 | 燕山大学 | Crowd counting method and system combining cross-modal information in complex scene |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117235453A (en) * | 2023-11-09 | 2023-12-15 | 天津大学 | Remote monitoring data restoration method and system for heavy diesel vehicle based on machine learning |
| CN117235453B (en) * | 2023-11-09 | 2024-01-26 | 天津大学 | Remote monitoring data restoration method and system for heavy diesel vehicle based on machine learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114662060B (en) | 2022-08-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109285348B (en) | Vehicle behavior identification method and system based on unmanned aerial vehicle and long-and-short memory network | |
| CN111832225A (en) | Method for constructing driving condition of automobile | |
| CN114662060B (en) | Vehicle-mounted nitrogen oxide sensor concentration measurement value correction method based on machine learning | |
| CN106218329A (en) | Vehicle parameter acquisition method based on wireless senser | |
| US20140012791A1 (en) | Systems and methods for sensor error detection and compensation | |
| CN110823585B (en) | Method for acquiring NOx emission factor in heavy vehicle tail gas based on OBD remote emission monitoring data | |
| CN108981781A (en) | For analyzing and detecting the hypothesis analysis system and method for machine sensor fault | |
| CN108197731B (en) | Motor vehicle exhaust remote measurement and vehicle inspection result consistency method based on co-training | |
| CN106143012A (en) | Vehicle part state real-time processing method | |
| CN108445866A (en) | LDW based on convolutional neural networks accidentally fails to report test method and test system | |
| CN113700558A (en) | Diesel engine air system fault detection method | |
| CN110827444B (en) | Heavy vehicle emission factor obtaining method suitable for OBD remote emission monitoring data | |
| CN110838192B (en) | Correction method of OBD remote online monitoring data | |
| CN114291081B (en) | Vehicle collision detection method based on artificial intelligence algorithm | |
| CN112486962A (en) | Extraction and combination short segment calculation heavy-duty diesel vehicle NOxMethod of discharging | |
| CN113436190B (en) | Lane line quality calculation method and device based on lane line curve coefficient and automobile | |
| CN110081890B (en) | Dynamic k nearest neighbor map matching method combined with deep network | |
| CN114060132B (en) | NO based on emission remote monitoring x Sensor cheating discrimination method | |
| CN114706704A (en) | Processing method of abnormal driving and transportation data | |
| CN116992247B (en) | Abnormal data detection method of tail gas analyzer | |
| CN117271992B (en) | Urban road carbon emission monitoring management system based on big data | |
| CN117708625B (en) | Dam monitoring historical data filling method under spent data background | |
| CN113158415B (en) | Vehicle track similarity evaluation method based on error analysis | |
| CN117588291B (en) | Online monitoring method for urea quality | |
| CN115796720A (en) | Heavy vehicle emission evaluation method and storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |