CN118010935A - Method and device for sniffing automobiles - Google Patents
Method and device for sniffing automobiles Download PDFInfo
- Publication number
- CN118010935A CN118010935A CN202410271115.8A CN202410271115A CN118010935A CN 118010935 A CN118010935 A CN 118010935A CN 202410271115 A CN202410271115 A CN 202410271115A CN 118010935 A CN118010935 A CN 118010935A
- Authority
- CN
- China
- Prior art keywords
- gas
- dictionary
- matrix
- sensor
- detail
- 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 27
- 230000004044 response Effects 0.000 claims abstract description 52
- 230000007613 environmental effect Effects 0.000 claims abstract description 29
- 238000007781 pre-processing Methods 0.000 claims abstract description 13
- 230000003044 adaptive effect Effects 0.000 claims abstract description 7
- 238000010276 construction Methods 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 46
- 239000013598 vector Substances 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 10
- 238000012937 correction Methods 0.000 claims description 7
- 238000003745 diagnosis Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 238000004891 communication Methods 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 238000012804 iterative process Methods 0.000 claims description 3
- 238000012417 linear regression Methods 0.000 claims description 3
- 230000001737 promoting effect Effects 0.000 claims description 3
- 230000002441 reversible effect Effects 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 abstract description 11
- 238000001514 detection method Methods 0.000 abstract description 5
- 239000007789 gas Substances 0.000 description 61
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 229910002091 carbon monoxide Inorganic materials 0.000 description 3
- 238000012805 post-processing Methods 0.000 description 3
- 238000013480 data collection Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 239000012855 volatile organic compound Substances 0.000 description 2
- 239000004215 Carbon black (E152) Substances 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000005034 decoration Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 229930195733 hydrocarbon Natural products 0.000 description 1
- 150000002430 hydrocarbons Chemical class 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000008786 sensory perception of smell Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Investigating Or Analyzing Materials By The Use Of Fluid Adsorption Or Reactions (AREA)
Abstract
The invention discloses a method and a device for automobile sniffing, and relates to the technical field of automobile gas sniffing detection, wherein the sniffing method comprises pretreatment, sparse dictionary construction, gas identification and post-treatment; capturing the multidimensional characteristics of the exhaust gas of the automobile by using a plurality of sensors of different types, recording environmental parameters at the same time, and preprocessing to obtain a data set; using the response mode of the known gas as an initial dictionary atom, adding the response mode of the new gas into the dictionary through an online learning algorithm along with the collection of new data, and finding the optimal sparse representation in the adaptive dictionary for the new sensor response; identifying the most relevant dictionary atoms according to the result of sparse coding; based on the identified gas components and the response intensity of the sensor, the concentration of each gas is estimated, and the odor intensity of the detected gas is analyzed and identified by combining the automobile emission standard and the identified gas concentration, so that the problem of interference caused by non-monitoring gas in a complex gas environment can be better processed.
Description
Technical Field
The invention relates to the technical field of automobile gas sniffing detection, in particular to an automobile sniffing method and an automobile sniffing device.
Background
The artificial olfactory system comprises a data acquisition module and a calculation unit module, wherein the calculation unit converts an instrument response value and a predicted value, and finally converts the electric signal strength into an observed value with practical physical meaning after fitting calculation is completed, so that an instrument detection task is completed.
In the field of automobile detection, conventionally, the monitoring of the internal environment of an automobile mainly depends on manual sampling and detection modes, the mode has the defects of low efficiency, high cost, untimely data acquisition and the like, along with the development of the Internet of things and sensing technology, an intelligent monitoring system is increasingly introduced in modern automobile production, an in-automobile odor intensity sniffing instrument used by the intelligent monitoring system is specially used for the internal environment of the automobile, detected gas is converted into an electric signal through a gas sensor and is processed, the odor intensity of the detected gas is analyzed and identified, and real-time monitoring and data acquisition of the internal environment of the automobile are realized.
In the actual application process, the sensor or the sensor array in the sniffing instrument is influenced by the ambient temperature and humidity, the interference gas components in the environment and the running drift of the sensor array, so that nonlinear deviation of a response value and a true value is caused, the interference of abnormal gas is reduced to the greatest extent on a data collection side for the interference of non-monitoring gas, but in the actual application environment, the types and the contents of the gas have high uncertainty, and the application cost is increased by adopting the sensor array with stronger anti-interference capability; on the other hand, from the viewpoint of actual gas distribution environment, the gas content in the environment is relevant, taking NMHC gas as an example, which is a mixed hydrocarbon gas generated in the organic chemical reaction process, when organic matters are generated, the contents of CO, CH 4 and the like in the environment are changed, and finally the dynamic balance in the air is achieved, so that the complex gas environment is a working normal state of an artificial olfaction system.
For this purpose, a method and a device for sniffing automobiles are provided to solve the above problems.
Disclosure of Invention
The invention aims to solve the problem that the automobile sniffing is difficult to face a complex gas environment in the prior art, and provides an automobile sniffing method and an automobile sniffing device.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
An automobile sniffing method comprises preprocessing, sparse dictionary construction, gas recognition and post-processing;
Capturing multidimensional characteristics of gas generated by an automobile by using a plurality of sensors of different types, recording environmental parameters at the same time, correcting the influence of the environmental parameters on the response of the sensors, and preprocessing to obtain a data set; using the response mode of the known gas as an initial dictionary atom, adding the response mode of the new gas into the dictionary through an online learning algorithm along with the collection of new data, and finding the optimal sparse representation in the adaptive dictionary for the new sensor response; identifying the most relevant dictionary atoms according to the result of sparse coding; based on the identified gas components and the sensor response intensity, the concentration of each gas is estimated and used for in-vehicle environment analysis, and the odor intensity of the detected gas is analyzed and identified.
Further, the preprocessing step includes removing noise and normalizing operation, wherein the wavelet transform decomposes the signal into components of different frequencies, the high frequency components represent noise, the sensor response matrix is decomposed by utilizing the two-dimensional wavelet transform, and the matrix is decomposed into a low frequency and a high frequency part based on the two-dimensional discrete wavelet transform;
a: filtering in the horizontal and vertical directions is achieved by a convolution operation, wherein a low-pass filter h and a high-pass filter g are used to extract approximation and detail information;
ApproxLL＝X*h*hT
DetailLH＝X*h*gT
DetailHL＝X*g*hT
DetailHH＝X*g*gT
Where, represents convolution operation, h T and g T represent transposes of the filter, respectively;
b: after each Approx and Detail is obtained, it is downsampled to reduce the size of the matrix;
ApproxLL＝dw(ApproxLL)
DetailLH＝dw(DetailLH)
DetailHL＝dw(DetailHL)
DetailHH＝dw(DetailHH)
Where dw () represents the half of the sample points removed for each row and column;
c: step A and step B are repeated step by step, the low-frequency sub-band is continuously decomposed to form a multi-layer decomposition structure, and approximation and detail information on different scales are obtained;
During wavelet transformation, noise is concentrated in the high frequency detail part, and the noise is removed by setting the wavelet coefficient threshold of the high frequency detail part to zero.
Further, the environmental parameters are corrected, the sensor response is corrected with a linear regression model, the model being expressed as:
Xcalibrated＝Xnormalized-W·E
w is a correction coefficient matrix, the objective of which is to correct the sensor response by the environmental parameters, estimated using the least squares method, i.e. minimizing the following loss function:
Wherein, F represents the Frobenius norm, namely, there is a loss function:
Calculating an optimal solution:
In the process, the matrix E T E needs to be kept as a reversible matrix, otherwise, the matrix E T E needs to be regularized, and an analytical solution of the correction coefficient matrix W is obtained by solving the equation;
W＝(ETE)-1ETXcalibrated
further, the sensor response was corrected with a nonlinear regression model, the RBF model used being expressed as:
Where J is the number of RBFs, representing the complexity of the model, w j is the weight of the jth RBF, RBF (E, c j,βj) is used to represent the output of the jth RBF, c j is the center, and β j is the width parameter;
estimating model parameters with a minimized loss function;
Further, constructing the adaptive sparse coding dictionary includes:
S1, selecting a group of initial dictionary atoms, using a response mode of known gas as the initial dictionary atoms, wherein each vector represents a response mode of gas, and forming the atoms into an initial dictionary matrix D 0;
S2, generating a new sensor response sample X new for a newly collected gas response denoted as X new;
The new sensor response sample X new is brought in, and the dictionary is updated through an online learning algorithm:
Wherein η t represents a learning rate for controlling a step size of dictionary updating;
For X new, performing coefficient coding, λ (λ > 0) is a regularization parameter used for balancing reconstruction errors and sparsity, obtaining a sparse representation matrix α t used for updating a dictionary, |·|| 2 representing an L2 norm (i.e., euclidean norm) used for balancing reconstruction errors; and 1 represents an L1 norm for promoting sparsity.
Further, the iterative process is as follows:
For each sample X i:
A: updating by using the least square form of the LASSO problem, initializing alpha t to be a zero vector matrix, or using the sparse representation of the former step as an initial value for alpha t;
B: soft thresholding alpha t, i.e., thresholding each component alpha t, i;
αt,i＝sign(αt,i)·max(|αt,i|-λ,0)
C: updating a residual r t;
rt＝Xnew-Dtαt
Updating the dictionary matrix using residual r t;
Judging whether a stopping condition is met, ending iteration if the stopping condition is met, otherwise, returning to continue iteration; the sparse representation coefficients α t and the dictionary matrix D 0 are continuously updated until the optimal solution is converged or a predetermined number of iterations is reached, and the optimal judgment standard may be one or more of judgment conditions based on a residual threshold, a sparse representation matrix change threshold, the number of iterations, and the like.
An automobile sniffing device comprises a sensor array, an environmental parameter sensor, data acquisition equipment, a computing unit, a user interface, storage equipment and a communication module;
The sensor array is used for capturing multidimensional characteristics of gas in the vehicle, the gas sensor comprises sensors for detecting various gas components, and the environmental parameter sensor is used for recording environmental parameters; the data acquisition equipment is used for receiving and processing the data of the sensor array and the environmental parameter sensor, and preprocessing and feature extraction are carried out; the computing unit is used for executing an algorithm and comprises computing steps of sparse representation, gas identification, concentration estimation, fault diagnosis and the like; the user interface is used for displaying the identification result, concentration estimation and fault diagnosis information and providing suggestions for decision support; the storage device is used for storing the sensor data, the model parameters and the history record so as to perform performance evaluation and model optimization; the communication module is used for carrying out data exchange and remote control with other systems or cloud.
Compared with the prior art, the invention has the following advantages:
The multi-sensor array is used for capturing multi-dimensional characteristics of gas in the vehicle, so that gas components can be analyzed more comprehensively, and the sensor response is corrected by combining data recorded by the environmental parameter sensor, so that the accuracy and stability of gas identification are improved.
By constructing the self-adaptive sparse coding dictionary, the problem of non-monitoring gas interference in a complex gas environment can be better processed, and the accuracy and the robustness of gas identification are improved. The comprehensive performance and practicality of the system are improved by comprehensively considering the factors such as sensor response, environmental parameters, sparse representation and gas identification, the gas components can be identified, the concentration of each gas can be estimated, the odor intensity of the detected gas can be analyzed, decision support information such as maintenance advice, emission strategy optimization and the like is provided, and a user is helped to manage and maintain the gas environment in the vehicle better.
Drawings
Fig. 1 is a flowchart of a method for sniffing an automobile according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments.
An automobile sniffing device comprises a sensor array, an environmental parameter sensor, data acquisition equipment, a computing unit, a user interface, storage equipment and a communication module;
The sensor array is used for capturing multidimensional characteristics of gas in the automobile, is mainly used for monitoring the air environment in the automobile in the production and delivery processes of the automobile, and the sensor comprises sensors capable of detecting various gas components, such as Volatile Organic Compounds (VOC), oxynitride (NO x), carbon monoxide (CO) and the like; the environmental parameter sensor is used for recording environmental parameters such as temperature, humidity, air pressure and the like so as to correct the influence of sensor response; the data acquisition equipment is used for receiving and processing the data of the sensor array and the environmental parameter sensor, and preprocessing and feature extraction are carried out; the computing unit is used for executing an algorithm and comprises computing steps of sparse representation, gas identification, concentration estimation, fault diagnosis and the like; the user interface is used for displaying the identification result, concentration estimation and fault diagnosis information and providing suggestions for decision support; the storage device is used for storing the sensor data, the model parameters and the history record so as to perform performance evaluation and model optimization; the communication module is used for carrying out data exchange and remote control with other systems or cloud.
The device can more accurately reflect the real content of various gases in the internal environment of the automobile when monitoring the gases in the automobile, wherein the sensor array can better adapt to complex gas environments, including the high uncertainty of the types and the contents of the gases and the correlation between the contents of the gases, so that the device can more effectively cope with the actual application environment; the sensor array and the data processing method are used for coping with nonlinear deviation of environmental factors on sensor response, so that the monitoring precision and stability of the sensor array are improved, and the sensor array is better suitable for complex gas environments.
Referring to fig. 1, an automobile sniffing method includes preprocessing, sparse dictionary construction, gas recognition and post-processing, and is applicable to the above device, and the specific method is as follows;
1. new data collection and data preprocessing:
1.1
Collecting raw data from a sensor array, including multidimensional characteristics of gas in a vehicle and environmental parameters (such as temperature, humidity and the like), sensor response vector representation and environmental parameter vector representation, and the sensor array receives signals, wherein the signals are expressed as response vectors X (M sensor groups);
X＝[X1,X2,K,XM]T
P environmental parameters, temperature, humidity, air pressure;
E＝[E1,E2,K,EP]T
1.2
And in the preprocessing stage, the sensor data is subjected to noise removal processing so as to improve the data quality and accuracy, and the sensor data is normalized to be in the same scale range so as to eliminate the dimensional difference among different sensors, thereby facilitating the subsequent processing.
Denoising by using wavelet transformation, wherein initial data is denoted as X 0、E0;
the wavelet transform decomposes the signal into components of different frequencies, where the high frequency components represent noise, decompose the sensor response matrix using a two-dimensional wavelet transform, decompose the matrix into low frequency and high frequency portions based on a two-dimensional discrete wavelet transform;
a: filtering in the horizontal and vertical directions is achieved by a convolution operation, wherein a low-pass filter h and a high-pass filter g are used to extract approximation and detail information;
ApproxLL＝X*h*hT
DetailLH＝X*h*gT
DetailHL＝X*g*hT
DetailHH＝X*g*gT
Where, represents convolution operation, h T and g T represent transposes of the filter, respectively;
b: after each subband (Approx and Detail) is obtained, it is downsampled to reduce the size of the matrix;
ApproxLL＝dw(ApproxLL)
DetailLH＝dw(DetailLH)
DetailHL＝dw(DetailHL)
DetailHH＝dw(DetailHH)
Where dw () represents the half of the sample points removed for each row and column;
c: step A and step B are repeated step by step, the low-frequency sub-band is continuously decomposed to form a multi-layer decomposition structure, and approximation and detail information on different scales are obtained;
During wavelet transformation, noise is concentrated in the high frequency detail part, and the noise is removed by setting the wavelet coefficient threshold of the high frequency detail part to zero.
The sensor data is normalized to the same scale range by the normalization operation, so that dimensional differences among different sensors are eliminated, and subsequent processing is facilitated;
Sensor response matrix X:
calculating a mean value:
mean(X)＝[mean(X1) mean(X2) Λ mean(XM)]
Wherein X i represents the i-th column of the matrix, i=1, 2, M;
Calculating standard deviation:
std(X)＝[std(X1) std(X2) Λ std(XM)]
finally, calculating to obtain a normalized sensor response matrix, wherein the data of each column has zero mean and unit variance;
1.3
Environmental parameter correction, correcting sensor response by a linear regression model, the model being expressed as:
Xcalibrated＝Xnormalized-W·E
w is a correction coefficient matrix, the objective of which is to correct the sensor response by the environmental parameters, estimated using the least squares method, i.e. minimizing the following loss function:
Wherein, F represents the Frobenius norm, namely, there is a loss function:
Calculating an optimal solution:
In the process, the matrix E T E needs to be kept as a reversible matrix, otherwise, the matrix E T E needs to be regularized, and an analytical solution of the correction coefficient matrix W is obtained by solving the equation;
W＝(ETE)-1ETXcalibrated
The sensor response was corrected with a nonlinear regression model, the RBF model used was expressed as:
Where J is the number of RBFs, representing the complexity of the model, w j is the weight of the jth RBF, RBF (E, c j,βj) is used to represent the output of the jth RBF, c j is the center, and β j is the width parameter;
likewise, model parameters are estimated with minimized loss functions;
After processing X calibrated, a response vector X pre is obtained, the quality and accuracy of sensor data can be improved through a preprocessing step, and a more reliable data base is provided for subsequent gas identification and concentration estimation.
2. Constructing an adaptive sparse coding dictionary:
2.1
Selecting a set of initial dictionary atoms, using the response pattern of the known gas as initial dictionary atoms, each vector representing the response pattern of a gas, forming the atoms into an initial dictionary matrix D0;
D0＝[ρ1,ρ2,K,ρN]
2.2
For the newly collected gas response denoted as X new, a new sensor response sample X new is generated;
The new sensor response sample X new is brought in, and the dictionary is updated through an online learning algorithm:
Wherein η t represents a learning rate for controlling a step size of dictionary updating;
For X new, performing coefficient coding, λ (λ > 0) is a regularization parameter used for balancing reconstruction errors and sparsity, obtaining a sparse representation matrix α t used for updating a dictionary, |·|| 2 representing an L2 norm (i.e., euclidean norm) used for balancing reconstruction errors; and 1 represents an L1 norm for promoting sparsity.
The iterative process is as follows:
For each sample X i:
A: updating by using the least square form of the LASSO problem, initializing alpha t to be a zero vector matrix, or using the sparse representation of the former step as an initial value for alpha t;
B: soft thresholding alpha t, i.e., thresholding each component alpha t, i;
αt,i＝sign(αt,i)·max(|αt,i|-λ,0)
C: updating a residual r t;
rt＝Xnew-Dtαt
Updating the dictionary matrix using residual r t;
Judging whether a stopping condition is met, ending iteration if the stopping condition is met, otherwise, returning to continue iteration; the sparse representation coefficients α t and the dictionary matrix D 0 are continuously updated until the optimal solution is converged or a predetermined number of iterations is reached, and the optimal judgment standard may be one or more of judgment conditions based on a residual threshold, a sparse representation matrix change threshold, the number of iterations, and the like.
The least square form of the LASO method is suitable for controlling sparsity, is suitable for sparsity requirements in sparse coding, and is more suitable for solving general sparse representation problems instead of sparsity constraints in LASO compared with a general residual iteration soft threshold algorithm, and is more suitable for updating a dictionary when constructing an adaptive sparse coding dictionary to handle non-monitoring gas interference problems in complex gas environments.
3. Gas identification
Based on the result α of sparse coding, the most relevant dictionary atoms, i.e. the most likely gas components, are identified, the threshold value of sparse coding is ε, and the identified gas components can be expressed as
GasID＝argmaxi|ai|
Where a i is the i-th vector of the sparse representation matrix vector α t, i is the index of the dictionary atom, and represents the i-th gas component.
4. Post-processing and decision support
C＝f(GasID,Xpre)
Wherein, C represents GasID the concentration of the corresponding gas, f is a concentration estimation function, and according to the concentration estimation result and the fault diagnosis information, an optimization suggestion is provided for the user, such as an adjustment of engine parameters, a replacement of parts, a decoration material, a maintenance, etc. is suggested.
The foregoing is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art, who is within the scope of the present invention, should make equivalent substitutions or modifications according to the technical scheme of the present invention and the inventive concept thereof, and should be covered by the scope of the present invention.
Claims (7)
1. The automobile sniffing method is characterized by comprising pretreatment, sparse dictionary construction, gas identification and post-treatment;
Capturing multidimensional characteristics of gas generated by an automobile by using a plurality of sensors of different types, recording environmental parameters at the same time, correcting the influence of the environmental parameters on the response of the sensors, and preprocessing to obtain a data set; using the response mode of the known gas as an initial dictionary atom, adding the response mode of the new gas into the dictionary through an online learning algorithm along with the collection of new data, and finding the optimal sparse representation in the adaptive dictionary for the new sensor response; identifying the most relevant dictionary atoms according to the result of sparse coding; based on the identified gas components and the sensor response intensity, the concentration of each gas is estimated, and the gas concentration is combined with the automobile emission standard and the identified gas concentration to be used for analyzing and identifying the odor intensity of the detected gas in the automobile environment.
2. The method of claim 1, wherein the preprocessing step includes removing noise and normalizing operations, wherein the wavelet transform decomposes the signal into components of different frequencies, the high frequency components represent noise, the sensor response matrix is decomposed using a two-dimensional wavelet transform, and the matrix is decomposed into low frequency and high frequency components based on a two-dimensional discrete wavelet transform;
a: filtering in the horizontal and vertical directions is achieved by a convolution operation, wherein a low-pass filter h and a high-pass filter g are used to extract approximation and detail information;
ApproxLL＝X*h*hT
DetailLH＝X*h*gT
DetailHL＝X*g*hT
DetailHH＝X*g*gT
Where, represents convolution operation, h T and g T represent transposes of the filter, respectively;
b: after each Approx and Detail is obtained, it is downsampled to reduce the size of the matrix;
ApproxLL＝dw(ApproxLL)
DetailLH＝dw(DetailLH)
DetailHL＝dw(DetailHL)
DetailHH＝dw(DetailHH)
Where dw () represents the half of the sample points removed for each row and column;
c: step A and step B are repeated step by step, the low-frequency sub-band is continuously decomposed to form a multi-layer decomposition structure, and approximation and detail information on different scales are obtained;
During wavelet transformation, noise is concentrated in the high frequency detail part, and the noise is removed by setting the wavelet coefficient threshold of the high frequency detail part to zero.
3. The method of claim 2, wherein the environmental parameter is corrected, and the sensor response is corrected using a linear regression model, the model being expressed as:
Xcalibrated＝Xnormalized-W·E
w is a correction coefficient matrix, the objective of which is to correct the sensor response by the environmental parameters, estimated using the least squares method, i.e. minimizing the following loss function:
Wherein, F represents the Frobenius norm, namely, there is a loss function:
Calculating an optimal solution:
In the process, the matrix E T E needs to be kept as a reversible matrix, otherwise, the matrix E T E needs to be regularized, and an analytical solution of the correction coefficient matrix W is obtained by solving the equation;
W＝(ETE)-1ETXcalibrated
4. The method of claim 2, wherein the sensor response is corrected using a nonlinear regression model, and the RBF model is expressed as:
Where J is the number of RBFs, representing the complexity of the model, w j is the weight of the jth RBF, RBF (E, c j,βj) is used to represent the output of the jth RBF, c j is the center, and β j is the width parameter;
estimating model parameters with a minimized loss function;
5. The method of claim 1, wherein constructing an adaptive sparse coding dictionary comprises:
S1, selecting a group of initial dictionary atoms, using a response mode of known gas as the initial dictionary atoms, wherein each vector represents a response mode of gas, and forming the atoms into an initial dictionary matrix D 0;
S2, generating a new sensor response sample X new for a newly collected gas response denoted as X new;
The new sensor response sample X new is brought in, and the dictionary is updated through an online learning algorithm:
Wherein η t represents a learning rate for controlling a step size of dictionary updating;
For X new, performing coefficient coding, λ (λ > 0) is a regularization parameter used for balancing reconstruction errors and sparsity, obtaining a sparse representation matrix α t used for updating a dictionary, |·|| 2 representing an L2 norm (i.e., euclidean norm) used for balancing reconstruction errors; and 1 represents an L1 norm for promoting sparsity.
6. The automobile sniffing method of claim 5, wherein the iterative process is as follows:
For each sample X i:
A: updating by using the least square form of the LASSO problem, initializing alpha t to be a zero vector matrix, or using the sparse representation of the former step as an initial value for alpha t;
B: soft thresholding alpha t, i.e., thresholding each component alpha t, i;
αt,i＝sign(αt,i)·max(|αt,i|-λ，0)
C: updating a residual r t;
rt＝Xnew-Dtαt
Updating the dictionary matrix using residual r t;
Judging whether a stopping condition is met, ending iteration if the stopping condition is met, otherwise, returning to continue iteration; the sparse representation coefficients α t and the dictionary matrix D 0 are continuously updated until the optimal solution is converged or a predetermined number of iterations is reached, and the optimal judgment standard may be one or more of judgment conditions based on a residual threshold, a sparse representation matrix change threshold, the number of iterations, and the like.
7. An automobile sniffing device for implementing the automobile sniffing method of any one of claims 1-6, comprising a sensor array, an environmental parameter sensor, a data acquisition device, a computing unit, a user interface, a storage device, and a communication module;
Wherein the sensor array is used for capturing multidimensional characteristics of gas in the vehicle, the sensor comprises a sensor capable of detecting various gas components, and the environmental parameter sensor is used for recording environmental parameters; the data acquisition equipment is used for receiving and processing the data of the sensor array and the environmental parameter sensor, and preprocessing and feature extraction are carried out; the computing unit is used for executing an algorithm and comprises computing steps of sparse representation, gas identification, concentration estimation, fault diagnosis and the like; the user interface is used for displaying the identification result, concentration estimation and fault diagnosis information and providing suggestions for decision support; the storage device is used for storing the sensor data, the model parameters and the history record so as to perform performance evaluation and model optimization; the communication module is used for carrying out data exchange and remote control with other systems or cloud.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410271115.8A CN118010935A (en) | 2024-03-11 | 2024-03-11 | Method and device for sniffing automobiles |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410271115.8A CN118010935A (en) | 2024-03-11 | 2024-03-11 | Method and device for sniffing automobiles |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118010935A true CN118010935A (en) | 2024-05-10 |
Family
ID=90953828
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410271115.8A Pending CN118010935A (en) | 2024-03-11 | 2024-03-11 | Method and device for sniffing automobiles |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118010935A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102866179A (en) * | 2012-09-13 | 2013-01-09 | 重庆大学 | Online recognition and inhibition method based on non-target interference smell in electronic nose of artificial intelligent learning machine |
| CN103868955A (en) * | 2014-04-09 | 2014-06-18 | 西南大学 | Portable in-automobile gas detection system and method |
| CN105241823A (en) * | 2015-09-30 | 2016-01-13 | 西安交通大学 | Thermal power plant flue gas spectral quantitative analysis method based on sparse representation |
| US20160123943A1 (en) * | 2013-06-05 | 2016-05-05 | Institute of Microelectronics, Chinese Academy of Sciences | Gas recognition method based on compressive sensing theory |
| CN109740666A (en) * | 2018-12-29 | 2019-05-10 | 西南大学 | The electronic nose feature extraction and classifying method towards the detection of orange blossom fragrance based on LKSVD |
-
2024
- 2024-03-11 CN CN202410271115.8A patent/CN118010935A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102866179A (en) * | 2012-09-13 | 2013-01-09 | 重庆大学 | Online recognition and inhibition method based on non-target interference smell in electronic nose of artificial intelligent learning machine |
| US20160123943A1 (en) * | 2013-06-05 | 2016-05-05 | Institute of Microelectronics, Chinese Academy of Sciences | Gas recognition method based on compressive sensing theory |
| CN103868955A (en) * | 2014-04-09 | 2014-06-18 | 西南大学 | Portable in-automobile gas detection system and method |
| CN105241823A (en) * | 2015-09-30 | 2016-01-13 | 西安交通大学 | Thermal power plant flue gas spectral quantitative analysis method based on sparse representation |
| CN109740666A (en) * | 2018-12-29 | 2019-05-10 | 西南大学 | The electronic nose feature extraction and classifying method towards the detection of orange blossom fragrance based on LKSVD |
Non-Patent Citations (1)
| Title |
|---|
| 李熙 等: "基于单个SnO2传感器的CO/H2混合气体定量分析", 《传感技术学报》, vol. 20, no. 10, 31 December 2007 (2007-12-31), pages 2169 - 2173 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111476339B (en) | Rolling bearing fault feature extraction method, intelligent diagnosis method and system | |
| CN113485302A (en) | Vehicle operation process fault diagnosis method and system based on multivariate time sequence data | |
| CN111783616B (en) | Nondestructive testing method based on data-driven self-learning | |
| CN116304549A (en) | Wavelet threshold denoising method for tunnel health monitoring data | |
| CN113703422B (en) | Gas turbine pneumatic actuator fault diagnosis method based on feature analysis processing | |
| CN108694375B (en) | Imaging white spirit identification method applicable to multi-electronic nose platform | |
| CN118010935A (en) | Method and device for sniffing automobiles | |
| CN116596326B (en) | Urban environment detection and comprehensive evaluation method based on remote sensing data | |
| CN109840386B (en) | Damage identification method based on factor analysis | |
| CN111251187A (en) | Method and device for fusing information and extracting characteristics of blade grinding burn | |
| CN111079591A (en) | Bad data restoration method and system based on improved multi-scale principal component analysis | |
| CN113836476B (en) | Pavement evenness measuring method based on smart phone | |
| CN115830368A (en) | Vehicle shock absorber defect diagnosis method, system, equipment and medium | |
| CN115905827A (en) | Steam turbine fault diagnosis method and device based on neural network | |
| CN115130500A (en) | Motor fault diagnosis system and method based on SDAE algorithm | |
| CN114299323A (en) | Printed matter defect detection method combining machine vision and deep learning technology | |
| CN112834194B (en) | Fault intelligent detection method based on soft target measurement under fault-free sample | |
| CN117934963B (en) | Gas sensor drift compensation method | |
| CN115288994B (en) | Improved DCGAN-based compressor abnormal state detection method | |
| CN117216489B (en) | Waterproof coating quality inspection analysis method and system based on Internet | |
| CN115659150B (en) | Signal processing method, device and equipment | |
| CN116975527A (en) | Fault diagnosis method based on soft thresholding and wavelet transformation denoising | |
| CN117523229A (en) | Hyperspectral small sample classification method and system based on space feature contrast loss | |
| CN117612277A (en) | Unsupervised Internet of vehicles anomaly detection method and system based on physical invariant subspace | |
| CN117474790A (en) | Pointer instrument image denoising method and device based on improved wavelet threshold function |
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 |