CN115266639A - Laser methane telemetering concentration calculation method and system supporting vector regression - Google Patents
Laser methane telemetering concentration calculation method and system supporting vector regression Download PDFInfo
- Publication number
- CN115266639A CN115266639A CN202210869738.6A CN202210869738A CN115266639A CN 115266639 A CN115266639 A CN 115266639A CN 202210869738 A CN202210869738 A CN 202210869738A CN 115266639 A CN115266639 A CN 115266639A
- Authority
- CN
- China
- Prior art keywords
- concentration
- methane
- vector regression
- laser
- concentration calculation
- 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
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/39—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using tunable lasers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
- G01N2201/126—Microprocessor processing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
- G01N2201/129—Using chemometrical methods
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Analytical Chemistry (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Optics & Photonics (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a laser methane telemetering concentration calculation method and system for a support vector regression machine, and belongs to the technical field of laser methane telemetering. Firstly, acquiring original signals at different detection distances and different methane gas concentrations, processing and transmitting the original signals to an upper computer to filter low-frequency noise and extract characteristics, then dividing a sample data set into a training set and a test set, inputting the training set into a support vector regression after disorder processing, training a model until an output prediction methane gas concentration result meets requirements, and finishing training; and finally, evaluating the generalization performance of the support vector regression algorithm constructed on the basis of the training set through the test set. The invention can learn the relation between the methane gas signal characteristics and the concentration, has high prediction precision on the methane concentration, is more accurate in detection, can adapt to detection under various working conditions, and reduces economic loss and potential safety hazard caused by methane leakage.
Description
Technical Field
The invention belongs to the technical field of laser methane telemetering, and particularly relates to a laser methane telemetering concentration calculation method and system based on a support vector regression.
Background
In recent years, natural gas has been widely used in industry and people's life as a clean energy source. The safety problem of natural gas is more prominent, especially natural gas gathering and transportation sites and natural gas pipelines, and because the economic loss and the casualties caused by natural gas leakage and explosion are serious, the safety of natural gas becomes the basic safety requirement of society. The main gas of natural gas is methane, the natural gas is mostly detected by detecting methane gas at present, the most widely used detection methods are a catalytic combustion method, a flame ion detector and a semiconductor gas sensor measurement method, however, the detection methods have the problems of short service life, low precision and poor stability, and a false alarm phenomenon often exists.
The tunable diode laser absorption spectrum technology is widely applied to methane sensors due to the advantages of high precision, quick response, accurate positioning and the like, but the data processing of the technology generally adopts methods such as a least square method, an L-M nonlinear fitting method, a peak fitting method and the like, and the methods are easily influenced by noise and have high fitting result errors. In the past, the concentration inversion is mostly carried out by adopting the peak value of an original signal, other characteristics of the original signal are not considered, and the influence of a detection distance on the concentration of methane gas is not considered, so that the problems of large detection error and low precision of a laser methane sensor are caused.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a laser methane telemetering concentration calculation method and a laser methane telemetering concentration calculation system based on a support vector regression.
A laser methane telemetering concentration calculation method supporting a vector regression machine comprises the following steps:
s1, collecting sample data, and sensing methane gas original signals at different detection distances and different methane gas concentrations. The laser methane sensor processes and transmits the methane gas original signal with concentration information obtained by each sensing to the upper computer.
In the upper computer, a high-pass filter is adopted to filter low-frequency noise of the processed methane gas original signal, then time domain characteristics and frequency domain characteristics are extracted, the extracted characteristics and the calibration concentration are used as a group of samples, data acquisition is carried out under various working conditions, a sample data set is constructed, and the sample set is standardized.
And (4) establishing a concentration calculation model by using a support vector regression mechanism.
And S2, dividing the sample data set into a training set and a testing set based on an independent distribution principle.
S3, in order to reduce the variance of the concentration calculation model and avoid overfitting of the concentration calculation model, carrying out disorder treatment on the training set; taking the time domain features and the frequency domain features in the training set as the input of a support vector regression machine, and taking the calibrated concentration value as a sample label of the support vector regression machine; and searching the optimal values of the SVR penalty factor c and the kernel function parameter g by a grid searching and cross verification method.
When the output prediction concentration of the concentration calculation model meets the requirement, the training is finished; if not, increasing the sample size and the sample type of the sample data set, and repeating the steps S1 to S3 until the requirements are met.
S4, inputting the test set into a concentration calculation model constructed on the basis of the training set to obtain the predicted methane concentration, evaluating the generalization performance, and if the requirements can be met, finishing the construction of the concentration calculation model; if not, increasing the sample size and the sample type of the sample data set, and repeating the steps S1 to S4 until the requirements are met.
A concentration calculation model based on a support vector regression machine is written into laser methane sensor calculation software, and methane gas detection is carried out in natural gas pipelines, natural gas transmission station yards, high-rise buildings and other yards by means of a laser methane remote measurement system.
A laser methane telemetering concentration calculation system supporting a vector regression machine is composed of a laser methane sensor, a signal amplification module, a data acquisition module and a computer.
The laser methane sensor is connected with the signal amplification module, the signal amplification module is connected with the data acquisition module, and the data acquisition module is connected with the computer.
The laser methane sensor is used for sensing a methane gas concentration signal; the signal amplification module is used for amplifying a methane gas concentration signal sensed by the laser methane sensor; the data acquisition module is used for carrying out A/D conversion and storage on the methane gas concentration signal amplified by the signal amplification module; the computer is provided with a computer system and concentration calculation software for analyzing and processing the methane gas concentration signal converted by the data acquisition module A/D and outputting a methane gas concentration calculation result, and the concentration calculation software comprises a support vector regression machine, parameters of the support vector regression machine and the like; the functions of the concentration calculation software comprise concentration real-time display, oscillogram real-time display, signal noise reduction, concentration early warning, detection database updating and the like.
The invention has the beneficial effects that: the invention can learn the relation between the signal characteristics and the concentration of the methane gas, so that the detection precision of the methane concentration is high, the false detection probability of the methane concentration is reduced, the invention can be more suitable for detection under various working conditions, and the economic loss and the potential safety hazard caused by methane leakage are reduced.
Drawings
FIG. 1 is a flow chart of the operation of a laser methane telemetry system;
FIG. 2 is a flow chart of the establishment of a laser methane telemetry concentration calculation model based on a support vector regression;
fig. 3 is a block diagram of laser methane telemetry system concentration calculation software.
Detailed Description
For a further understanding of the invention, reference will now be made to the preferred embodiments of the present invention by way of example, but it is to be understood that the description is intended to illustrate further features and advantages of the invention, rather than to limit the invention to the claims. The following description is merely a preferred embodiment of the present invention, and various operating conditions include, but are not limited to, the operating conditions of the embodiment. The present invention may be subject to several modifications without departing from the scope of the invention, and such modifications and improvements are intended to fall within the scope of the claims.
A laser methane telemetering concentration calculation method of a support vector regression machine is provided, which comprises the steps of adopting a support vector regression mechanism to build a concentration calculation model, selecting 21 characteristics of an original signal with concentration information as a sample set independent variable, calibrating concentration as a sample set dependent variable, and mapping data to a high-dimensional space by using nonlinear mapping. The 21 features of the original signal include time domain features including entropy, energy, maximum, minimum, mean, root mean square value, root mean square magnitude, variance, standard deviation, skewness factor, kurtosis, form factor, impulse factor, peak factor, and margin factor, and frequency domain features including center of gravity frequency, mean square frequency, root mean square frequency, frequency variance, and frequency standard deviation. The original signals of the methane gas measured at different distances with the same concentration or different concentrations with the same distance have differences, the original signals cannot be completely expressed by a single characteristic, and the 21 characteristics basically cover most of information and characteristics of the original signals of the methane gas, so that the original signals of the methane gas are expressed by the 21 characteristics.
A concentration calculation method for laser methane remote measurement based on a support vector regression mechanism is shown in FIG. 2 and specifically comprises the following steps:
1) Data is collected. The experimental detection distance is 70mm to 3500mm, the distance and the methane gas concentration are changed, and the experiment is repeated. Each experiment gas was measured at the same concentration and distance in 2 groups. The method comprises the steps of sequentially testing methane gas pools with 50ppm · m, 100ppm · m, 200ppm · m, 500ppm · m, 1000ppm · m, 1700ppm · m, 2000ppm · m, 5000ppm · m and 10000ppm · m 9 specifications, finally constructing a sample data set comprising 858 groups of samples, standardizing the sample data set, and enabling features of different dimensions to have certain comparability in terms of values so as to improve the convergence speed and detection accuracy of a model.
And (3) processing the methane gas original signal with concentration information measured by the laser methane sensor each time, and transmitting the processed signal to the upper computer software.
And (3) carrying out noise reduction treatment on the processed methane gas original signal by adopting a high-pass filter in an upper computer, extracting characteristics, and taking the extracted characteristics as a group of data, wherein the extracted characteristics comprise 21 characteristics of time domain characteristics and frequency domain characteristics of the methane gas original signal.
2) And then, the sample set is divided into a training set and a testing set based on an independent distribution principle, so that the generalization performance of a concentration calculation model based on a Support Vector Regression (SVR) is improved. In this embodiment 858 sets of concentration sample data sets are divided into training sets and test sets, the number of which is 675:183.
3) And carrying out disorder processing on the training set so as to reduce the variance of the concentration calculation model and avoid overfitting of the concentration calculation model, wherein the testing set has no influence on the concentration calculation model and does not carry out disorder processing, the training set is used for learning the parameters of the concentration calculation model, and the testing set is used for evaluating the overall generalization performance of the concentration calculation model. In the embodiment, an epsilon-SVM type support vector regression machine is adopted as a calculation model; the kernel function used is the radial kernel function (RBF).
The penalty factor c and the kernel function parameter g are trained and optimally selected through a training set, a grid search method is adopted, the search range is between-10 and 10, the step length is 0.5, and the c and the g take values in the range. And for the determined c and g, obtaining the accuracy of the training set under the group c and g by using a k-fold cross validation method for the training set, and finally taking the group c and g which enable the prediction result on the training set to have the highest accuracy as the optimal parameters, wherein 5-fold cross validation is adopted. The value of the loss function epsilon takes 0.01.
During training, inputting the training set subjected to disorder processing into a support vector regression machine, using 21 features of the time domain features and the frequency domain features extracted in the step 1) as the input of the support vector regression machine, and outputting the input as a predicted concentration value. When the deviation of the training meets the requirement, the training is finished, and c =5.6569 and g =0.70711 are performed.
4) Inputting the test set into a support vector regression machine constructed based on a training set to obtain the predicted methane concentration, wherein the number of samples in the test set is 183, the average absolute percentage error is 9.504%, the model prediction result meets the requirement, the model test meets the deviation requirement, the establishment of the prediction method of the concentration calculation model based on the support vector regression machine is finished, and a flow chart is shown in figure 2.
After a concentration calculation model based on a support vector regression machine is established, the concentration calculation model based on the support vector regression machine is written into laser methane telemetering system calculation software, and methane gas detection is carried out in natural gas pipelines, natural gas transmission station sites, high-rise buildings and other sites by means of the laser methane telemetering system.
A laser methane telemetering concentration calculation system based on a support vector regression consists of a laser methane sensor, a signal amplification module, a data acquisition module and a computer, wherein the signal amplification module is connected with the laser methane telemetering system in a wireless mode, and the laser methane telemetering concentration calculation system is connected with the computer in a wireless mode.
The laser methane sensor is connected with the signal amplification module, the signal amplification module is connected with the data acquisition module, and the data acquisition module is connected with the computer.
The laser methane sensor is used for sensing a methane gas concentration signal; the signal amplification module is used for amplifying a methane gas concentration signal sensed by the laser methane sensor; the data acquisition module is used for carrying out A/D conversion and storage on the methane gas concentration signal amplified by the signal amplification module; the computer is provided with a computer system and concentration calculation software for analyzing and processing the methane gas concentration signal converted by the data acquisition module A/D and outputting a methane gas concentration calculation result, and the concentration calculation software comprises a support vector regression machine, parameters of the support vector regression machine and the like; the functions of the concentration calculation software comprise concentration real-time display, oscillogram real-time display, signal noise reduction, concentration early warning, detection database updating and the like.
Preferably, the laser in the laser methane sensor is a distributed feedback laser (DFB), the working temperature of the laser is controlled to be 25 ℃, the driving current is set to be 80mA +/-20 mA, and the central wavelength of laser emitted by the laser is 1653.72nm at about 80mA, namely the absorption peak of methane gas.
By adopting a wavelength modulation spectrum technology, a hardware circuit is used for generating a 20KHz high-frequency sine wave and a 20Hz low-frequency triangular wave, and the sine wave and the triangular wave are injected into a distributed feedback laser to enable the laser to emit modulation wavelength. The low-frequency triangular wave is used for scanning laser wavelength near the center of a gas absorption spectrum line, and the high-frequency sine wave is used for improving the anti-noise capacity of the system, so that the extraction of original signals containing methane concentration information in the spectrum after methane gas absorption is facilitated.
FIG. 3 is a block diagram of the software for calculating the concentration of the laser methane telemetry system, FIG. 1 is a flow chart of the operation of the laser methane telemetry system, and the specific steps of the invention are as follows as can be seen from FIGS. 1 and 3;
(1) The laser methane sensor senses a methane gas signal;
(2) The signal amplification module is used for amplifying the methane gas signal in the step (1);
(3) Performing A/D conversion on the methane gas signal obtained in the step (2) based on a data acquisition module, and storing data;
(4) Sending the output signal of the data acquisition unit in the step (3) to a computer;
(5) And (3) displaying the waveform of the output signal in the step (4) in real time by concentration calculation software in a computer, extracting time domain characteristics and frequency domain characteristics of the signal, inputting the time domain characteristics and the frequency domain characteristics and the calibrated concentration value into the constructed concentration calculation model to obtain a calculation result, comparing the calculation result with an early warning value, displaying the methane concentration and the early warning result, and updating a detection database.
Claims (7)
1. A laser methane telemetering concentration calculation method supporting a vector regression machine is characterized by comprising the following specific steps:
s1, sensing methane gas original signals at different detection distances and different methane gas concentrations by a laser methane sensor, and processing and transmitting the sensed methane gas original signals to an upper computer;
filtering low-frequency noise of the processed methane gas original signal by adopting a high-pass filter in an upper computer, extracting time domain characteristics and frequency domain characteristics, and taking the extracted characteristics and the calibration concentration as a group of samples;
carrying out data acquisition under various working conditions, constructing a sample data set, and standardizing the sample data set;
constructing a concentration calculation model by adopting a Support Vector Regression (SVR);
s2, dividing the sample data set into a training set and a testing set;
s3, carrying out disorder processing on the training set, taking the time domain features and the frequency domain features in the training set as the input of the support vector regression machine, and calibrating a concentration value as a sample label of the support vector regression machine;
searching the penalty factor c and the optimal value of the kernel function parameter g of the Support Vector Regression (SVR) through a grid searching and cross verification method;
when the output prediction concentration of the concentration calculation model meets the requirement, the training is finished; if not, increasing the sample size and the sample type of the sample data set, and repeating the steps S1 to S3 until the requirements are met;
s4, inputting the test set into a concentration calculation model constructed based on the training set to obtain the predicted methane concentration, evaluating the generalization performance of the concentration calculation model, and finishing the establishment of the concentration calculation model if the generalization performance can meet the requirements; if not, increasing the sample size and the sample type of the sample data set, and repeating the steps 1 to 4 until the requirements are met.
2. The laser methane telemetry concentration calculation method based on the support vector regression machine according to claim 1, characterized in that: the time domain features in the step S1 comprise entropy, energy, maximum value, minimum value, average value, root mean square amplitude, variance, standard deviation, skewness factor, kurtosis, form factor, pulse factor, peak factor and margin factor;
the frequency domain characteristics comprise center of gravity frequency, mean square frequency, root mean square frequency, frequency variance and frequency standard deviation.
3. The method for calculating the telemetering concentration of methane by laser supporting a vector regression machine according to claim 1, wherein: the original signal in step S1 is processed by amplifying the signal first and then performing a/D conversion on the amplified signal.
4. The method for calculating the telemetering concentration of methane by laser supporting a vector regression machine according to claim 1, wherein: in the step S3, the time domain characteristic and the frequency domain characteristic are used as independent variables of the sample set, the calibration concentration is used as a dependent variable of the sample set, and the data are mapped to a high-dimensional space by using nonlinear mapping;
before the training of the concentration calculation model, a loss parameter epsilon and a kernel function are determined.
5. A laser methane telemetering concentration calculation system supporting a vector regression machine is characterized in that: the laser methane telemetering concentration calculation system supporting the vector regression machine comprises a laser methane sensor, a signal amplification module, a data acquisition module and a computer;
the laser methane sensor is connected with the signal amplification module, the signal amplification module is connected with the data acquisition module, and the data acquisition module is connected with the computer;
the laser methane sensor is used for sensing a methane gas concentration signal;
the signal amplification module is used for amplifying a methane gas concentration signal sensed by the laser methane sensor;
the data acquisition module is used for carrying out A/D conversion and storage on the methane gas concentration signal amplified by the signal amplification module;
and the computer is used for analyzing and processing the methane gas concentration signal subjected to A/D conversion by the data acquisition module and outputting a methane gas concentration calculation result.
6. The system for calculating the telemetered concentration of methane by using a support vector regression machine according to claim 5, wherein: the computer is provided with a computer system and concentration calculation software;
the concentration calculation software comprises a support vector regression machine and parameters of the support vector regression machine.
7. The system for calculating the telemetered concentration of methane by using a support vector regression machine according to claim 5, wherein: the functions of the concentration calculation software comprise concentration real-time display, oscillogram real-time display, signal noise reduction, concentration early warning and detection database updating.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210869738.6A CN115266639A (en) | 2022-07-22 | 2022-07-22 | Laser methane telemetering concentration calculation method and system supporting vector regression |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210869738.6A CN115266639A (en) | 2022-07-22 | 2022-07-22 | Laser methane telemetering concentration calculation method and system supporting vector regression |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115266639A true CN115266639A (en) | 2022-11-01 |
Family
ID=83769451
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210869738.6A Pending CN115266639A (en) | 2022-07-22 | 2022-07-22 | Laser methane telemetering concentration calculation method and system supporting vector regression |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115266639A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117890327A (en) * | 2024-03-14 | 2024-04-16 | 鞍山天汇科技有限公司 | Laser methane sensor stability evaluation method |
-
2022
- 2022-07-22 CN CN202210869738.6A patent/CN115266639A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117890327A (en) * | 2024-03-14 | 2024-04-16 | 鞍山天汇科技有限公司 | Laser methane sensor stability evaluation method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tian et al. | Retrieval of gas concentrations in optical spectroscopy with deep learning | |
| CN111520192B (en) | Non-contact tunnel engineering construction rock burst real-time forecasting optimization method | |
| CN114563150B (en) | Bridge health online detection module generation method, detection method, tool box and device | |
| CN103412004B (en) | Method for detecting storage time of citrus sinensis | |
| CN111007035B (en) | Method, system and device for detecting concentration of high-temperature gas in secondary combustion chamber based on convolutional neural network and storage medium | |
| CN115266639A (en) | Laser methane telemetering concentration calculation method and system supporting vector regression | |
| CN116881749B (en) | Pollution site construction monitoring method and system | |
| CN116985183B (en) | Quality monitoring and management method and system for near infrared spectrum analyzer | |
| CN116181416A (en) | Multi-parameter advanced early warning system and method for rock burst | |
| CN104516991A (en) | Gamma sensor full-temperature range compensation method | |
| CN111898744A (en) | TDLAS trace gas concentration detection method based on pooled LSTM | |
| CN111539393B (en) | Oil-gas pipeline third-party construction early warning method based on EMD decomposition and LSTM | |
| CN115949891B (en) | Intelligent control system and control method for LNG (liquefied Natural gas) station | |
| CN115582637B (en) | Automatic detection system for laser cutting leakage process | |
| CN117214133A (en) | Methane detection method with ranging function based on TDLAS | |
| CN111798638A (en) | Auxiliary system fire information processing method based on information fusion | |
| CN114397273B (en) | Gas concentration measuring device and method based on combination of second harmonic wave and fourth harmonic wave | |
| CN103454114B (en) | The diagnosis of mesolow gas pressure regulator, governor operation troubles and safety early warning device and method | |
| CN111220571B (en) | Second harmonic signal fitting method and system based on amplitude dispersion | |
| CN112903617B (en) | Calibration method and system of carbon dioxide gas sensor based on curved surface fitting | |
| CN103424524B (en) | Method of detecting storage time of Annona squamosa | |
| CN117723917B (en) | Monitoring application method based on optical fiber extrinsic Fabry-Perot type ultrasonic sensor | |
| CN218512308U (en) | Gas concentration detection device based on multi-harmonic information fusion laser absorption spectroscopy technology | |
| CN116975768B (en) | Data anomaly detection method for fertilizer safety production | |
| CN103412005B (en) | Detection method for storage time of sweet oranges |
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 |