CN117747008B - Baseline fitting and noise reduction method and system for gas laser absorption spectrum - Google Patents
Baseline fitting and noise reduction method and system for gas laser absorption spectrum Download PDFInfo
- Publication number
- CN117747008B CN117747008B CN202410171508.1A CN202410171508A CN117747008B CN 117747008 B CN117747008 B CN 117747008B CN 202410171508 A CN202410171508 A CN 202410171508A CN 117747008 B CN117747008 B CN 117747008B
- Authority
- CN
- China
- Prior art keywords
- signal
- baseline
- noise
- neural network
- absorbance
- 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.)
- Active
Links
- 230000009467 reduction Effects 0.000 title claims abstract description 90
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000000862 absorption spectrum Methods 0.000 title claims abstract description 41
- 238000013528 artificial neural network Methods 0.000 claims abstract description 122
- 238000012549 training Methods 0.000 claims abstract description 82
- 238000012360 testing method Methods 0.000 claims abstract description 75
- 230000001629 suppression Effects 0.000 claims abstract description 53
- 238000001514 detection method Methods 0.000 claims abstract description 40
- 238000012545 processing Methods 0.000 claims abstract description 23
- 238000002835 absorbance Methods 0.000 claims description 160
- 238000010521 absorption reaction Methods 0.000 claims description 65
- 230000005540 biological transmission Effects 0.000 claims description 23
- 238000004088 simulation Methods 0.000 claims description 19
- 230000003595 spectral effect Effects 0.000 claims description 14
- 238000012886 linear function Methods 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 6
- 230000000694 effects Effects 0.000 abstract description 15
- 238000003672 processing method Methods 0.000 abstract description 3
- 239000007789 gas Substances 0.000 description 71
- 238000010586 diagram Methods 0.000 description 20
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 16
- 230000003287 optical effect Effects 0.000 description 10
- OTMSDBZUPAUEDD-UHFFFAOYSA-N Ethane Chemical compound CC OTMSDBZUPAUEDD-UHFFFAOYSA-N 0.000 description 8
- 238000000411 transmission spectrum Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 5
- 238000005259 measurement Methods 0.000 description 5
- 238000004847 absorption spectroscopy Methods 0.000 description 4
- 238000001914 filtration Methods 0.000 description 4
- 239000000203 mixture Substances 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000002238 attenuated effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 229910052757 nitrogen Inorganic materials 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
Classifications
-
- 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
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The application relates to the technical field of signal processing, and provides a baseline fitting and noise reduction method and system for a gas laser absorption spectrum. The method comprises the following steps: transmitting a detection light signal to the gas to be detected; establishing a first data set and a baseline fitting neural network; training and testing a baseline fitting neural network by adopting a first data set to obtain a baseline fitting signal; establishing a second data set and a noise suppression neural network; and training and testing the noise suppression neural network by adopting a second data set to obtain a noise reduction signal. The baseline fitting and noise reduction method does not need to additionally increase hardware equipment, does not introduce additional hardware noise, does not have frequency limitation on noise, improves the problems of baseline signal drift and noise interference existing in gas concentration detection compared with the existing processing method, and has the advantages of fast processing and good noise reduction effect. The method is not limited by the structure, is suitable for small-sized systems and short-path equipment, effectively reduces the equipment cost and has a wider application range.
Description
Technical Field
The application relates to the technical field of signal processing, in particular to a baseline fitting and noise reduction method and system for a gas laser absorption spectrum.
Background
The detection requirement of gas concentration is widely applied to various fields, and the gas concentration is generally obtained by adopting a direct absorption spectrum technology in an absorption spectrum.
The direct absorption spectroscopy (Direct Absorption Spectroscopy, DAS) technology has a great potential in the aspect of miniaturization of a sensing system due to the characteristics of simple system structure, convenient operation and relatively low requirements on a detector. However, in practical use, the DAS technique requires a group of non-absorption signals as baseline signals, and the baseline signals are usually obtained by selecting "non-absorption region" data in the transmission spectrum signals and performing low-order polynomial fitting, but the selection of "non-absorption region" data and the selection of polynomials are performed empirically, and there is no fixed rule. The baseline signal obtained by the polynomial fitting method tends to drift with greater uncertainty, which can cause measurement errors.
In order to obtain a more accurate baseline signal, two beams of detection light signals can be adopted, one beam is used for normal detection, and the other beam is used as reference light, so that the obtained baseline signal is more accurate, interference noise is easy to generate, and the detection precision of the gas concentration can be reduced due to the existence of the interference noise.
Disclosure of Invention
The application provides a baseline fitting and noise reduction method and system for a gas laser absorption spectrum, which are used for solving the technical problems of baseline signal drift and noise interference in gas concentration detection.
The baseline fitting and noise reduction method for the gas laser absorption spectrum provided by the first aspect of the application comprises the following steps: transmitting a detection light signal to the gas to be detected; establishing a first data set and a baseline fitting neural network; training and testing a baseline fitting neural network by adopting a first data set to obtain a baseline fitting signal; the input of the baseline fitting neural network is a transmitted light signal, the output of the baseline fitting neural network is a non-absorption baseline signal, and the transmitted light signal and the non-absorption baseline signal are obtained through detection light signal processing; establishing a second data set and a noise suppression neural network; wherein the second data set comprises absorbance signals, the absorbance signals being the ratio of the non-absorbing baseline signal to the transmitted light signal; training and testing a noise suppression neural network by adopting a second data set to obtain a noise reduction signal, wherein the input of the noise suppression neural network is an absorbance signal, the output of the noise suppression neural network is a high signal-to-noise ratio absorbance signal or a noise-free simulation absorbance signal, the high signal-to-noise ratio absorbance signal is obtained through absorbance signal processing, and the noise-free simulation absorbance signal is obtained through simulation program processing.
In some possible implementations, establishing the first data set includes: acquiring a plurality of baseline training data and a plurality of baseline test data; the baseline training data and the baseline test data are obtained by detecting optical signals, wherein the baseline training data comprise a first transmitted optical signal and a first non-absorption baseline signal, and the baseline test data comprise a second transmitted optical signal and a second non-absorption baseline signal; and selecting the baseline training data and the baseline test data according to a first characteristic selection principle to obtain a first data set.
In some possible implementations, establishing the second data set includes: acquiring a plurality of noise reduction training data and a plurality of noise reduction test data, wherein the noise reduction training data comprises a first absorbance signal and a first high signal-to-noise ratio absorbance signal or a first noise-free simulation absorbance signal, and the noise reduction test data comprises a second absorbance signal and a second high signal-to-noise ratio signal or a second noise-free simulation absorbance signal; and selecting the noise reduction training data and the noise reduction test data according to the principle of selecting the second features to obtain a second data set.
In some possible implementations, the first feature selection principle is to set values corresponding to absorption peaks of the first transmitted light signal and the second transmitted light signal to zero; the second characteristic selection principle is that absorption peaks of the first absorbance signal and the second absorbance signal are fitted by using a Lorentz linear function; wherein the Lorentz line functionThe method comprises the following steps:
；
is the center frequency of the molecular absorbance spectral line of the gas to be detected,/> Full width at half maximum of molecular absorbance spectrum line representing gas to be measured,/>Is the outgoing light frequency of the laser.
In some possible implementations, the baseline fitting neural network and the noise suppression neural network each include an input layer, a first hidden layer, a second hidden layer, and an output layer; the activation functions between the input layer and the first hidden layer, between the first hidden layer and the second hidden layer and between the second hidden layer and the third hidden layer are Sigmoid functions:
；
Wherein, Is a natural constant,/>Input spectral data for the input layer or the first hidden layer or the second hidden layer; the activation function between the third hidden layer and the output layer is/>; Wherein/>And outputting spectral data for the third hidden layer.
In some possible implementations, the loss functions of the baseline fitting neural network and the noise suppressing neural network are the same; the training method of the baseline fitting neural network and the noise suppression neural network is the same.
In some possible implementations, the baseline fitting neural network is tested using a mean square error, MSE, which is:
；
Wherein, Fitting the output value of the neural network to the trained baseline,/>Is a second non-absorptive baseline signal; i is the ith spectral data, and n is the number of input spectral data.
In some possible implementations, the noise suppression neural network is tested using a signal-to-noise ratio function, SNR, which is:
；
Wherein, Absorption peak value of second absorbance signal,/>Is the non-absorption peak portion of the second absorbance signal.
In some possible implementations, the baseline fitting and denoising method of gas laser absorption spectra further includes: and calculating the concentration of the gas to be detected according to the noise reduction signal.
The application provides a baseline fitting and noise reduction system for a gas laser absorption spectrum, which comprises a sending module; configured to transmit a detection light signal into a gas to be detected; a first establishing module configured to establish a first data set and a baseline fitting neural network; the first training and testing module is configured to train and test the baseline fitting neural network by adopting a first data set to obtain a baseline fitting signal; the input of the baseline fitting neural network is a transmitted light signal, the output of the baseline fitting neural network is a non-absorption baseline signal, and the transmitted light signal and the non-absorption baseline signal are obtained through detection light signal processing; a second establishing module configured to establish a second data set and a noise suppressing neural network; wherein the second data set comprises absorbance signals, the absorbance signals being the ratio of the non-absorbing baseline signal to the transmitted light signal; the second training and testing module is configured to train and test the noise suppression neural network by adopting a second data set to obtain a noise reduction signal; the input of the noise suppression neural network is an absorbance signal, the output of the noise suppression neural network is a high signal-to-noise ratio absorbance signal, the absorbance signal is obtained through absorbance signal processing, the noise-free simulation absorbance signal is obtained through simulation program processing.
The baseline fitting and noise reduction method for the gas laser absorption spectrum can adaptively fit a baseline signal according to the transmitted light signal, and can effectively inhibit noise in an original absorbance signal. The method is not limited by the structure, is suitable for small-sized systems and short-path equipment, and has a wide application range.
Drawings
In order to more clearly illustrate the technical solution of the present application, the drawings that are needed in the embodiments will be briefly described below, and it will be obvious to those skilled in the art that other drawings can be obtained from these drawings without inventive effort.
Fig. 1 is a flow chart of a baseline fitting and noise reduction method for a gas laser absorption spectrum according to an embodiment of the present application;
FIG. 2a is a schematic diagram of a transmitted light signal according to an embodiment of the present application;
FIG. 2b is a schematic diagram of feature selection using a first feature selection principle according to an embodiment of the present application;
FIG. 3 is a schematic flow chart of training and testing a baseline fitting neural network according to an embodiment of the present application;
FIG. 4 is a schematic diagram of a simulation versus original absorbance signal provided by an embodiment of the application;
FIG. 5a is a schematic diagram of the original absorbance signal of the mixed gas according to the embodiment of the present application;
FIG. 5b is a graph showing the absorbance of the scaled mixture gas according to the embodiment of the present application;
FIG. 6 is a schematic diagram of feature selection using a second feature selection principle according to an embodiment of the present application;
FIG. 7 is a schematic flow chart of training and testing a noise suppression neural network according to an embodiment of the present application;
FIG. 8 is a flowchart of a specific application of a baseline fitting and noise reduction method for a gas laser absorption spectrum according to an embodiment of the present application;
FIG. 9 is a schematic illustration of the effect of a baseline fit provided by an embodiment of the present application;
FIG. 10a is a schematic diagram of a raw absorbance signal provided by an embodiment of the application;
FIG. 10b is a schematic diagram of a denoised absorbance signal according to an embodiment of the application;
FIG. 11 is a block diagram of a baseline fitting and noise reduction system for gas laser absorption spectra according to an embodiment of the present application;
fig. 12 is a schematic structural diagram of a baseline fitting and noise reduction system for gas laser absorption spectrum according to an embodiment of the present application.
The graphic indicia:
100-baseline fitting and noise reduction system; 101-a transmitting module; 102-a first building module; 103-a first training and testing module; 104-a second building module; 105-second training and testing module.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly described below with reference to the drawings in the embodiments of the present application. It will be apparent that the described embodiments are some, but not all, embodiments of the application. Based on the embodiments of the present application, other embodiments that may be obtained by those of ordinary skill in the art without making any inventive effort are within the scope of the present application.
Hereinafter, the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first", "a second", etc. may explicitly or implicitly include one or more such feature. In the description of the present application, unless otherwise indicated, the meaning of "a plurality" is two or more.
Furthermore, in the present application, the terms "upper," "lower," "inner," "outer," and the like are defined relative to the orientation in which the components are schematically depicted in the drawings, and it should be understood that these directional terms are relative concepts, which are used for descriptive and clarity relative thereto, and which may be varied accordingly with respect to the orientation in which the components are depicted in the drawings.
The gas concentration detection requirement widely exists in the fields of industrial production safety, environment monitoring, deep space and deep sea detection, energy production and utilization and the like, and the optical detection method becomes one of the most promising technical routes in the current gas detection technology due to the high response speed, high safety, non-invasive measurement and great potential in miniaturization and low power consumption.
Direct Absorption Spectroscopy (DAS) has a great potential in the aspect of miniaturization of a sensing system due to the characteristics of simple system structure, convenient operation and relatively low requirements on a detector. However, in practical use, the DAS technique requires a group of non-absorption signals as baseline signals, and the baseline signals are usually obtained by selecting "non-absorption region" data in the transmission spectrum signals and performing low-order polynomial fitting, but the selection of "non-absorption region" data and the selection of polynomials are performed empirically, and there is no fixed rule. The baseline signal obtained by the polynomial fitting method tends to drift with greater uncertainty, which can cause measurement errors. Under weak absorption conditions such as low concentration of the gas to be measured, weak absorption line intensity, too short effective optical path and the like, measurement errors caused by polynomial fitting of the base line are often unacceptable.
To calculate the gas concentration, the original absorbance signal is calculated by comparing the baseline fitting signal and the transmission spectrum signal, and the original absorbance signal is inevitably interfered by various noises including various noises on the optical path and the circuit, and the existence of the noises can reduce the detection performance of the system, especially the influence caused by the noises under the conditions of miniaturization and short optical path of the system is relatively larger, so that the noise reduction treatment on the detection signal is necessary. The common noise reduction method includes multiple signal accumulation average and S-G filtering. The multiple averaging method has very good noise reduction effect on white noise, but in theory, the noise reduction effect often depends on the average frequency, but in practical application, the average frequency cannot be increased infinitely due to the limitation of measurement response time requirements, hardware cost and the like. S-G filtering is widely applied to data smoothing processing, and can ensure that the shape of a signal is unchanged while filtering high-frequency noise, but the processing of local abrupt signals can cause larger errors, and in addition, the filtering of low-frequency noise often causes phenomena of signal distortion and amplitude reduction, so that the noise reduction effect is poor.
In order to solve the technical problems, the embodiment of the application provides a baseline fitting and noise reduction method for a gas laser absorption spectrum. The method does not need to add extra hardware equipment, does not introduce extra hardware noise, has no frequency limitation on noise, and has the advantages of quick processing and good noise reduction effect compared with the existing processing method.
Fig. 1 is a flow chart of a baseline fitting and noise reduction method for a gas laser absorption spectrum according to an embodiment of the present application.
Referring to fig. 1, the method may be implemented by the following steps S100 to S500.
Step S100: and sending a detection light signal to the gas to be detected.
Step S200: a first dataset and a baseline fitting neural network are established.
In step S200, a first dataset may be established followed by establishing a baseline fitting neural network.
In establishing the first data set, the following steps S201 and S202 may be employed.
Step S201: a plurality of baseline training data and a plurality of baseline test data are acquired.
The baseline training data and the baseline testing data are obtained by detecting the optical signals, the baseline training data are used for executing training operation on the baseline fitting neural network subsequently, and the baseline testing data are used for executing testing operation on the baseline fitting neural network subsequently.
Specifically, the baseline training data includes a first transmitted light signal and a first non-absorbed baseline signal, and the baseline test data includes a second transmitted light signal and a second non-absorbed baseline signal. The transmitted light signal provided in the embodiment of the application is a spectrum signal.
That is, both baseline training data and baseline test data include transmitted light signals and non-absorbed baseline signals.
Specifically, the direct absorption spectroscopy technology generally needs a group of stable non-absorption signals as baseline signals, but due to the use of the uncooled photoelectric detector, the detection effect is easily affected by the change of the external environment temperature and the accumulated heat caused by the self-operation, so that the baseline signals deviate, and detection errors are caused. However, the temperature change caused by the external environment temperature and the detector operation is negligible in a short time, so in the embodiment of the application, the baseline training data and the baseline test data can be obtained by continuously measuring the first transmission light signals of different concentration gases and the corresponding first non-absorption baseline signals, the second transmission light signals and the corresponding second non-absorption baseline signals in a short time and establishing the mapping relation between the first transmission light signals and the second transmission light signals and the corresponding second non-absorption baseline signals.
Thus, the number of the first transmission light signals and the second transmission light signals is multiple, and the first transmission light signals and the second transmission light signals can be transmission light signals of the gas to be detected under different concentration conditions. The number of first non-absorptive baseline signals is the same as the number of first transmissive light signals, and the number of second non-absorptive baseline signals is the same as the number of second transmissive light signals.
Step S202: and selecting the baseline training data and the baseline test data according to a first characteristic selection principle to obtain a first data set.
The principle of the first feature selection is as follows: the values corresponding to the absorption peaks of the first transmitted light signal and the second transmitted light signal are set to zero.
Specifically, the first transmitted light signal and the second transmitted light signal are affected by the concentration effect, so that the correlation between the transmitted light signal and the non-absorption baseline signal can be reduced, and the difficulty of network training is increased. Therefore, in order to eliminate the influence of the concentration effect as much as possible, the concentration-affected portions of the first and second transmitted light signals, that is, the corresponding values in the absorption peak range, are set to zero, and the rest portions are not processed.
Fig. 2a is a schematic diagram of a transmitted light signal according to an embodiment of the present application.
Fig. 2b is a schematic diagram of feature selection using the first feature selection principle according to an embodiment of the present application.
In one specific implementation, referring to fig. 2a and 2b, the feature selection results of a baseline fitting neural network are shown, where the abscissa is wave number and the ordinate is transmitted light signal. In fig. 2a, the transmitted light signal is obtained by detecting the detected light signal, the transmitted light signal generates a recess due to the absorption of the gas to be measured, and the concentration of the gas to be measured is different, so that the concentration effect is introduced. In order to reduce the influence of concentration effect, the value of the concave part is set to 0, the rest part of the transmitted light signal is taken as a characteristic, so that the characteristic selection of the input signal of the baseline fitting neural network is completed, and the final input signal of the baseline fitting neural network is shown as a dashed curve in fig. 2b.
Thus, the first data set is obtained after processing the first transmitted light signal and the second transmitted light signal. In particular, the first data set may comprise a first training data set comprising the processed first transmitted light signal and the first non-absorptive baseline signal and a first test data set comprising the processed second transmitted light signal and the second non-absorptive baseline signal.
And after the first database is established, a baseline fitting neural network may be established.
The establishment of the baseline fitting neural network mainly takes transmission spectrum signals of gases to be measured with different concentrations as input signals, and corresponding non-absorption baseline signals are taken as output signals of the baseline fitting neural network.
Step S300: and training and testing the baseline fitting neural network by adopting the first data set to obtain a baseline fitting signal.
The input of the baseline fitting neural network is a transmitted light signal, the output of the baseline fitting neural network is a non-absorption baseline signal, and the transmitted light signal and the non-absorption baseline signal are obtained through detection light signal processing.
Specifically, the baseline fitting neural network may be trained using a first training data set in the first data set, and then the trained baseline fitting neural network may be tested using a first test data set in the first data set.
Fig. 3 is a schematic flow chart of training and testing a baseline fitting neural network according to an embodiment of the present application.
Referring to fig. 3, in one specific implementation, step S300 may be implemented by the following steps S301 to S306.
Step S301: the detector detects the obtained transmission spectrum signal (training data), and the target baseline signal (label) corresponding to the transmission spectrum signal.
The transmission spectrum signal is a transmission light signal, and the target baseline signal is a non-absorption baseline signal.
In step S301, training data is used as input data for the baseline fitting neural network, and a label is used as output data for the baseline fitting neural network.
Step S302: the initialized baseline fits the neural network.
Step S303: training.
Step S304: a baseline signal was fitted.
The fitted baseline signal is the baseline fitting signal.
Step S305: and judging whether the mean square error between the fitted base line and the target base line meets the requirement. If yes, the training and testing process is completed, and if no, step S306 is executed.
Step S306: and adjusting the structure and parameters of the baseline fitting neural network. And performs step S303.
In step S306, a Mean Square Error (MSE) may be used to determine whether the trained network meets the requirements, where the mean square error expression is as follows:
；
Wherein, Fitting the output value of the neural network to the predicted value, i.e., the baseline after training,/>Is the label value, i.e. the target baseline signal, i.e. the second non-absorbing baseline signal, i is the i-th spectral data, n is the number of input spectral data.
Specifically, after training is completed, the baseline fitting neural network may also be tested through the first test data set, thereby achieving optimization of the baseline fitting neural network.
Step S400: a second data set and a noise suppression neural network are established.
Wherein the second data set comprises an absorbance signal, the absorbance signal being a ratio of the non-absorbing baseline signal to the transmitted light signal.
Specifically, in step S400, the second data set may be established first, and then the noise suppression neural network may be established.
Establishing the second data set may be achieved by the following steps S401 to S402.
Step S401: and acquiring a plurality of noise reduction training data and a plurality of noise reduction test data.
The noise reduction training data comprises a first absorbance signal and a first high signal-to-noise ratio absorbance signal or a first noise-free simulation absorbance signal, and the noise reduction test data comprises a second absorbance signal and a second high signal-to-noise ratio absorbance signal or a second noise-free simulation absorbance signal. The first absorbance signal may be calculated from the first non-absorption baseline signal to the first transmission light signal, and the second absorbance signal may be calculated from the second non-absorption baseline signal to the second transmission light signal.
The noise-free simulated absorbance signals are preferentially output, and when the noise-free simulated absorbance signals cannot be output or the noise-free simulated absorbance signals cannot be output better, the high signal-to-noise ratio absorbance signals are selected to be output. Because there is sometimes a non-negligible difference between the original absorbance signal and the simulated absorbance signal under the same condition, the simulated absorbance signal at this time cannot be used as the output signal of the noise suppression neural network, and at this time, a high-concentration original absorbance signal having a linear relationship with the input signal is often used as the output signal by a linear scaling method. Wherein the original absorbance signal represents an absorbance signal which is not subjected to noise reduction processing, that is, an input signal of the noise suppression neural network.
Fig. 4 is a schematic diagram of a comparison of a simulated and original absorbance signal provided by an embodiment of the application.
Referring to fig. 4, taking the absorbance signal of the mixed gas of 50ppm ethane and 40ppm methane at normal temperature and pressure as an example, the case that there is a significant difference between the simulated absorbance of the mixed gas and the original absorbance signal is shown.
The direct absorption spectrum technology is based on the beer-lambert law, when the wavelength of light emitted by a laser covers the characteristic absorption spectrum line of the gas to be detected, photon energy at the response wavelength can be absorbed by the gas molecule to be detected, so that the light power of the emergent light on the detection end is attenuated, and the following formula is satisfied:
；
Wherein the method comprises the steps of The transmitted light intensity is represented, and the transmitted light intensity is the detection signal after the absorption of the gas to be detected; /(I)The emergent light intensity is represented, and the detection signal is obtained when nitrogen is introduced in the actual detection; /(I)Is the outgoing light frequency of the laser,/>For the absorption coefficient,/>For the concentration of the gas to be measured,/>Is an effective optical path.
In actual detection, the gas concentration is inverted by absorbance, and the absorbance is generally used for detectionThe definition is as follows:
；
wherein S (T) is the intensity of an absorption spectrum line, and T is the temperature; p is the pressure intensity of the light, As a lorentz linear function.
From the above formula, it can be found that, under otherwise identical conditions, there is a linear relationship between the absorbance signal and the gas concentration. Therefore, in theory, the absorbance signal with low concentration can be obtained by linearly scaling the absorbance signal with high concentration according to the concentration ratio, and since the system noise is fixed and the absorbance signal is larger under the condition of high concentration, the noise is relatively smaller, the absorbance signal with higher signal-to-noise ratio can be obtained by the method.
Step S402: and selecting the noise reduction training data and the noise reduction test data according to the principle of selecting the second features to obtain a second data set.
The second characteristic selection principle is that the values corresponding to the absorption peaks in the first absorbance signal and the second absorbance signal can be fitted through a Lorentz linear function, and the values of the rest parts are set to be zero.
In particular, lorentzian linear functionsThe method comprises the following steps:
；
is the center frequency of the molecular absorbance spectral line of the gas to be detected,/> Is the full width at half maximum of the molecular absorbance spectrum line of the gas to be measured,/>Is the outgoing light frequency of the laser.
Fig. 5a is a schematic diagram of the original absorbance signal of the mixed gas according to the embodiment of the present application.
Fig. 5b is a schematic diagram of absorbance of the scaled mixed gas according to the embodiment of the present application.
For the problems in fig. 4 described above, see fig. 5a and 5b, the original absorbance signal of the mixture of 500ppm ethane and 400ppm methane can be corrected by reducing the absorbance signal by a factor of 10, which approximates the absorbance signal of the mixture of 50ppm ethane and 40ppm methane with high signal-to-noise ratio.
Thus, the steps of detecting and processing the gas to be detected with different concentration gradients are carried out for a plurality of times, and a plurality of noise reduction training data and a plurality of noise reduction test data can be obtained preliminarily. And obtaining a second data set after the second characteristic is selected.
Fig. 6 is a schematic diagram of feature selection using a second feature selection principle according to an embodiment of the present application.
In a specific implementation, referring to fig. 6, the abscissa is the wave number, the ordinate is the absorbance, and the solid curve in fig. 6 is the original absorbance signal, but because the absorption of the gas to be detected exists in the range of the using wave band, the original signal is all greater than 0, so in the embodiment of the application, the absorption peak is corrected, pulled down to be near 0 value, and finally the lorentz linear function is used to fit the absorption peak to obtain a graph dash-dot curve, and the dash-dot curve is the input signal of the noise suppression neural network. In this example, the absorbance signal of a mixed gas of 50ppm ethane and 40ppm methane, wherein the absorption peaks at the positions 2986.6cm -1 and 2990cm -1 are ethane absorption peaks, so that all information about the concentration of ethane gas can be obtained by fitting only one of the absorption peaks. The methane at the position of 2989cm -1 has a plurality of absorption peaks, and the shape of the absorption peaks does not conform to the Lorentzian line type due to superposition of the plurality of absorption peaks, but part of the absorption peaks still conform to the Lorentzian line type, so that only part of the methane peaks need to be subjected to the Lorentzian line type fitting operation. Thus, feature selection is completed.
Thus, the second data set is obtained after processing the first absorbance signal and the second absorbance signal. Specifically, the second data set may include a second training data set and a second test data set, where the second training data set includes the processed first absorbance signal and the first high signal-to-noise absorbance signal or the noiseless first simulated absorbance signal; the second test data set includes the processed second absorbance signal and either the second high signal to noise absorbance signal or the noise free second simulated absorbance signal.
And after the second database is established, a noise suppression neural network may be established.
The noise suppression neural network is established to establish a mapping relationship between the first absorbance signal and the second absorbance signal converted from the detected optical signal, and the absorbance signal with high signal-to-noise ratio or without noise. The first absorbance signal and the second absorbance signal are used as input signals of the noise suppression neural network, and the output signals generally adopt absorbance signals simulated under the same condition as the input signals or absorbance signals with high signal to noise ratio after linear scaling. The high signal-to-noise ratio absorbance signal is obtained through absorbance signal processing, and the noise-free simulated absorbance signal is obtained through simulation program processing.
In a specific implementation, the training methods of the baseline fitting neural network and the noise suppression neural network are the same, and the baseline fitting neural network and the noise suppression neural network each include an input layer, a first hidden layer, a second hidden layer, a third hidden layer, and an output layer.
The activation functions between the input layer and the first hidden layer, between the first hidden layer and the second hidden layer and between the second hidden layer and the third hidden layer are Sigmoid functions:
；
Wherein, Is a natural constant,/>Input spectral data for the input layer or the first hidden layer or the second hidden layer;
The activation function between the third hidden layer and the output layer is ; Wherein/>And outputting spectral data for the third hidden layer.
The loss functions of the baseline fitting neural network and the noise suppression neural network may be the same, and both may be mean square error functions as the loss functions.
Step S500: and training and testing the noise suppression neural network by adopting the second data set to obtain the noise reduction signal.
The input of the noise suppression neural network is an absorbance signal, and the output of the noise suppression neural network is a high signal-to-noise ratio absorbance signal or a noise-free simulation absorbance signal.
Specifically, the noise suppression neural network is trained by using a second training data set in the second data set, and then the trained noise suppression neural network is tested by using a second test data set in the second data set.
In a specific implementation, the training methods of the baseline fitting neural network and the noise suppression neural network may be the same, and a bayesian regularization method may be used. In Bayesian regularization training algorithm, a priori distribution of model parameters needs to be defined first. Wherein/>For/>Network parameters of the layer. The posterior distribution of each layer of network is then calculated by observing the data set D according to the bayesian formula. As/>The output of the layer network model is/>The corresponding target value is/>。
According to the Bayesian formula, there are:
；
Wherein, Representing the probability that a target value is observed given the model output and parameters.
Further, it is necessary to maximize the posterior probability, i.e., to select a parameter value with the greatest posterior probability. Then, the parameters are updated according to the maximized posterior probability result.
And finally, applying the obtained model to test the effect of the model. And adjusting and optimizing according to the experimental result.
Fig. 7 is a schematic flow chart of training and testing a noise suppression neural network according to an embodiment of the present application.
Referring to fig. 7, in a specific implementation, step S500 may be implemented by the following steps S501 to S506.
Step S501: noisy absorbance signal (training data), noiseless absorbance signal (label). The absorbance signal containing noise is the original absorbance signal, namely the ratio of the absorbance signal without absorption baseline signal to the transmitted light signal, and the absorbance signal without noise is the absorbance signal with high signal to noise ratio or the simulated absorbance signal without noise.
In step S501, training data is used as input data of the noise suppression neural network, and a tag is used as output data of the noise suppression neural network.
Step S502: the initialized noise suppressing neural network.
Step S503: training.
Step S504: and (5) reducing the absorbance signal after noise.
Step S505: and judging whether the signal to noise ratio requirement is met. If yes, the training and testing process is completed, and if no, step S506 is executed.
Step S506: and adjusting the structure and parameters of the noise suppression neural network. And performs step S503.
In step S506, the signal-to-noise ratio function SNR may be employed as:
；
Wherein, Absorption peak value of second absorbance signal,/>Is the non-absorption peak portion of the second absorbance signal.
Specifically, after the trained noise suppression neural network meets the condition of the signal-to-noise ratio function SNR, the training and testing process of the noise suppression neural network is completed.
In some possible implementations, the baseline fitting and noise reduction method may further include step S600.
Step S600: and calculating the concentration of the gas to be detected according to the noise reduction signal.
In step S600, after training and testing the two neural networks, the concentration of the gas to be measured can be calculated by the obtained noise reduction signal.
Fig. 8 is a flowchart of a specific application of a baseline fitting and noise reduction method for a gas laser absorption spectrum according to an embodiment of the present application.
Referring to fig. 8, in a specific implementation, the calculation of the concentration of the gas to be measured may be performed by the baseline fitting and noise reduction method of the gas laser absorption spectrum provided by the embodiment of the present application, which may be specifically implemented by the following steps S001 to S600-1.
Step S001: the transmitted light signal obtained by the detector.
Specifically, the transmitted light signal may be converted from the detected light signal.
Step S201-1: and selecting the characteristics of the transmitted light signals.
Wherein step S201-1 may be a sub-step in step S201.
Step S300-1: the trained baselines fit to the neural network.
Step S300-2: a baseline signal was fitted.
The steps 300-1 and 300-2 may be two steps in the step S300.
Step S401-1: and comparing the fitted baseline signal with the transmitted light signal to obtain an absorbance signal of the gas to be detected.
After the fitted baseline signal is obtained, the fitted baseline signal is required to be compared with the transmission spectrum signal, and the original absorbance signal is obtained.
Step S402-1: the absorbance signal is feature selected.
The steps S401-1 and S402-1 may be two steps in the step S400.
Step S501-1: the trained noise suppressing neural network.
Step S502-1: and (5) denoising the absorbance signal of the gas to be detected.
The steps S501-1 and S502-1 may be two steps in the step S500.
Step S600-1: and inverting the concentration of the gas to be detected according to the peak value of the absorption peak of the gas to be detected in the absorbance signal.
Wherein, the step S600-1 may be the step in the step S600.
Fig. 9 is a schematic diagram showing the effect of baseline fitting according to an embodiment of the present application.
Referring to fig. 9, the signals after feature selection are input into a trained and tested baseline fitting neural network, the baseline fitting neural network outputs baseline fitting signals, and the fitting effect is shown in fig. 9.
Fig. 10a is a schematic diagram of an original absorbance signal according to an embodiment of the application.
Fig. 10b is a schematic diagram of an absorbance signal after noise reduction according to an embodiment of the application.
Referring to fig. 10a and 10b, the absorbance signal after feature selection is used as an input signal of the noise suppression neural network, and the training and testing noise suppression neural network outputs the absorbance signal after noise reduction.
And (5) inverting the concentration of the gas to be detected according to the peak value of the absorbance signal after noise reduction. Wherein, the SNR of methane is improved from 16.6dB to 20.5dB before noise reduction, and the SNR of ethane is improved from 22.69dB to 26.26dB before noise reduction according to the peak value of methane and ethane.
Specifically, the baseline fitting and noise reduction method for the gas laser absorption spectrum provided by the embodiment of the application can be used for adaptively fitting a baseline signal according to a transmitted light signal, so that the problems of baseline signal drift and noise interference existing in gas concentration detection are solved, and the noise in an absorbance signal can be effectively inhibited. The method does not need to add extra hardware equipment, does not introduce extra hardware noise, has no frequency limitation on noise, and has the advantages of quick processing and good noise reduction effect compared with the existing processing method. The method is not limited by the structure, is suitable for small-sized systems and short-path equipment, effectively reduces the equipment cost and has a wider application range.
Fig. 11 is a block diagram of a baseline fitting and noise reduction system for gas laser absorption spectrum according to an embodiment of the present application.
Referring to fig. 11, corresponding to the above embodiment, the present application further provides an embodiment of a baseline fitting and noise reduction system 100 for a gas laser absorption spectrum, which is applied to the baseline fitting and noise reduction method provided in any one of the above embodiments. The baseline fitting and noise reduction system 100 includes a transmission module 101, a first setup module 102, a first training and testing module 103, a second setup module 104, and a second training and testing module 105.
The transmitting module 101 is configured to transmit a detection light signal to a gas to be detected.
The first setup module 102 is for establishing a first dataset and a baseline fitting neural network.
The first training and testing module 103 is configured to train and test the baseline fitting neural network using the first data set to obtain a baseline fitting signal; the input of the baseline fitting neural network is a transmitted light signal, the output of the baseline fitting neural network is a non-absorption baseline signal, and the transmitted light signal and the non-absorption baseline signal are obtained through detection light signal processing.
The second establishing module 104 is configured to establish a second data set and a noise suppression neural network; wherein the second data set comprises absorbance signals. The absorbance signal is the ratio of the non-absorbing baseline signal to the transmitted light signal.
The second training and testing module 105 is configured to train and test the noise suppression neural network using the second data set to obtain a noise reduction signal; the input of the noise suppression neural network is an absorbance signal, the output of the noise suppression neural network is a high signal-to-noise ratio absorbance signal or a noise-free simulation absorbance signal, and the high signal-to-noise ratio absorbance signal or the noise-free simulation absorbance signal is obtained through absorbance signal processing.
Fig. 12 is a schematic structural diagram of a baseline fitting and noise reduction system according to an embodiment of the present application.
In a specific implementation, referring to fig. 12, a modulation signal generated by a signal generator is input into a laser controller, so as to tune the laser, so that an output wavelength of the laser covers a target detection band, then an outgoing detection light signal of the laser is collimated by a collimating lens and then is injected into a gas chamber, the gas chamber is filled with gas to be detected, the detected light signal is converted into an electric signal by a photoelectric detection module and amplified, and finally a data acquisition card acquires a transmission light signal and uploads the transmission light signal to a computer.
The laser may be the transmitting module 101 in the foregoing embodiment, and the computer may include the first establishing module 102, the first training and testing module 103, the second establishing module 104, and the second training and testing module 105 provided in the foregoing embodiment. For regulating the temperature and pressure, the air chamber may be provided with a thermometer and a manometer.
The computer may further include a calculation module for calculating a concentration of the gas to be measured according to the noise reduction signal.
Specifically, the baseline fitting and noise reduction system for the gas laser absorption spectrum provided by the embodiment of the application can adaptively fit a baseline signal according to a transmitted light signal, so that the problems of baseline signal drift and noise interference existing in gas concentration detection are solved, and noise in an original absorbance signal can be effectively inhibited. The miniaturization of the equipment can be realized, and the production cost is effectively reduced.
It is noted that other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. This application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the application pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope of the application being indicated by the following claims.
It is to be understood that the application is not limited to the precise arrangements and instrumentalities shown in the drawings, which have been described above, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the application is limited only by the appended claims.
Claims (8)
1. A baseline fitting and noise reduction method for a gas laser absorption spectrum, comprising:
Transmitting a detection light signal to the gas to be detected;
establishing a first data set and a baseline fitting neural network;
Training and testing the baseline fitting neural network by adopting the first data set to obtain a baseline fitting signal; the input of the baseline fitting neural network is a transmitted light signal, the output of the baseline fitting neural network is a non-absorption baseline signal, and the transmitted light signal and the non-absorption baseline signal are obtained through the detection light signal processing;
Establishing a second data set and a noise suppression neural network; wherein the second data set comprises absorbance signals, the absorbance signals being the ratio of the non-absorbing baseline signal to the transmitted light signal;
Training and testing the noise suppression neural network by adopting the second data set to obtain a noise reduction signal; the input of the noise suppression neural network is the absorbance signal, the output of the noise suppression neural network is a high signal-to-noise ratio absorbance signal or a noise-free simulation absorbance signal, the high signal-to-noise ratio absorbance signal is obtained through the absorbance signal processing, and the noise-free simulation absorbance signal is obtained through the simulation program processing;
the establishing a first data set includes:
acquiring a plurality of baseline training data and a plurality of baseline test data; the baseline training data and the baseline test data are obtained through the detection light signals, the baseline training data comprise a first transmission light signal and a first non-absorption baseline signal, and the baseline test data comprise a second transmission light signal and a second non-absorption baseline signal;
Selecting the baseline training data and the baseline test data according to a first characteristic selection principle to obtain the first data set;
the establishing the second data set includes:
Acquiring a plurality of noise reduction training data and a plurality of noise reduction test data; the noise reduction training data comprises a first absorbance signal and a first high signal-to-noise ratio absorbance signal or a first noise-free simulation absorbance signal, and the noise reduction test data comprises a second absorbance signal and a second high signal-to-noise ratio absorbance signal or a second noise-free simulation absorbance signal;
Selecting the noise reduction training data and the noise reduction test data according to a second characteristic selection principle to obtain a second data set;
the first characteristic selection principle is to set values corresponding to absorption peaks of the first transmission light signal and the second transmission light signal to zero;
the second characteristic selection principle is to fit absorption peaks of the first absorbance signal and the second absorbance signal by using a Lorentzian linear function.
2. The method for baseline fitting and noise reduction of gas laser absorption spectrum according to claim 1, wherein,
The Lorentz linear functionThe method comprises the following steps:
for the center frequency of the molecular absorbance spectrum line of the gas to be detected,/> For the full width at half maximum of the molecular absorbance spectrum line of the gas to be detected,/>Is the outgoing light frequency of the laser.
3. The method for baseline fitting and noise reduction of gas laser absorption spectrum according to claim 2, wherein,
The baseline fitting neural network and the noise suppression neural network each comprise an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer; the activation functions between the input layer and the first hidden layer, between the first hidden layer and the second hidden layer, and between the second hidden layer and the third hidden layer are Sigmoid functions:
wherein e is a natural constant,/> Input spectral data for the input layer or the first hidden layer or the second hidden layer;
the activation function between the third hidden layer and the output layer is that ; Wherein/>And outputting spectral data for the third hidden layer.
4. The method for baseline fitting and noise reduction of gas laser absorption spectrum according to claim 3,
The baseline fitting neural network and the noise suppression neural network have the same loss function;
the baseline fitting neural network and the noise suppression neural network are the same in training method.
5. The method for baseline fitting and noise reduction of gas laser absorption spectrum according to claim 4, wherein,
The baseline fitting neural network is tested by adopting a Mean Square Error (MSE), wherein the MSE is as follows:
Wherein/> Fitting the output values of the neural network to the baseline after training,Is the second non-absorptive baseline signal; i is the ith spectral data, and n is the number of input spectral data.
6. The method for baseline fitting and noise reduction of gas laser absorption spectrum according to claim 5, wherein,
The noise suppression neural network is tested by adopting a signal-to-noise ratio function, and the SNR of the signal-to-noise ratio function is as follows:
Wherein/> For the absorption peak value of the second absorbance signal,/>Is the non-absorption peak portion of the second absorbance signal.
7. The method for baseline fitting and noise reduction of gas laser absorption spectrum according to claim 1, wherein,
The baseline fitting and noise reduction method for the gas laser absorption spectrum further comprises the following steps: and calculating the concentration of the gas to be detected according to the noise reduction signal.
8. A baseline fitting and noise reduction system for a gas laser absorption spectrum, characterized in that it is applied to the baseline fitting and noise reduction method for a gas laser absorption spectrum according to any one of claims 1 to 7, and the baseline fitting and noise reduction system for a gas laser absorption spectrum comprises:
A transmitting module; configured to transmit a detection light signal into a gas to be detected;
a first establishing module configured to establish a first data set and a baseline fitting neural network;
The first training and testing module is configured to train and test the baseline fitting neural network by adopting the first data set to obtain a baseline fitting signal; the input of the baseline fitting neural network is a transmitted light signal, the output of the baseline fitting neural network is a non-absorption baseline signal, and the transmitted light signal and the non-absorption baseline signal are obtained through the detection light signal processing;
A second establishing module configured to establish a second data set and a noise suppressing neural network; wherein the second data set comprises an absorbance signal that is a ratio of the non-absorbing baseline signal to the transmitted light signal;
The second training and testing module is configured to train and test the noise suppression neural network by adopting the second data set to obtain a noise reduction signal; the input of the noise suppression neural network is the absorbance signal, the output of the noise suppression neural network is a high signal-to-noise ratio absorbance signal or a noise-free simulation absorbance signal, the high signal-to-noise ratio absorbance signal is obtained through the absorbance signal processing, and the noise-free simulation absorbance signal is obtained through the simulation program processing;
the establishing a first data set includes: acquiring a plurality of baseline training data and a plurality of baseline test data; the baseline training data and the baseline test data are obtained through the detection light signals, the baseline training data comprise a first transmission light signal and a first non-absorption baseline signal, and the baseline test data comprise a second transmission light signal and a second non-absorption baseline signal; selecting the baseline training data and the baseline test data according to a first characteristic selection principle to obtain the first data set;
The establishing the second data set includes: acquiring a plurality of noise reduction training data and a plurality of noise reduction test data; the noise reduction training data comprises a first absorbance signal and a first high signal-to-noise ratio absorbance signal or a first noise-free simulation absorbance signal, and the noise reduction test data comprises a second absorbance signal and a second high signal-to-noise ratio absorbance signal or a second noise-free simulation absorbance signal; selecting the noise reduction training data and the noise reduction test data according to a second characteristic selection principle to obtain a second data set;
the first characteristic selection principle is to set values corresponding to absorption peaks of the first transmission light signal and the second transmission light signal to zero; the second characteristic selection principle is to fit absorption peaks of the first absorbance signal and the second absorbance signal by using a Lorentzian linear function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410171508.1A CN117747008B (en) | 2024-02-07 | 2024-02-07 | Baseline fitting and noise reduction method and system for gas laser absorption spectrum |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410171508.1A CN117747008B (en) | 2024-02-07 | 2024-02-07 | Baseline fitting and noise reduction method and system for gas laser absorption spectrum |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117747008A CN117747008A (en) | 2024-03-22 |
| CN117747008B true CN117747008B (en) | 2024-05-03 |
Family
ID=90257774
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410171508.1A Active CN117747008B (en) | 2024-02-07 | 2024-02-07 | Baseline fitting and noise reduction method and system for gas laser absorption spectrum |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117747008B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2092713A1 (en) * | 1990-10-12 | 1992-04-13 | James Milton Brown | Spectral data measurement and correction |
| JP2012137429A (en) * | 2010-12-27 | 2012-07-19 | Mitsubishi Heavy Ind Ltd | Laser measuring device |
| CN104535528A (en) * | 2014-11-26 | 2015-04-22 | 东南大学 | Method for real time extraction of TDLAS gas absorption spectrum absorbance by BP neural network |
| CA2915137A1 (en) * | 2014-12-26 | 2016-06-26 | General Electric Company | Systems and methods to analyze a gas-mixture |
| CN108981953A (en) * | 2018-10-24 | 2018-12-11 | 北京航空航天大学 | A kind of laser absorption spectrum temp measuring method and system based on interference modulations principle |
| CN110987870A (en) * | 2019-12-20 | 2020-04-10 | 天津同阳科技发展有限公司 | System and method for monitoring gas concentration in real time based on wavelength modulation spectrum technology |
| CN111678599A (en) * | 2020-07-07 | 2020-09-18 | 安徽大学 | Laser spectrum noise reduction method and device based on deep learning optimization S-G filtering |
| WO2021055902A1 (en) * | 2019-09-20 | 2021-03-25 | Seekops Inc. | Spectral fitting of compact laser-based trace gas sensor measurements for high dynamic range (hdr) |
| CN114993990A (en) * | 2022-08-03 | 2022-09-02 | 安徽岑锋科技有限公司 | Integrated small open-circuit greenhouse gas flux monitoring method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9068933B2 (en) * | 2012-06-08 | 2015-06-30 | Ut-Battelle, Llc | EGR distribution and fluctuation probe based on CO2 measurements |
| US11614402B2 (en) * | 2019-03-07 | 2023-03-28 | The Regents of the Univ. of Colorado, a body corp. | Baseline-free quantitative absorption spectroscopy |
-
2024
- 2024-02-07 CN CN202410171508.1A patent/CN117747008B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2092713A1 (en) * | 1990-10-12 | 1992-04-13 | James Milton Brown | Spectral data measurement and correction |
| JP2012137429A (en) * | 2010-12-27 | 2012-07-19 | Mitsubishi Heavy Ind Ltd | Laser measuring device |
| CN104535528A (en) * | 2014-11-26 | 2015-04-22 | 东南大学 | Method for real time extraction of TDLAS gas absorption spectrum absorbance by BP neural network |
| CA2915137A1 (en) * | 2014-12-26 | 2016-06-26 | General Electric Company | Systems and methods to analyze a gas-mixture |
| CN108981953A (en) * | 2018-10-24 | 2018-12-11 | 北京航空航天大学 | A kind of laser absorption spectrum temp measuring method and system based on interference modulations principle |
| WO2021055902A1 (en) * | 2019-09-20 | 2021-03-25 | Seekops Inc. | Spectral fitting of compact laser-based trace gas sensor measurements for high dynamic range (hdr) |
| CN110987870A (en) * | 2019-12-20 | 2020-04-10 | 天津同阳科技发展有限公司 | System and method for monitoring gas concentration in real time based on wavelength modulation spectrum technology |
| CN111678599A (en) * | 2020-07-07 | 2020-09-18 | 安徽大学 | Laser spectrum noise reduction method and device based on deep learning optimization S-G filtering |
| CN114993990A (en) * | 2022-08-03 | 2022-09-02 | 安徽岑锋科技有限公司 | Integrated small open-circuit greenhouse gas flux monitoring method |
Non-Patent Citations (4)
| Title |
|---|
| Design of Resolution-Tunable Neural Network-Based Integrated Reconstructive Spectrometer;Anish Mondal等;IEEE Sensors Journal;20211230;第22卷(第3 期);全文 * |
| 基于光谱吸收的CO2气体检测系统;宫卫华 等;山东科学;20221231;第35卷(第6期);全文 * |
| 基于深度学习的红外光谱气体智能检测技术研究;田遴博;万方学位论文;20231007;全文 * |
| 基于超光谱层析成像技术重建燃烧场温度及气体浓度二维分布;臧益鹏;中国博士学位论文全文数据库;20210915;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117747008A (en) | 2024-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6981817B2 (en) | Spectroscopic analyzer and spectroscopic analysis method | |
| CN116818739A (en) | Indocyanine green detection method based on optics | |
| US20230394651A1 (en) | Defect detecting apparatus and method | |
| CN110261328A (en) | Calibrate method and device, the gas concentration analyzer of optical maser wavelength | |
| CN110736719A (en) | gas concentration calibration and measurement method based on TDLAS direct absorption spectrum | |
| CN117747008B (en) | Baseline fitting and noise reduction method and system for gas laser absorption spectrum | |
| Weigl et al. | Effects of ambient parameters and cross-sensitivities from O2, CO2 and H2O on the photoacoustic detection of acetone in the UV region | |
| Zhang et al. | SNR enhancement for Brillouin distributed optical fiber sensors based on asynchronous control | |
| CN112378883B (en) | TDLAS gas concentration calibration method based on relative error least square method | |
| US20160238461A1 (en) | Temperature measuring system and temperature measuring method | |
| Hongyan et al. | Wavefront reconstruction of vortex beam propagation in atmospheric turbulence based on deep learning | |
| CN113496100A (en) | System and computer-implemented method for extrapolating a calibration spectrum | |
| Lu et al. | Beam-steering-effect immune laser absorption spectrum extraction for precise and robust temperature measurement | |
| CN112539772A (en) | Positioning method of Sagnac distributed optical fiber sensing system based on convolutional neural network integrated learning | |
| Wang et al. | Interference fringe suppression in tunable diode laser absorption spectroscopy based on CEEMDAN-WTD | |
| CN116840187A (en) | TDLAS joint detection method | |
| CN115792136A (en) | Gas concentration detection method and device, terminal equipment and storage medium | |
| CN114826396B (en) | Method and device for controlling pumping by monitoring Raman amplifier reflected power | |
| Shang et al. | Pink noise removal and spectral distortion correction based fiber Bragg grating demodulation algorithm | |
| CN114038371B (en) | Gamma adjustment method, related device and storage medium | |
| Zhu et al. | Improved Savitzky-Golay filtering algorithm for measuring a pharmaceutical vial’s oxygen content based on wavelength modulation spectroscopy | |
| CN115266583A (en) | Ambient light filtering method, ambient light filtering system, computer device and computer readable storage medium | |
| US11614402B2 (en) | Baseline-free quantitative absorption spectroscopy | |
| CN114295577B (en) | Terahertz detection signal processing method, device, equipment and medium | |
| Qi et al. | Optimized VMD algorithm for signal noise reduction based on TDLAS |
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 |