Interference elimination and temperature detection method and system based on second harmonic

Abstract

The invention relates to the technical field of temperature detection, and provides an interference elimination and temperature detection method and system based on second harmonics. The method comprises: establishing an interference elimination model; training an interference elimination model by adopting the training data set; the training data set comprises a second harmonic signal of the training mixed gas, and a second harmonic signal of the first gas and a second harmonic signal of the second gas corresponding to the second harmonic signal of the training mixed gas; the training mixed gas comprises a first gas and a second gas, and inputting a real-time second harmonic signal of the collected mixed gas into the trained interference elimination model to obtain a real-time second harmonic signal of the first gas; the temperature of the first gas is inverted based on the real-time second harmonic signal of the first gas. The method can effectively ensure the accuracy of temperature detection and improve the detection efficiency.

Description

Technical Field

[0001] The present application relates to the field of temperature detection technology, and in particular to a second harmonic-based interference elimination and temperature detection method and system. Background Art

[0002] Photoacoustic spectroscopy (PAS) is a highly sensitive detection method based on the photoacoustic effect. The photoacoustic effect refers to the heat generated by gas after absorbing light energy, which causes the local temperature to rise, which in turn causes the thermal expansion of the gas and generates sound waves. Photoacoustic spectroscopy can analyze the absorption characteristics of a substance and its relationship with temperature by detecting the intensity and frequency characteristics of this sound wave signal. This method was first used for gas detection, and with the advancement of technology, its application scope has continued to expand to multiple fields such as component analysis and biomedical imaging.

[0003] Common temperature detection methods include thermocouples, infrared temperature measurement, resistance temperature detectors (RTD), thermistors, and fiber optic temperature measurement. Thermocouples and RTDs require direct contact with the sample and have a slow response speed; although infrared temperature measurement is non-contact, it is easily interfered by ambient light; and although fiber optic temperature measurement is suitable for harsh environments, it is more expensive. In contrast, photoacoustic spectroscopy temperature detection has unique advantages. Its high sensitivity and selectivity enable photoacoustic technology to detect temperature changes in specific areas, reduce interference from external factors, and make the detection results more accurate. Photoacoustic spectroscopy also has a fast response characteristic, can monitor temperature changes in real time in a dynamic environment, and has broad application prospects in harsh environments such as high temperature and high pressure.

[0004] In photoacoustic spectroscopy, gases will produce different absorption peaks at different temperatures. By detecting the second harmonic amplitude information generated by the gas in the current state and performing inversion analysis, the temperature information of the temperature measurement area can be obtained. However, since different gases have similar absorption characteristics in certain bands, when the target gas used for temperature detection is combined with other non-target gases that also absorb in this band, the light absorption of the non-target gas will produce additional photoacoustic signals, resulting in complexity and uncertainty of the detection signal. Especially in a multi-component gas environment, overlapping absorption bands will make it difficult to extract the signal of the target gas alone, thus affecting the accuracy of temperature measurement. Summary of the invention

[0005] The present application provides a second harmonic-based interference elimination and temperature detection method and system, which can effectively ensure the accuracy of temperature detection and improve detection efficiency.

[0006] The first aspect of the present application provides a second harmonic-based interference elimination and temperature detection method, comprising: establishing an interference elimination model; using a training data set to train the interference elimination model; the training data set includes a second harmonic signal of a training mixed gas and a second harmonic signal of a first gas and a second harmonic signal of a second gas corresponding to the second harmonic signal of the training mixed gas; the training mixed gas includes a first gas and a second gas, the second gas being an interference gas of the first gas; inputting the real-time second harmonic signal of the collected mixed gas into a trained convolutional neural network to obtain a real-time second harmonic signal of the first gas; and inverting the temperature of the first gas based on the real-time second harmonic signal of the first gas.

[0007] In some feasible implementations, the interference cancellation model includes a plurality of sequentially connected convolution combination layers, where the convolution combination layer includes a connected convolution layer and an activation layer; the convolution formula of the convolution layer is: ; in, is the i-th element in the convolution output, k is the size of the convolution kernel, is the jth weight parameter of the convolution kernel, i and j are positive integers, and the second harmonic signal x =[ x 1 , x 2 ,⋯, x n ] , convolution kernel w =[ w 1 , w 2 ,⋯, w k ] , n is the magnitude of the second harmonic signal x, is the bias term; the activation function of the activation layer is the ReLU function.

[0008] In some feasible implementations, the interference elimination model further includes a pooling layer connected to the convolution combination layer, the pooling layer is a maximum pooling layer, and the pooling formula of the pooling layer is: ; in, is the pooling result of the i-th layer, and p is the size of the pooling window.

[0009] In some feasible implementations, the interference cancellation model also includes a fully connected layer connected to the pooling layer, and the output formula of the fully connected layer is: ; ; is the second harmonic signal of the first gas output by the interference elimination model, A second harmonic signal of the second gas output by the interference elimination model; is a linear activation function, ; Extract the jth input feature of the second harmonic signal in the mth layer; is the jth weight parameter of the first gas in the mth layer, is the jth weight parameter of the second gas in the mth layer, is the bias term of the first gas in the mth layer, is the bias term of the second gas of the mth layer; where m is the number of convolutional combination layers, and m is less than or equal to 5.

[0010] In some feasible implementations, the second harmonic-based interference elimination and temperature detection method also includes: normalizing the real-time second harmonic signal of the collected mixed gas; inverting the temperature of the first gas based on the real-time second harmonic signal of the first gas includes: denormalizing the real-time second harmonic signal of the first gas; and inverting the temperature of the first gas based on the real-time second harmonic signal of the first gas after denormalization.

[0011] In some feasible implementations, the normalization formula is: ; X is the original signal, X max is the maximum value of the original signal, X min is the minimum value of the original signal, X nom is the normalized signal; the formula for denormalization is: ; is the anti-normalized signal; is the scaling factor.

[0012] In some feasible implementations, the loss function of the interference cancellation model is: ; is the actual second harmonic signal of the first gas, is the actual second harmonic signal of the second gas.

[0013] In some feasible implementations, the first gas is CO2 and the second gas is H2O.

[0014] The interference elimination and temperature detection method based on second harmonic provided in the first aspect of the present application is universal and can be extended to other related technical fields. The main contents include the establishment of training data sets and feature selection, the construction of model structure, the training of interference elimination models, and the use of interference elimination models to calibrate the photoacoustic spectroscopy temperature detection system without adding additional hardware equipment, effectively ensuring the accuracy of temperature detection and improving detection efficiency.

[0015] The second aspect of the present application provides a second harmonic-based interference elimination and temperature detection system, which adopts the second harmonic-based interference elimination and temperature detection method provided by the first aspect. The second harmonic-based interference elimination and temperature detection system includes: an establishment module, which is configured to establish an interference elimination model; a training module, which is configured to: use a training data set to train the interference elimination model; the training data set includes a second harmonic signal of a training mixed gas and a second harmonic signal of a first gas and a second harmonic signal of a second gas corresponding to the second harmonic signal of the training mixed gas; the training mixed gas includes a first gas and a second gas, and the second gas is an interference gas of the first gas; a separation module, which is configured to: input the real-time second harmonic signal of the collected mixed gas into the trained interference elimination model to obtain a real-time second harmonic signal of the first gas; and an inversion module, which is configured to: invert the temperature of the first gas based on the real-time second harmonic signal of the first gas.

[0016] In some feasible implementations, the interference cancellation model includes a convolutional combination layer, a pooling layer, and a fully connected layer connected in sequence, the number of the convolutional combination layers is multiple, and the convolutional combination layer includes connected convolutional layers and activation layers.

[0017] The second aspect of the present application provides a second harmonic-based interference elimination and temperature detection system, which adopts the second harmonic-based interference elimination and temperature detection method provided in the first aspect. Its beneficial technical effects can be found in the first aspect and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0019] Figure 1 It is a flow chart of a method for interference elimination and temperature detection based on second harmonic provided in an embodiment of the present application; Figure 2 This is a schematic diagram of the cross-interference between CO2 and H2O molecules; Figure 3 It is a schematic diagram of the second harmonic signal of CO2 under the same concentration but different temperature conditions; Figure 4 It is a schematic diagram of the second harmonic signal of CO2 interfered by H2O molecules of different concentrations under the same concentration and temperature conditions; Figure 5 It is a flowchart of a detection method in a specific implementation method provided in an embodiment of the present application; Figure 6This is a schematic diagram of the effect of separating the second harmonic signals of CO2 and H2O molecules provided in an embodiment of the present application; Figure 7 It is a linear fitting diagram of the CO2 second harmonic signal and temperature after separation by the interference elimination model provided in the embodiment of the present application; Figure 8 It is a structural block diagram of an interference elimination and temperature detection system based on second harmonic provided by an embodiment of the present application; Fig. 9 It is a structural schematic diagram of an interference elimination and temperature detection system based on second harmonic provided in an embodiment of the present application.

[0020] Graphic marking: 100-interference elimination and temperature detection system based on second harmonic; 10-signal generator; 11-first output terminal; 12-second output terminal; 20-laser controller; 30-distributed feedback laser; 40-microphone; 50-phase-locked amplifier; 60-data acquisition device; 70-processor; 701-establishment module; 702-training module; 703-separation module; 704-inversion module; 200-temperature detection area. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be described clearly below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments of the present application, other embodiments obtained by ordinary technicians in this field without making creative work all belong to the protection scope of the present application.

[0022] In the following, the terms "first", "second", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", etc. may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0023] In addition, in the present application, directional terms such as "upper", "lower", "inner" and "outer" are defined relative to the orientation of the components schematically placed in the drawings. It should be understood that these directional terms are relative concepts. They are used for relative description and clarification, and they can change accordingly according to the changes in the orientation of the components placed in the drawings.

[0024] See also Figure 1 As shown, the interference elimination and temperature detection method based on second harmonic provided in the embodiment of the present application can be implemented by the following steps S1 to S4.

[0025] Step S1: Establish an interference cancellation model.

[0026] Step S2: Using the training data set to train the interference cancellation model.

[0027] The establishment of the training data set may be before or after the establishment of the interference elimination model. The training data set may include two parts, one part is the training data and the other part is the label. The training data includes the second harmonic signal of the training mixed gas. The label includes the second harmonic signal of the first gas and the second harmonic signal of the second gas corresponding to the second harmonic signal of the training mixed gas.

[0028] Specifically, the training data is obtained based on the established second harmonic interference elimination and temperature detection system, and the temperature can be detected by the absorption spectra of the characteristic absorption molecules of different gases. In the temperature detection area, the second harmonic signals of the mixed gas of the first gas and the second gas of different temperatures and concentrations, as well as the experimentally obtained second harmonic signals of the first gas and the second harmonic signals of the second gas are collected. Among them, the training data can be obtained by system detection, and the detection environment of the training data can be indoors or outdoors. The label can be obtained by the system in experimental detection. The first gas and the second gas cannot be detected separately outdoors, so the first gas and the second gas can be obtained by experimental detection in the laboratory respectively.

[0029] In this way, after training is completed, the input of the interference elimination model can be the second harmonic signal of the mixed gas, and the output of the interference elimination model can be the second harmonic signal of the first gas and the second harmonic signal of the second gas, thereby achieving separation of the second harmonic signals of the first gas and the second gas.

[0030] The following uses CO2 as the first gas and H2O as the second gas as an example to introduce the influence of interfering gas on the detection accuracy of the target gas.

[0031] See also Figure 2 As shown in the figure, there is serious interference from H2O molecules at the absorption peak of CO2. When CO2 is used as the target gas to detect temperature, it will be interfered by H2O molecules, resulting in inaccurate test results.

[0032] See also Figure 3 As shown, the second harmonic signal of CO2 changes with the temperature, and the signal amplitude will change. Therefore, the temperature information required for detection can be inverted by measuring the second harmonic signal of CO2.

[0033] See also Figure 4 As shown in the figure, due to the interference of H2O molecules, the amplitude of the second harmonic signal of CO2 is affected, resulting in inaccurate temperature inversion.

[0034] The embodiment of the present application uses a training data set to train the interference elimination model, with the aim of separating the second harmonic signal of the first gas and the second harmonic signal of the second gas from the second harmonic signal of the mixed gas. The first gas can be used as the target gas, and the second gas can be used as the interfering gas of the target gas, thereby avoiding interference of the second gas to the second harmonic signal of the first gas when detecting the second harmonic signal of the first gas, thereby ensuring detection accuracy.

[0035] It is worth noting that the above-mentioned first gas is CO2 and the second gas is H2O is only an example. In other implementations, the first gas and the second gas may also be other different gases.

[0036] Specifically, the established interference elimination model can be a one-dimensional convolutional neural network, including structural layers connected in sequence, and the output value of each structural layer is used as the input value of the next layer.

[0037] The interference removal model consists of a combination of convolutional layers, pooling layers, and fully connected layers.

[0038] There are multiple convolution combination layers, and each convolution combination layer includes a connected convolution layer and an activation layer.

[0039] In a one-dimensional convolution, the convolution kernel w slides over the input second harmonic signal x and performs a dot product operation at each position to form the convolution output. x =[ x 1 , x 2 ,⋯, x n ] and convolution kernel w =[ w 1 , w 2 ,⋯, w k ] , the i-th element in the convolution output y can be expressed as: ; in, is the i-th element in the convolution output, k is the length of the convolution kernel, is the jth weight parameter of the convolution kernel, i and j are positive integers, and the second harmonic signal x =[ x 1 , x 2 ,⋯, x n ] , convolution kernel w =[ w 1 , w 2 ,⋯, w k ] , n is the length of the second harmonic signal x, is the bias term; i ranges from 1 to n k+1, that is, the length of the convolution output is equal to n k+1, assuming the step size is 1.

[0040] For a stride of s, the length of the convolution output is It can be expressed as: .

[0041] After the convolution operation is completed, the output value of the convolution layer will be input to the activation layer as the input value, and pass through a nonlinear activation function. The ReLU function can be selected as the activation function, and its formula is: ; This step introduces a nonlinear activation function, which helps the network learn complex signal features.

[0042] The number of convolution combination layers is multiple, and the number of convolution layers is also multiple. Suppose the input feature map of the lth convolution layer is , the jth weight parameter of the lth convolution kernel is , then the output of this layer for: ; is the size of the convolution kernel of the lth layer; is the bias term of the lth layer.

[0043] Among them, the stacking of convolutional layers allows the interference elimination model to gradually extract features at different levels, and gradually build more complex and higher-level abstract features from simple and local features, thereby facilitating the subsequent separation of the second harmonic signals of the first gas and the second gas.

[0044] In some feasible implementations, l is less than or equal to 5. For example, if there are 5 convolutional layers in total, when l is 3, it represents the convolution output value of the third layer.

[0045] After multiple convolutional layers are processed, the output value of the convolutional layer will be used as the input value of the pooling layer.

[0046] The pooling layer is a dimensionality reduction layer used to reduce the size of the feature map and retain significant features. The pooling layer can be a maximum pooling layer, and the output formula of the maximum pooling layer is: ; The size of the pooling window is p, is the pooling result of the i-th layer.

[0047] The length of the feature map after pooling is the original length divided by the pooling window size, which can be used to retain the most important local features and reduce the amount of calculation.

[0048] The output value of the pooling layer will be used as the input value of the fully connected layer, and the fully connected layer is used to separate the second harmonic signals of the first gas and the second gas.

[0049] The output formula of the fully connected layer is: ; ; To eliminate the interference of the model output of the first gas second harmonic signal, A second harmonic signal of a second gas outputted by the interference elimination model; is a linear activation function, ; Extract the jth input feature of the second harmonic signal in the mth layer; is the jth weight parameter of the first gas in the mth layer, is the jth weight parameter of the second gas in the mth layer, is the bias term of the first gas in the mth layer, is the bias term of the second gas of the mth layer, where m is the number of convolutional combination layers, and m is less than or equal to 5.

[0050] In some feasible implementations, when training the interference cancellation model, the input signal of the interference cancellation model can be normalized. Normalization can make the features learned by the model more general and reduce the dependence on numerical values. During the training process, the interference cancellation model can better capture the structural characteristics of the data, rather than being limited to the size of the numerical value, which helps to improve the generalization ability of the model and perform better on new data. The normalization formula is: ; X is the original signal, X max is the maximum value of the original signal, X min is the minimum value of the original signal, X nom When the training data is normalized by this formula, X can be the second harmonic signal of the training mixed gas, and X max can be the maximum value of the second harmonic signal of the training mixed gas, Xmin It can be the minimum value of the second harmonic signal of the training mixed gas, X nom It can be a normalized signal of the second harmonic signal of the training mixed gas.

[0051] After the interference elimination model is trained, its output value is denormalized. The denormalization formula is: ; is the anti-normalized signal; is a scaling factor, which is used to adapt or adjust the normalized signal. Wherein, Y can be the second harmonic signal of the first gas after anti-normalization.

[0052] In this way, during the training of the interference elimination model, the pattern learned by the fully connected layer can automatically identify and separate the second harmonic signals of the first gas and the second gas, eliminate the interference of the second gas on the spectrum of the first gas, and achieve a correction effect.

[0053] See also Figure 5 As shown, in a specific implementation, the interference elimination and temperature detection method based on second harmonic provided in the embodiment of the present application can be implemented by the following steps S100 to S600.

[0054] Step S100: Acquire the second harmonic signal actually detected by the system as training data, and use the second harmonic signals of the first gas and the second gas detected experimentally as labels.

[0055] Step S200: Initialize the interference cancellation model.

[0056] Step S300: training the interference cancellation model.

[0057] Step S400: The interference cancellation model completes training. Steps S100 to S400 may correspond to step S2 in the above implementation.

[0058] Step S500: Determine whether the separated interference elimination model can effectively separate the second harmonic signals of the first gas and the second gas. If the result is yes, the training is completed, and if the result is no, step S600 is executed.

[0059] In this way, the separation effect can be guaranteed.

[0060] Step S600: Adjust interference cancellation model parameters and structure.

[0061] Specifically, the loss function of the interference cancellation model can be a mean square error function, which is used to calculate the difference between the estimated signal and the true signal in order to optimize the network neural parameters. The loss function can be expressed as follows: ; is the actual second harmonic signal of the first gas, is the actual second harmonic signal of the second gas, wherein the actual second harmonic signal can be the second harmonic signal detected in the experiment, that is, the real signal. The estimated signal is the output signal of the interference elimination model. In this way, through the loss function, it can be ensured that the separated second harmonic signals of the first gas and the second gas are as close to the real source signal as possible.

[0062] During the training of the interference elimination model, if the interference elimination model can complete the separation operation of the first gas and the second gas, the training is completed. If it does not meet the requirements, the interference elimination model is further trained by adjusting the parameters and structure of the interference elimination model.

[0063] Still taking the first gas as CO2 and the second gas as H2O as an example, see Figure 6 As shown, it can be seen that the signal after separation by the interference elimination model achieves a good separation effect, corrects the influence of the H2O second harmonic signal on the CO2 second harmonic signal, and facilitates the accuracy of the subsequent temperature inversion.

[0064] Step S3: inputting the real-time second harmonic signal of the collected mixed gas into the trained interference elimination model to obtain the real-time second harmonic signal of the first gas.

[0065] Specifically, step S3 may include step S31 and step S32.

[0066] Step S31: normalizing the real-time second harmonic signal of the collected mixed gas.

[0067] In this step, the normalization formula may be the same as the normalization formula in the above training process.

[0068] In this step, X can be the real-time second harmonic signal of the mixed gas, X max It can be used to collect the maximum value of the real-time second harmonic signal of the mixed gas, X min It can be used to collect the minimum value of the real-time second harmonic signal of the mixed gas, X nom It can be a normalized signal of the real-time second harmonic signal of the collected mixed gas.

[0069] Step S32: inputting the normalized real-time second harmonic signal of the collected mixed gas into the trained interference elimination model to obtain the real-time second harmonic signal of the first gas.

[0070] In this way, the second harmonic signal of the first gas and the second harmonic signal of the second gas can be separated by the interference elimination model.

[0071] Of course, the interference elimination model can also obtain the real-time second harmonic signal of the second gas, so as to perform comparative analysis on the real-time second harmonic signals of the two gases.

[0072] Step S4: Inverting the temperature of the first gas based on the real-time second harmonic signal of the first gas.

[0073] Specifically, in step S4, the second harmonic signal of the first gas may be first denormalized, and then the temperature of the first gas may be inverted using the denormalized second harmonic signal of the first gas. In this step, Y in the denormalization formula is the real-time second harmonic signal of the first gas after denormalization.

[0074] See also Figure 7 The figure shows the fitted linear relationship between the second harmonic signal of CO2 and temperature after the photoacoustic spectroscopy is used for temperature detection and the interference elimination model is processed. The CO2 gas absorption peak is used as the characteristic peak for temperature detection, and the second harmonic signal is obtained through wavelength modulation technology to perform temperature inversion, which effectively improves the accuracy of temperature detection. It can be seen that the amplitude of the second harmonic signal of the first gas after the interference elimination model algorithm is processed shows a good linear relationship with the temperature.

[0075] The interference elimination and temperature detection method based on second harmonic provided in the embodiment of the present application is universal and can be extended to other related technical fields. The main contents include the establishment of training data sets and feature selection, the construction of model structure, the training and use of interference elimination models to calibrate the photoacoustic spectroscopy temperature detection system, effectively ensure the accuracy of temperature detection, and improve detection efficiency.

[0076] Corresponding to the aforementioned embodiment of the interference elimination and temperature detection method based on second harmonic, the present application also provides an embodiment of the interference elimination and temperature detection system based on second harmonic.

[0077] See also Figure 8 The second harmonic-based interference elimination and temperature detection system 100 provided in the embodiment of the present application includes: an establishment module 701 , a training module 702 , a separation module 703 and an inversion module 704 .

[0078] The establishing module 701 is configured to establish an interference cancellation model.

[0079] That is, the establishing module 701 is used to execute step S1 in the above-mentioned second harmonic-based interference elimination and temperature detection method embodiment.

[0080] The training module 702 is configured to: use a training data set to train the interference elimination model; the training data set includes a second harmonic signal of a training mixed gas and a second harmonic signal of a first gas and a second harmonic signal of a second gas corresponding to the second harmonic signal of the training mixed gas; the training mixed gas includes a first gas and a second gas, and the second gas is an interfering gas of the first gas.

[0081] That is, the training module 702 is used to execute step S2 in the above-mentioned second harmonic-based interference elimination and temperature detection method embodiment.

[0082] The separation module 703 is configured to: input the real-time second harmonic signal of the collected mixed gas into the trained interference elimination model to obtain the real-time second harmonic signal of the first gas.

[0083] That is, the separation module 703 is used to execute step S3 in the above-mentioned second harmonic-based interference elimination and temperature detection method embodiment.

[0084] The inversion module 704 is configured to invert the temperature of the first gas based on the real-time second harmonic signal of the first gas.

[0085] That is, the inversion module 704 is used to execute step S4 in the above-mentioned second harmonic-based interference elimination and temperature detection method embodiment.

[0086] The interference cancellation model includes a plurality of sequentially connected convolutional combination layers, a pooling layer, and a fully connected layer; wherein the convolutional combination layer includes a connected convolutional layer and an activation layer. The calculation formulas of each structural layer of the interference cancellation model can be found in the above method embodiment, which will not be repeated here.

[0087] For some possible implementations, see Fig. 9 The second harmonic-based interference elimination and temperature detection system also includes a signal generator 10 , a laser controller 20 , a distributed feedback laser 30 (Distributed Feedback Laser, DFB), a microphone 40 , a lock-in amplifier 50 , a data acquisition device 60 and a processor 70 .

[0088] The signal generator 10 includes a first output terminal 11 and a second output terminal 12 .

[0089] The first output terminal 11 is connected to the input terminal of the laser controller 20. The signal generator 10 is configured to generate a low-frequency sawtooth wave and a high-frequency sine wave, and the generated sawtooth wave and sine wave signals are processed by an addition circuit and act on the laser controller 20 as modulation signals.

[0090] The output end of the laser controller 20 is connected to the input end of the distributed feedback laser 30 . The laser controller 20 is configured to send a driving signal to the distributed feedback laser 30 . The driving signal includes parameters of the distributed feedback laser 30 such as wavelength and frequency.

[0091] The distributed feedback laser 30 is configured to generate a detection light signal, the wavelength scanning range of the detection light signal covers the absorption wavelength of the first gas and the second gas. The wavelength of the distributed feedback laser 30 is controlled by the laser controller 20 through temperature and current.

[0092] The microphone 40 is arranged in the temperature detection area 200, and the detection light signal is absorbed by the gas in the temperature detection area 200 to generate a photoacoustic signal. The microphone 40 is configured to convert the photoacoustic signal into an electrical signal and send it to the lock-in amplifier 50. The temperature detection area 200 can be a closed detection area or an open detection area.

[0093] The input end of the phase-locked amplifier 50 is connected to the output end of the microphone 40 and the second output end 12 of the signal generator 10 respectively. The phase-locked amplifier 50 is configured to demodulate the electrical signal according to the reference signal sent by the signal generator 10 to obtain a second harmonic signal.

[0094] The input end of the data collector 60 is connected to the output end of the lock-in amplifier 50, and is used to collect the second harmonic signal and send it to the processor 70. The data collector 60 may be a data acquisition card.

[0095] The input end of the processor 70 is connected to the output end of the data collector 60, and the processor 70 includes a building module 701, a training module 702, a separation module 703 and an inversion module 704. The processor 70 can be a computer or other devices with processing functions.

[0096] It should be noted that those skilled in the art will easily think of other embodiments of the present application after considering the specification and practicing the application disclosed herein. The present application is intended to cover any variation, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary technical means in the art that are not disclosed in the present application.

[0097] It should be understood that the present application is not limited to the precise construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof, the true scope being indicated by the present application.

Claims

1. A method for interference elimination and temperature detection based on second harmonic, characterized in that: include: Establish interference cancellation model; The interference elimination model is trained using a training data set; the training data set includes a second harmonic signal of a training mixed gas and a second harmonic signal of a first gas and a second harmonic signal of a second gas corresponding to the second harmonic signal of the training mixed gas; the training mixed gas includes the first gas and the second gas, and the second gas is an interfering gas of the first gas; Inputting the real-time second harmonic signal of the collected mixed gas into the trained interference elimination model to obtain the real-time second harmonic signal of the first gas; The temperature of the first gas is inverted based on the real-time second harmonic signal of the first gas.

2. The interference elimination and temperature detection method based on second harmonic according to claim 1, characterized in that: The interference elimination model includes a plurality of sequentially connected convolutional combination layers, wherein the convolutional combination layer includes connected convolutional layers and activation layers; The convolution formula of the convolution layer is: ； in, is the i-th element in the convolution output, k is the size of the convolution kernel, is the jth weight parameter of the convolution kernel, i and j are positive integers, and the second harmonic signal , convolution kernel , n is the magnitude of the second harmonic signal x, is the bias term; The activation function of the activation layer is the ReLU function.

3. The interference elimination and temperature detection method based on second harmonic according to claim 2, characterized in that: The interference elimination model also includes a pooling layer connected to the convolution combination layer, the pooling layer is a maximum pooling layer, and the pooling formula of the pooling layer is: ； in, is the pooling result of the i-th layer, and p is the size of the pooling window.

4. The interference elimination and temperature detection method based on second harmonic according to claim 3, characterized in that: The interference elimination model also includes a fully connected layer connected to the pooling layer, and the output formula of the fully connected layer is: ; ; is the second harmonic signal of the first gas output by the interference elimination model, A second harmonic signal of the second gas output by the interference elimination model; is a linear activation function, ; Extract the jth input feature of the second harmonic signal in the mth layer; is the jth weight parameter of the first gas in the mth layer, is the jth weight parameter of the second gas in the mth layer, is the bias term of the first gas in the mth layer, is the bias term of the second gas in the mth layer; wherein m is the number of layers of the convolution combination layer, and m is less than or equal to 5.

5. The interference elimination and temperature detection method based on second harmonic according to claim 4, characterized in that: The interference elimination and temperature detection method based on second harmonic also includes: Normalizing the real-time second harmonic signal of the collected mixed gas; The inverting the temperature of the first gas based on the real-time second harmonic signal of the first gas comprises: Performing a denormalization process on the real-time second harmonic signal of the first gas; The temperature of the first gas is inverted based on the real-time second harmonic signal of the first gas after denormalization.

6. The interference elimination and temperature detection method based on second harmonic according to claim 5, characterized in that: The formula for the normalization process is: ； X is the original signal, X max is the maximum value of the original signal, X min is the minimum value of the original signal, X nom is the normalized signal; The formula for the denormalization process is: ； is the anti-normalized signal; is the scaling factor.

7. The interference elimination and temperature detection method based on second harmonic according to claim 6, characterized in that: The loss function of the interference cancellation model is: ； is the actual second harmonic signal of the first gas, is the actual second harmonic signal of the second gas.

8. The interference elimination and temperature detection method based on second harmonic according to claim 7, characterized in that: The first gas is CO2, and the second gas is H2O.

9. An interference elimination and temperature detection system based on second harmonic, characterized in that: The interference elimination and temperature detection method based on second harmonic according to any one of claims 1 to 8 is adopted, and the interference elimination and temperature detection system based on second harmonic comprises: A building module configured to build an interference cancellation model; A training module is configured to: train the interference elimination model using a training data set; the training data set includes a second harmonic signal of a training mixed gas and a second harmonic signal of a first gas and a second harmonic signal of a second gas corresponding to the second harmonic signal of the training mixed gas; the training mixed gas includes the first gas and the second gas, and the second gas is an interfering gas of the first gas; A separation module is configured to: input the real-time second harmonic signal of the collected mixed gas into the trained interference elimination model to obtain the real-time second harmonic signal of the first gas; The inversion module is configured to invert the temperature of the first gas based on the real-time second harmonic signal of the first gas.

10. The interference elimination and temperature detection system based on second harmonic according to claim 9, characterized in that: The interference elimination model includes a convolutional combination layer, a pooling layer and a fully connected layer connected in sequence, the number of the convolutional combination layers is multiple, and the convolutional combination layer includes connected convolutional layers and activation layers.

Images