Method for identifying concentration of each substance in mixed substance of ethanol and methanol
Through near-infrared spectral testing and CNN model processing based on micro-nano structure, the problem of difficulty in accurately distinguishing the concentration of ethanol and methanol is solved, and high-precision concentration detection is achieved.
Patent Information
- Application Number
- CN202510479727.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-17
- Publication Date
- 2025-07-11
AI Technical Summary
The prior art is difficult to accurately distinguish and quantify the concentration of ethanol and methanol, especially at low concentrations, the resolution and sensitivity of traditional spectral analysis instruments are insufficient, and the chemical analysis methods lack specificity.
Near-infrared spectral test based on micro-nano structure is used to combine the CNN model and the least squares method. By training the CNN model, the spectral characteristics are enhanced by the micro-nano structure, the detection signal is amplified, and the micro-nano structure is used to enhance the local electromagnetic field of light and matter interaction, the detection signal is amplified, and the detection accuracy is improved by training the CNN model.
The accuracy and efficiency of detection of the concentrations of each substance in the mixed solution of ethanol and methanol is significantly improved, especially in the case of low concentrations, which can better distinguish and quantify each alcohol component.
Smart Images
Figure CN120293909A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of solution concentration identification, in particular to a method for identifying the concentration of each substance in a mixture of ethanol and methanol. Background Art
[0002] Spectroscopic technology has gradually emerged in the field of alcohol component detection due to its advantages of being fast and non-destructive. It obtains the spectral information of a substance by analyzing its absorption, emission and scattering characteristics of light of different wavelengths, and then infers its internal composition and concentration. However, when faced with the spectrum formed by a mixture of polyols, the spectra of each alcohol substance overlap highly, making it difficult for traditional analysis methods to accurately distinguish and quantify each alcohol component. In particular, for low-concentration alcohol substances, their weak spectral signals are easily masked by the spectra of other high-concentration substances, which greatly reduces the detection accuracy.
[0003] Traditional chemical analysis methods, such as acid-base titration and redox titration, are mainly based on the chemical properties of substances. However, due to the similar chemical properties of methanol and ethanol, these chemical methods are difficult to provide sufficient specificity to accurately distinguish between the two. They can only give an approximate alcohol concentration, but cannot determine the specific concentrations of methanol and ethanol. In addition, spectral analysis instruments, such as ultraviolet-visible spectrophotometers and infrared spectrometers, have limited resolution and sensitivity. For substances with similar spectral characteristics such as methanol and ethanol, it is difficult for the instrument to capture the subtle spectral differences between them, resulting in low resolution sensitivity and low accuracy in methanol and ethanol detection. Summary of the invention
[0004] The object of the present invention is to provide a method for identifying the concentration of each substance in a mixture of ethanol and methanol.
[0005] The technical solution of the present invention is as follows:
[0006] A method for identifying the concentration of each substance in a mixture of ethanol and methanol, comprising the following operations:
[0007] The mixed solution to be tested is subjected to near infrared spectroscopy test based on micro-nano structure to obtain the near infrared spectrum of the solution to be tested, and the ethanol concentration and methanol concentration are obtained through training CNN model processing and least squares method processing;
[0008] The operation of training the CNN model is as follows: Near-infrared spectroscopy tests based on micro-nano structures are respectively performed on methanol solutions with different concentrations, ethanol solutions with different concentrations, and mixed solutions obtained by mixing methanol and ethanol in different ratios, to obtain a number of methanol near-infrared spectra, a number of ethanol near-infrared spectra, and a number of mixed-solution near-infrared spectra; Resonance peak position spectra in a number of mixed-solution near-infrared spectra are obtained to get a number of preferred spectra of the mixed solutions; Using a number of preferred spectra of the mixed solutions as inputs, and the methanol near-infrared spectra and ethanol near-infrared spectra corresponding to their respective concentrations as mapping labels, the CNN model is trained until the prediction error is less than the error threshold, and then the trained CNN model is obtained.
[0009] The operation of near-infrared spectroscopy test based on micro-nano structure is as follows: The near-infrared light source is refracted by the lens and passes through the micro-nano structure placed in the mixed solution to be measured, and the detector converts the received optical signal into an electrical signal to obtain the near-infrared spectrum; The micro-nano structure is composed of a glass layer and a silver layer plated on it, and there are a number of grooves distributed on the surface of the silver layer.
[0010] The thickness of the glass layer is 600 - 1000 nm, the thickness of the silver layer is 80 - 120 nm, and the thickness of the groove is 80 - 120 nm.
[0011] The manufacturing method of the micro-nano structure is as follows: The polystyrene spheres are emulsified, and after standing still to make the polystyrene spheres evenly dispersed and stable, an emulsified polystyrene sphere solution is obtained; The emulsified polystyrene sphere solution is centrifuged, the supernatant is poured out, deionized water is added to the remaining polystyrene sphere solution, and after ultrasonic treatment, a polystyrene sphere suspension is obtained; The polystyrene sphere suspension is pushed between two dry glasses until the liquid fills the area between the two glasses, and after air drying, one glass is removed to obtain a close-packed structure of polystyrene spheres; The side of the close-packed structure of polystyrene spheres containing polystyrene spheres is treated by reactive ion etching, and then a silver layer is evaporated on the surface to obtain the initial structure; After washing off the polystyrene spheres on the surface of the initial structure and performing ultrasonic cleaning and drying, the micro-nano structure is obtained.
[0012] The particle size of the polystyrene spheres is 250 - 350 nm.
[0013] When the polystyrene sphere suspension is pushed between the two glasses, the polystyrene sphere suspension is pushed obliquely and uniformly from one corner between the two glasses.
[0014] The operation of drying the glass is specifically as follows: The glass is cleaned with anhydrous ethanol, then ultrasonic cleaned and rinsed with ionized water, and the surface of the glass is dried to obtain the dry glass.
[0015] The wavelength band of the near-infrared light source is 900 - 1700 nm.
[0016] The center of the lens and the center of the micro-nano structure are at the same height.
[0017] The beneficial effects of the present invention are:
[0018] The present invention provides a method for identifying the concentration of each substance in a mixture of ethanol and methanol. First, methanol solutions of different concentrations, ethanol solutions of different concentrations, and mixed solutions obtained by mixing methanol and ethanol in different proportions are respectively subjected to near-infrared spectroscopy tests based on micro-nano structures to obtain a plurality of methanol near-infrared spectra, a plurality of ethanol near-infrared spectra, and a plurality of mixed solution near-infrared spectra as data sets; adding micro-nano structures to the near-infrared spectroscopy test can greatly improve the resolution of alcohol substance detection, and the micro-nano structures enhance the local electromagnetic field of the interaction between light and matter, amplify the detection signal, and amplify the specific peak intensity of different substances, so as to better Low concentration and mixed substances can be distinguished; then, several resonance peak position spectra with obvious characteristic differences in the near-infrared spectra of several mixed solutions are obtained to obtain several preferred spectra of the mixed solutions; several preferred spectra of the mixed solutions are used as input, and the near-infrared spectra of methanol and ethanol of their corresponding concentrations are used as output to train the CNN model until the prediction error is less than the error threshold, thereby obtaining the trained CNN model; finally, the mixed solution to be tested is subjected to near-infrared spectroscopy test based on micro-nano structure, training CNN model processing and least squares method processing to obtain the ethanol concentration and methanol concentration, thereby improving the accuracy and efficiency of the detection results of ethanol concentration and methanol concentration in the mixed solution. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] By reading the detailed description of the preferred embodiment below, the scheme and advantages of the present application will become clear to those skilled in the art. The accompanying drawings are only for the purpose of illustrating the preferred embodiment and are not to be considered as limiting the present invention.
[0020] In the attached picture:
[0021] Figure 1 This is a spectrum diagram obtained without adding micro-nano structures in the near infrared spectrum test in the embodiment;
[0022] Figure 2 In the embodiment, the spectrum obtained by adding the micro-nano structure prepared in this embodiment to the near infrared spectrum test;
[0023] Figure 3 An electron microscope image of a micro-nano structure in an embodiment;
[0024] Figure 4 Spectrum diagram of three resonance peak positions in the embodiment. DETAILED DESCRIPTION
[0025] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings.
[0026] This embodiment provides a method for identifying the concentrations of substances in a mixture of ethanol and methanol, including the following operations:
[0027] The test mixture solution is subjected to near-infrared spectroscopy based on micro-nano structures to obtain the near-infrared spectrum of the test solution, which is then processed by a trained CNN model and the least squares method to obtain the ethanol concentration and methanol concentration.
[0028] The operation of training the CNN model is as follows: Near-infrared spectroscopy based on micro-nano structures is performed on methanol solutions with different concentrations, ethanol solutions with different concentrations, and mixed solutions obtained by mixing methanol and ethanol in different proportions, respectively, to obtain several methanol near-infrared spectra, several ethanol near-infrared spectra, and several mixed solution near-infrared spectra; The resonance peak position spectra in several mixed solution near-infrared spectra are obtained to obtain several preferred spectra of the mixed solutions; Using several preferred spectra of the mixed solutions as inputs, and the methanol near-infrared spectra and ethanol near-infrared spectra corresponding to their respective concentrations as mapping expressions, the CNN model is trained until the prediction error is less than the error threshold, and then the trained CNN model is obtained.
[0029] The operation of near-infrared spectroscopy based on micro-nano structures is as follows: After the near-infrared light source is refracted by a lens and passes through the vertically placed micro-nano structure immersed in the test mixture solution, the detector converts the received optical signal into an electrical signal to obtain the near-infrared spectrum; The micro-nano structure is composed of a glass layer and a silver layer plated on it, and several grooves are distributed on the surface of the silver layer; The grooves are regularly distributed on the surface of the silver layer, and the thickness of its deepest part is approximately equal to the thickness of the silver layer.
[0030] The specific operation process of a specific embodiment is as follows.
[0031] S1. Near-infrared spectroscopy based on micro-nano structures is performed on methanol solutions with different concentrations, ethanol solutions with different concentrations, and mixed solutions obtained by mixing methanol and ethanol in different proportions, respectively, to obtain several methanol near-infrared spectra, several ethanol near-infrared spectra, and several mixed solution near-infrared spectra.
[0032] Near-infrared spectroscopy based on micro-nano structures is performed on methanol solutions with different concentrations, ethanol solutions with different concentrations, and mixed solutions obtained by mixing methanol and ethanol in different proportions, respectively, to obtain several methanol near-infrared spectra, several ethanol near-infrared spectra, and several mixed solution near-infrared spectra, which are used as a data set for training the CNN model.
[0033] The above operation of near-infrared spectroscopy based on micro-nano structures is as follows: After the near-infrared light source is refracted by a lens and passes through the vertically placed micro-nano structure immersed in the solution, the detector converts the received optical signal into an electrical signal to obtain the near-infrared spectrum.
[0034] Specifically, the vertically placed micro-nano structure is immersed in a solution (methanol solution or ethanol solution or a mixed solution of methanol and ethanol). A lens (convex lens) is vertically placed in front of the micro-nano structure, and the center of the lens is at the same height as the center of the micro-nano structure, that is, the height of the lens is located at the middle height of the micro-nano structure, which is convenient for concentrating the near-infrared light source on the micro-nano structure (the near-infrared light source hits the silver layer of the micro-nano structure and then passes through the glass layer). A detector is placed behind the micro-nano structure to convert the optical signal into an electrical signal. After the near-infrared light source is refracted by the lens and passes through the vertically placed micro-nano structure immersed in the solution, different substances have different degrees of absorption and scattering characteristics for light of different wavelengths. This attenuation situation contains the key information of the components of the substance to be measured. Therefore, the micro-nano structure enhances the spectral characteristic response of the solution. When the attenuated light finally hits the detector, the detector (preferably a linear array CCD detector) will convert the optical signal into an electrical signal and further generate the near-infrared spectrum of the solution.
[0035] At the same time, the micro-nano structure, the lens and the detector are placed in a sealed light-tight box to improve the accuracy of the spectral data. The wavelength band of the near-infrared light source is 900 - 1700 nm. In addition, by preparing a gel, the micro-nano structure is clamped in the detection liquid tank, so that the position of the micro-nano structure does not change during each test, preventing errors caused by the change of the structure position.
[0036] The above micro-nano structure is prepared based on polystyrene spheres. The micro-nano structure is composed of a glass layer and a silver layer plated on it. The thickness of the glass layer is 600 - 1000 nm (in this embodiment, a thickness of 800 nm is selected), the thickness of the silver layer is 80 - 120 nm (in this embodiment, a thickness of 100 nm is selected), the total thickness of the micro-nano structure is 680 - 1120 nm (in this embodiment, a thickness of 900 nm is selected). There are several grooves distributed on the surface of the silver layer, and the grooves are arranged in a hexagonal periodic pattern in the neighborhood range. The thickness of the grooves is 80 - 120 nm (in this embodiment, a thickness of 100 nm is selected), which can enhance the interaction between the substance and light at the resonance peak position of the alcohol to be measured. The characteristic resonance peak of the alcohol originates from the specific vibration or energy level transition of the functional groups in the molecule. By forming an SPP resonance mode on its surface through the micro-nano structure, the detection signal is amplified, which is convenient for better distinguishing the spectral characteristics of different alcohol substances. See Figure 1 the spectrum obtained without the micro-nano structure in Figure 2 and the spectrum obtained with the micro-nano structure in
[0037] It can be clearly seen that after adding the micro-nano structure prepared in this embodiment, the spectral differences of different concentration solutions are significantly enhanced.
[0038] Step 1: Emulsify polystyrene spheres. After standing still to evenly disperse and stabilize the polystyrene spheres, an emulsified polystyrene sphere solution is obtained. Centrifuge the emulsified polystyrene sphere solution, pour off the supernatant, add deionized water to the remaining polystyrene sphere solution, and perform ultrasonic treatment to obtain a polystyrene sphere suspension.
[0039] Specifically, take out polystyrene spheres with a particle size of 250 - 350 nm (the particle size of 300 nm is selected in this example) for emulsification treatment to promote the full dispersion of polystyrene spheres in the solution. After emulsification, let the solution stand overnight to evenly disperse and stabilize the polystyrene spheres, and then an emulsified polystyrene sphere solution is obtained. Use a brand - new syringe to extract the emulsified polystyrene sphere solution and centrifuge it. After centrifugation, pour off the supernatant, select the polystyrene sphere solution at the bottom after centrifugation, then add deionized water to this solution and perform ultrasonic treatment to fully mix and homogenize the polystyrene spheres and deionized water to form a polystyrene sphere suspension.
[0040] Step 2: Push the polystyrene sphere suspension between two dry glass sheets until the liquid fills the area between the two glass sheets, and then air - dry. After removing one glass sheet, a close - packed structure of polystyrene spheres is obtained. After the side of the close - packed structure of polystyrene spheres containing polystyrene spheres is treated by reactive ion etching, a silver layer is evaporated on the surface to obtain an initial structure.
[0041] Specifically, place two dry glass sheets with a thickness of 600 - 1000 nm (the thickness of 800 nm is selected in this example) staggeredly. Place a wire loop around the overlapping area to form a gap between the two glass sheets, providing a generation site for the micro - nano structure. Use a syringe to extract the polystyrene sphere suspension and slowly and evenly push the polystyrene sphere solution along the gap between the glass sheets until the liquid completely fills the area between the two glass sheets. Place the assembled sample in a well - ventilated environment and air - dry for a period of time to allow the polystyrene spheres to self - assemble into a hexagonal close - packed structure on the glass sheet surface. After removing one glass sheet, a close - packed structure of polystyrene spheres is obtained. After the structure is stable, perform reactive ion etching treatment on the side of the close - packed structure of polystyrene spheres containing polystyrene spheres, and control the etching time to 200 - 280 s (the time of 240 s is selected in this example) to increase the distance between polystyrene spheres. After etching, evaporate a silver layer with a thickness of 80 - 120 nm (the thickness of 100 nm is selected in this example) on the surface of the side containing polystyrene spheres to obtain an initial structure.
[0042] Among them, to improve the generation quality of the micro-nano structure, the operation of drying the glass is as follows: after cleaning the glass with absolute ethanol, ultrasonic cleaning and rinsing with deionized water are carried out, and then the surface of the glass is dried to obtain dry glass. Specifically, select a glass sheet of appropriate size, initially clean it with absolute ethanol to ensure that there is no visible impurity residue on the surface of the glass sheet. Then place the glass sheet in an ultrasonic cleaning device and clean it with alcohol to thoroughly remove the dirt and tiny impurity particles attached to the surface of the glass sheet by ultrasonic waves. Rinse its surface with deionized water multiple times to completely remove the residual alcohol. After the rinsing is completed, use a rubber ear bulb to dry the surface of the glass sheet to ensure that the surface of the glass sheet is in a dry state.
[0043] Meanwhile, to ensure the uniformity of the micro-nano structure, when pushing the polystyrene sphere suspension between two glass sheets, the polystyrene sphere suspension is pushed in obliquely and uniformly from one corner between the two glass sheets. Specifically, at the lower right corner or lower left corner of the overlapping area formed between the two glass sheets, push the polystyrene sphere suspension obliquely in the direction of the upper left corner or upper right corner at a uniform speed. After the polystyrene sphere suspension touches the upper part of the overlapping area formed between the two glass sheets, it naturally flows downwards, so that during the pushing process, the polystyrene sphere suspension can fill and cover the overlapping area formed between the two glass sheets, preventing the appearance of voids and improving the generation uniformity of the subsequent micro-nano structure.
[0044] Step 3: After washing off the polystyrene spheres on the surface of the initial structure, perform ultrasonic cleaning, and after drying, obtain the micro-nano structure. Specifically, place the initial structure in a tetrahydrofuran solution, and ultrasonically treat to wash off the polystyrene spheres on the surface. Perform ultrasonic cleaning with an acetone solution, then rinse with alcohol, and dry the initial structure with a rubber ear bulb. After the above treatment, hexagonal periodic grooves can be formed on the surface of the initial structure to obtain the micro-nano structure. The electron microscope image of the micro-nano structure prepared from polystyrene spheres with a particle size of 300 nm is shown in Figure 3 From the electron microscope image, it can be seen that the grooves of the micro-nano structure are evenly distributed and arranged regularly and orderly, and the holes form a hexagonal periodic arrangement within the neighborhood range.
[0045] S2: Obtain the spectral positions of the resonance peaks in the near-infrared spectra of several mixed solutions to obtain several preferred spectra of the mixed solutions; use several preferred spectra of the mixed solutions as inputs, and the corresponding methanol near-infrared spectrum and ethanol near-infrared spectrum as mapping labels to train the CNN model until the prediction error is less than the error threshold, and then obtain the trained CNN model.
[0046] First, summarize the near-infrared spectra of several mixed solutions to form a spectrogram with the wavelength on the abscissa and the transmittance on the ordinate. In the spectrogram, obtain the spectral positions of the resonance peaks in the near-infrared spectra of several mixed solutions to obtain several preferred spectra of the mixed solutions, as shown in Figure 4Spectra of the middle three resonance peak positions.
[0047] The spectra at the resonance peak positions, due to the attenuation of the near-infrared light source by the micro-nano structure, make the spectral differences between solutions with different concentrations more obvious compared to the spectral differences without the influence of the micro-nano structure (see Figure 1 and Figure 2 ), which is convenient for subsequent analysis and improves the detection accuracy.
[0048] The method for obtaining the resonance peak positions is as follows: traverse the spectral curve in the spectrogram with a window of fixed size and fixed step length. During the traversal, the abscissa position corresponding to the maximum value point within the window is taken as the peak point, and the abscissa positions within the neighborhood range of the peak point are taken as the resonance peak positions; the neighborhood range can be divided or varied according to the actual situation.
[0049] Then, several optimized spectra of the mixed solutions are used as inputs, and the near-infrared spectra of methanol and ethanol corresponding to their respective concentrations are used as mapping labels (outputs) to train the CNN model. After the prediction error is less than the error threshold, the trained CNN model is obtained.
[0050] In the CNN model, the optimized spectral data of each mixed solution first enters the convolutional layer. The convolutional kernel slides continuously on the data according to the set step length for convolutional operations to extract local features in the data. Then, the data enters the pooling layer for dimensionality reduction processing. By reducing the data dimension, the computational amount and the complexity of the model are effectively reduced. The data after convolutional and pooling processing is finally concatenated in the fully connected layer, and various features extracted previously are comprehensively analyzed. At the same time, a regression task is performed in the fully connected layer. By continuously optimizing the model parameters, the model gradually learns the internal mapping relationship between the spectra of the mixed substances and the spectra of each substance, and finally outputs the spectral data of each substance in the mixed substance.
[0051] During the training process, first, segmental training and learning are carried out based on the characteristic positions in the spectra to obtain the spectral mapping relationship between the mixed substances and the single substances therein. Through model training, the spectral information related to different substances can be initially separated. Then, the single-substance spectra are separated specifically, and finally, the spectral data of each substance alone is output, and the spectral characteristic information of each substance is retained. Since single-alcohol substances have fingerprint characteristics and change monotonically with the change of concentration, the lower the concentration of the alcohol, the lower the light intensity of the spectrum. Through the step of separating the single-substance spectra from the mixed-substance spectra, due to the fingerprint characteristics of the peak-valley positions of the spectra of this substance at different concentrations and relying on the uniqueness of the spectral characteristics, the generated spectra are close to the real spectra, and thus the concentration prediction is more accurate. When the spectra overlap severely and the samples are complex, separating the spectra first and then combining the calibration model to calculate the concentration is more reliable than directly predicting the concentration.
[0052] S3. The mixed solution to be measured is subjected to near-infrared spectroscopy based on micro-nano structures to obtain the near-infrared spectrum of the solution to be measured. After being processed by the trained CNN model and the least squares method, the ethanol concentration and methanol concentration are obtained.
[0053] The mixed solution to be measured is subjected to near-infrared spectroscopy based on micro-nano structures to obtain the near-infrared spectrum of the solution to be measured; the near-infrared spectrum of the solution to be measured is processed by the trained CNN model to obtain the near-infrared spectrum results of methanol and ethanol; the near-infrared spectrum results of methanol and ethanol are processed by the least squares method to obtain the ethanol concentration and methanol concentration results.
[0054] To verify the effectiveness of the method in this embodiment, the method in this embodiment is compared with existing advanced methods (including traditional CNN methods, BP neural networks, MLP networks, random forests; among them, the input in the traditional CNN method is the entire near-infrared spectrum of the mixed solution, and the micro-nano structure is composed of a glass layer with a thickness of 800 nm and a silver layer with a thickness of 100 nm), and the discrimination results of the methanol and ethanol concentrations in the mixed solution are compared. And from four important indicators of RMSE (root mean square error), MAE (mean absolute error), and R 2 (coefficient of determination) to evaluate the discrimination effects of different methods, as shown in Table 1. It can be clearly seen from the data in Table 1 that for substance 1 ethanol and substance 2 methanol, the RMSE and MAE of the method in this embodiment are the smallest overall among all models, indicating that the method in this embodiment has the best prediction performance.
[0055] Table 1 Summary of the discrimination effects of methanol and ethanol concentrations in different mixed solutions by different methods
[0056]
[0057] The present embodiment provides a method for identifying the concentration of each substance in a mixture of ethanol and methanol. First, methanol solutions of different concentrations, ethanol solutions of different concentrations, and mixed solutions obtained by mixing methanol and ethanol in different proportions are subjected to near-infrared spectroscopy tests based on micro-nano structures, respectively, to obtain a number of methanol near-infrared spectra, a number of ethanol near-infrared spectra, and a number of mixed solution near-infrared spectra as data sets. Adding micro-nano structures to near-infrared spectroscopy tests can greatly improve the resolution of alcohol substance detection. The micro-nano structures enhance the local electromagnetic field of the interaction between light and matter, amplify the detection signal, and amplify the specific peak intensity of different substances, which can better Low-concentration and mixed substances are distinguished; then, several resonance peak position spectra with obvious characteristic differences in the near-infrared spectra of several mixed solutions are obtained to obtain several preferred spectra of the mixed solutions; the several preferred spectra of the mixed solutions are used as input, and the methanol near-infrared spectra and ethanol near-infrared spectra of their respective corresponding concentrations are used as output to train the CNN model until the prediction error is less than the error threshold, thereby obtaining the trained CNN model; finally, the mixed solution to be tested is subjected to near-infrared spectroscopy testing based on micro-nano structures, training CNN model processing and least squares processing to obtain the ethanol concentration and methanol concentration, thereby improving the accuracy and efficiency of the detection results of the ethanol concentration and methanol concentration in the mixed solution.
Claims
1. A method for identifying the concentrations of each substance in a mixture of ethanol and methanol, characterized in that, Including the following operations: The mixed solution to be measured is subjected to near-infrared spectroscopy based on a micro-nano structure to obtain the near-infrared spectrum of the solution to be measured. After being processed by a trained CNN model and the least squares method, the ethanol concentration and methanol concentration are obtained; The operation of training the CNN model is as follows: Methanol solutions with different concentrations, ethanol solutions with different concentrations, and mixed solutions obtained by mixing methanol and ethanol in different proportions are respectively subjected to near-infrared spectroscopy based on a micro-nano structure to obtain a number of methanol near-infrared spectra, a number of ethanol near-infrared spectra, and a number of mixed solution near-infrared spectra; Resonance peak position spectra in a number of mixed solution near-infrared spectra are obtained to obtain a number of preferred spectra of the mixed solution; Using a number of preferred spectra of the mixed solution as inputs, and the methanol near-infrared spectra and ethanol near-infrared spectra corresponding to their respective concentrations as mapping labels, the CNN model is trained until the prediction error is less than the error threshold, and then the trained CNN model is obtained; The operation of near-infrared spectroscopy based on a micro-nano structure is as follows: The near-infrared light source passes through the micro-nano structure placed in the mixed solution to be measured after being refracted by a lens, and the detector converts the received optical signal into an electrical signal to obtain the near-infrared spectrum; The micro-nano structure is composed of a glass layer and a silver layer plated on it, and a number of grooves are distributed on the surface of the silver layer.
2. The method for identifying the concentration of each substance in a mixture of ethanol and methanol according to claim 1, characterized in that, The thickness of the glass layer is 600 - 1000 nm, the thickness of the silver layer is 80 - 120 nm, and the thickness of the groove is 80 - 120 nm.
3. The method for identifying the concentration of each substance in the ethanol and methanol mixture according to claim 1, characterized in that, The manufacturing method of the micro-nano structure is as follows: The polystyrene spheres are emulsified, and after standing still to make the polystyrene spheres uniformly dispersed and stable, an emulsified polystyrene sphere solution is obtained; The emulsified polystyrene sphere solution is centrifuged, the supernatant is poured off, deionized water is added to the remaining polystyrene sphere solution, and after ultrasonic treatment, a polystyrene sphere suspension is obtained; The polystyrene sphere suspension is pushed between two dry glasses until the liquid fills the area between the two glasses, and after air drying, one glass is removed to obtain a close-packed structure of polystyrene spheres; The side of the close-packed structure of polystyrene spheres containing polystyrene spheres is subjected to reactive ion etching treatment, and then a silver layer is evaporated on the surface to obtain an initial structure; After washing off the polystyrene spheres on the surface of the initial structure and then performing ultrasonic cleaning and drying, the micro-nano structure is obtained.
4. The method for identifying the concentrations of each substance in a mixture of ethanol and methanol according to claim 3, characterized in that, The particle size of the polystyrene spheres is 250 - 350 nm.
5. The method for identifying the concentration of each substance in a mixture of ethanol and methanol according to claim 3, characterized in that, When the polystyrene sphere suspension is pushed between the two glasses, the polystyrene sphere suspension is pushed obliquely and uniformly from one corner between the two glasses.
6. The method for identifying the concentrations of each substance in a mixture of ethanol and methanol according to claim 1, characterized in that, The operation of drying the glass is specifically as follows: The glass is cleaned with absolute ethanol, then subjected to ultrasonic cleaning and rinsing with ionized water, and the surface of the glass is dried to obtain dry glass.
7. The method for identifying the concentration of each substance in the ethanol and methanol mixture according to claim 1, wherein The wavelength band of the near-infrared light source is 900 - 1700 nm.
8. The method for identifying the concentrations of each substance in a mixture of ethanol and methanol according to claim 1, characterized in that, The center of the lens is at the same height as the center of the micro-nano structure.