Method for identifying biomass based on machine vision and Raman laser coupling

Abstract

The invention discloses a method for identifying biomass based on machine vision and Raman laser coupling. The method comprises the following steps: acquiring images, Raman spectrograms and physicochemical property data of various biomasses; extracting image feature parameters of the image data to obtain a first feature data set; raman characteristic parameters of the Raman spectrogram are extracted, and a second characteristic data set is obtained; based on the first feature data set and the second feature data set, respectively establishing a Hardgrove grindability prediction model, a moisture content prediction model and a hydrocarbon element ratio prediction model; and acquiring an image, a Raman spectrum and physicochemical property data of the biomass to be detected, and identifying the biomass to be detected by using the three prediction models. The method has the rapidity of machine vision recognition and the accuracy of Raman laser recognition, the grindability and the water content of the biomass to be recognized can be obtained, and the chemical components of the biomass to be recognized can also be obtained; compared with a single method for identification, the method has the advantages that biomass detection results are expanded, and interference of impurities on the surface of the to-be-detected object on Raman laser identification can be overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of biomass detection, and particularly relates to a method for identifying biomass based on the coupling of machine vision and Raman laser. Background Art

[0002] As a renewable energy source, biomass has the advantage in its power generation technology of reducing the consumption of fossil energy and improving the energy utilization structure. When a power plant uses biomass fuel for power generation, the power generation efficiency and the stable operation of the power plant are directly related to the quality of the biomass fuel. Especially when using a mixture of multiple types of biomass for combustion, it is necessary to make the combustion quality of the mixed biomass closer to the combustion characteristics of the combustion boiler. Therefore, it is necessary to detect the biomass before it is fed into the boiler for combustion to determine its type and chemical composition.

[0003] Traditional biomass detection methods involve sampling and performing chemical industrial analysis on the samples in the laboratory to determine their composition and chemical components. Although the results obtained by traditional biomass detection methods are accurate, they cannot achieve on-line real-time non-destructive detection and cannot be coupled with the intelligent control system of the power plant, which is not conducive to improving the power generation efficiency of the power plant.

[0004] Therefore, there is an urgent need for a non-destructive and on-line rapid biomass detection method, which is not only conducive to improving the detection efficiency but also conducive to reducing the loss of raw materials while ensuring the detection accuracy. Summary of the Invention

[0005] In view of the above technical problems, the present invention provides a method for identifying biomass based on the coupling of machine vision and Raman laser to achieve real-time, non-destructive on-line detection of biomass.

[0006] The technical solutions provided by the present application are as follows:

[0007] In a first aspect, a method for identifying biomass based on the coupling of machine vision and Raman laser is provided, including the following steps:

[0008] Obtain images, Raman spectra, and physical and chemical property data of various biomasses;

[0009] Extract image feature parameters of the image data to obtain a first feature dataset; and extract Raman feature parameters of the Raman spectrum to obtain a second feature dataset;

[0010] Based on the first feature dataset and the second feature dataset, establish a Hardgrove grindability prediction model, a moisture content prediction model, and a carbon-hydrogen element ratio prediction model respectively;

[0011] Obtain images, Raman spectra, and physical and chemical property data of the biomass to be measured, and use the three prediction models to identify the biomass to be measured.

[0012] In a possible implementation manner, the image is an image of biomass ground into particles.

[0013] In a possible implementation manner, the image feature parameters include average particle size and grayscale.

[0014] In a possible implementation manner, the physical and chemical property data includes Hardgrove grindability index, moisture content, and carbon-hydrogen element ratio.

[0015] In a possible implementation manner, the method for obtaining the Raman characteristic parameters includes:

[0016] Obtaining the total area S of the first-order Raman peak through Raman spectroscopy 1 and the total area S of the second-order Raman peak 2 ; and the peak areas of the characteristic peaks within the range of the second-order Raman spectrum;

[0017] Selecting the area ratio that can represent the characteristics of biomass as the Raman characteristic parameter.

[0018] In a possible implementation manner, the method for establishing the Hardgrove grindability prediction model includes:

[0019] Taking the first characteristic data set and the second characteristic data set as inputs, and the Hardgrove grindability index corresponding to the biomass as the output;

[0020] Training with a neural network to obtain the Hardgrove grindability prediction model.

[0021] In a possible implementation manner, the method for establishing the moisture content prediction model includes:

[0022] Coupling the particle size feature and the grayscale feature in the first characteristic data set to obtain a picture coupling feature;

[0023] Taking the picture coupling feature and the second characteristic data set as inputs, and the mass moisture content as the output;

[0024] Training with a neural network to obtain the moisture content prediction model.

[0025] In a possible implementation manner, the method for establishing the carbon-hydrogen element ratio prediction model includes:

[0026] Coupling the particle size feature and the grayscale feature in the first characteristic data set to obtain a picture coupling feature;

[0027] Taking the particle size feature and the grayscale feature as inputs, and the picture coupling feature as the output, and fitting to obtain the first equation;

[0028] Taking the image coupling feature and the second feature dataset as inputs and the carbon-hydrogen element ratio prediction as the output, a second equation is obtained by fitting.

[0029] Substituting the first equation into the second equation, a carbon-hydrogen element ratio prediction model is obtained.

[0030] Furthermore, the method for coupling the particle size feature and the grayscale feature in the first feature dataset to obtain the image coupling feature includes:

[0031] Feature fusion: The particle size feature and the grayscale feature are weighted and summed and normalized to obtain a fused feature set.

[0032] Depth information coupling: Calculating the depth information of each pixel point of each image in the image data, and forming a binary array with the corresponding fused feature set to obtain the image coupling feature.

[0033] In a second aspect, a device for identifying biomass based on machine vision and Raman laser coupling is provided, including:

[0034] An acquisition module for obtaining images, Raman spectra, and physical and chemical property data of various biomasses.

[0035] An extraction module for extracting image feature parameters of the image data to obtain a first feature dataset; and extracting Raman feature parameters of the Raman spectrum to obtain a second feature dataset.

[0036] A construction module for respectively establishing a Hardgrove grindability prediction model, a moisture content prediction model, and a carbon-hydrogen element ratio prediction model based on the first feature dataset and the second feature dataset.

[0037] An identification module for obtaining images, Raman spectra, and physical and chemical property data of the biomass to be measured, and identifying the biomass to be measured using the three prediction models.

[0038] Advantages of the present application:

[0039] (1) The method of the present application combines the advantages of Raman laser identification and machine vision identification, and the identification process is accurate, fast, and non-destructive.

[0040] (2) Compared with using only Raman laser identification, when the surface of the biomass is contaminated, the detection of Raman laser will fail, while the method of the present application can still accurately reveal the type of biomass through machine vision means.

[0041] (3) The method of the present application can obtain more comprehensive and accurate physical and chemical information of biomass, such as grindability index, moisture content, and carbon-hydrogen element ratio, and has a wide application range. Description of the Drawings

[0042] Figure 1Schematic diagram of the process of the present invention;

[0043] Figure 2 Schematic diagram of a part of the biomass sample in the embodiment of the present invention;

[0044] Figure 3 In the embodiment of the present invention, the particle size picture obtained by image processing of the collected sample biomass picture; among them, the upper left is peanut straw powder, the lower left is bagasse, the upper right is corncob powder, and the lower right is peanut shell;

[0045] Figure 4 Original picture, grayscale picture and grayscale value of peanut shell powder with different moisture contents;

[0046] Figure 5 Spectrum diagram obtained by laser Raman test of a part of the collected biomass sample;

[0047] Figure 6 Schematic diagram of the Raman spectrum diagram of the biomass to be measured and the baseline correction process;

[0048] Figure 7 Schematic diagram of the peak splitting of the second-order Raman spectrum diagram in the embodiment of the present invention;

[0049] Figure 8(a) is a schematic diagram of the structure of the BP neural network model in the Hardgrove grindability prediction model; Figure 8(b) is a schematic diagram of the construction of the Hardgrove grindability prediction model;

[0050] Figure 9 Fitting effect of the Hardgrove grindability index in the database;

[0051] Figure 10 Schematic diagram of the construction of the moisture content prediction model;

[0052] Figure 11 Fitting effect of the mass moisture content index in the database;

[0053] Figure 12 Schematic diagram of the construction of the carbon-hydrogen element ratio C / H prediction model;

[0054] Figure 13 Fitting effect of the carbon-hydrogen element ratio in the database;

[0055] Figure 14 Comparison of the predicted values and the true values of the Hardgrove grindability index, mass moisture content, and carbon-hydrogen element ratio of the biomass to be detected;

[0056] Figure 15 Schematic diagram of the structure of a device for identifying biomass based on the coupling of machine vision and Raman laser provided by the present invention. Detailed implementation manners

[0057] The following further describes the content of the present application in conjunction with specific embodiments. The content of the present application is not limited thereto.

[0058] Refer to Figure 1 , which is a method for identifying biomass based on the coupling of machine vision and Raman laser, including the following steps:

[0059] S100, obtaining images, Raman spectra, and physical and chemical property data of various biomasses.

[0060] In a possible implementation manner, the image is an image of biomass after being ground into particles.

[0061] It can be understood that in order to ensure the accuracy of subsequent model training, the particle size distribution of the ground biomass should be as consistent as possible to a certain extent. For example, the relevant parameters of grinding can be unified to ensure the consistency of grinding.

[0062] In a possible implementation manner, the physical and chemical property data includes Hardgrove grindability index, moisture content, and carbon-hydrogen element ratio.

[0063] S200, extracting image feature parameters of the image data to obtain a first feature dataset; and extracting Raman feature parameters of the Raman spectrum to obtain a second feature dataset.

[0064] In a possible implementation manner, the image feature parameters include average particle size and grayscale.

[0065] Furthermore, the method for obtaining the average particle size includes:

[0066] Image preprocessing: Preprocessing the collected image of biomass, including denoising, enhancing contrast, etc., to improve the image quality;

[0067] Edge detection: Using an edge detection algorithm (such as the Canny algorithm) to identify the edges of biomass particles in the image;

[0068] Particle segmentation: Separating the biomass particles from the background through the edge detection result to form independent particle images;

[0069] Particle size calculation: Calculating the size of the segmented particle images and measuring the diameter of each particle;

[0070] Average particle size calculation: Calculating the average value of all particle diameters to obtain the average particle size.

[0071] Furthermore, the method for obtaining the grayscale includes:

[0072] Image grayscale conversion: Converting the collected image into a grayscale image to reduce the interference of color information;

[0073] Gray normalization: Calculate the maximum and minimum pixel values in the grayscale image, and then map each pixel value to the range of 0 to 255 to enhance the contrast of the image;

[0074] Gray feature extraction: Use methods such as gray-level co-occurrence matrix (GLCM) to extract the gray features of the image; The gray features include the contrast, uniformity, energy, etc. of the texture;

[0075] Gray projection: Calculate the gray value projection of the image in the horizontal and vertical directions to obtain the gray distribution characteristics of the image.

[0076] In a possible implementation manner, the method for obtaining the Raman characteristic parameters includes:

[0077] Obtain the total area S of the first-order Raman peak through Raman spectroscopy 1 and the total area S of the second-order Raman peak 2 ; and the peak areas of the characteristic peaks within the range of the second-order Raman spectrum;

[0078] Select the area ratio that can represent the biomass characteristics as the Raman characteristic parameter.

[0079] S300, Based on the first feature dataset and the second feature dataset, establish a Hardgrove grindability prediction model, a moisture content prediction model, and a carbon-hydrogen element ratio prediction model respectively.

[0080] In a possible implementation manner, the method for establishing the Hardgrove grindability prediction model includes:

[0081] Use the first feature dataset and the second feature dataset as inputs, and the Hardgrove grindability index corresponding to the biomass as the output;

[0082] Train using a neural network to obtain the Hardgrove grindability prediction model.

[0083] In a possible implementation manner, the method for establishing the moisture content prediction model includes:

[0084] Couple the particle size feature and the gray feature in the first feature dataset to obtain a picture coupling feature;

[0085] Use the picture coupling feature and the second feature dataset as inputs, and the mass moisture content as the output;

[0086] Train using a neural network to obtain the moisture content prediction model.

[0087] In a possible implementation manner, the method for establishing the carbon-hydrogen element ratio prediction model includes:

[0088] Couple the particle size feature and the gray feature in the first feature dataset to obtain a picture coupling feature;

[0089] Taking the particle size feature and the grayscale feature as inputs and the image coupling feature as the output, a first equation is obtained by fitting.

[0090] Taking the image coupling feature and the second feature dataset as inputs and the prediction of the carbon-hydrogen element ratio as the output, a second equation is obtained by fitting.

[0091] Substituting the first equation into the second equation, a prediction model of the carbon-hydrogen element ratio is obtained.

[0092] Furthermore, the method for coupling the particle size feature and the grayscale feature in the first feature dataset to obtain the image coupling feature includes:

[0093] Feature fusion: Weighted summation and normalization of the particle size feature and the grayscale feature are performed to obtain a fused feature set.

[0094] Depth information coupling: Calculating the depth information of each pixel point in the image data, and forming a binary array with the corresponding fused feature set to obtain the image coupling feature.

[0095] Specifically, the steps of feature fusion are as follows:

[0096] Weighted summation: Weighted summation of the particle size feature and the grayscale feature is performed, and the formula is: weighted value = w 1 × particle size feature + w 2 × grayscale feature, where w 1 and w 2 are weight coefficients.

[0097] Normalization processing: Normalization processing of the weighted value is performed to ensure that the feature values are within an appropriate range, and a fused feature set is obtained.

[0098] It should be noted that the information depth of the pixel point refers to the additional information obtained by calculating the depth value of each pixel point in the image during the image processing process. The depth information can be used to enhance the expression ability of the feature, specifically:

[0099] Depth value: In a three-dimensional image or a depth image, the depth value of each pixel point represents the distance of the point from the observation plane.

[0100] Contrast information: By calculating the depth difference between adjacent pixel points in the image, the contrast information of the image is obtained.

[0101] Exemplarily, the particle size feature is PZ = 3.11 (average particle size), the grayscale feature is G = 105 (grayscale value), assuming the weights w 1 = 0.6, w 2 = 0.4, the weighted value F = w 1 × PZ + w 2×G = 0.6×3.11 + 0.4×105 = 43.666。

[0102] After calculating the weighted value F of all image data, perform normalization processing on it. The fused feature set obtained after normalization The value of F' is between 0 and 1, and the normalized data is not affected by units.

[0103] After that, calculate the distance L of each pixel point in the image data relative to the reference plane of the camera lens, and combine F' with the corresponding L to form a binary array (F', L), which is the picture coupling feature.

[0104] Furthermore, the neural network includes the BP neural network algorithm and the random forest algorithm; the performance of the neural network model is evaluated through cross-validation.

[0105] S400, obtain the image, Raman spectrum and physicochemical property data of the biomass to be measured, and use three prediction models to identify the biomass to be measured.

[0106] The following is illustrated with more specific embodiments as follows.

[0107] A method for identifying biomass based on the coupling of machine vision and Raman laser includes the following steps:

[0108] (1) Obtain the images, Raman spectra and physicochemical property data of various biomasses, and extract characteristic parameters

[0109] Select 10 types of biomass particles as samples to establish a database. The selected samples are all air-dried basis. Obtain the pictures of these biomasses respectively, and after image processing, obtain their average particle size and picture gray value, and measure their Hardgrove grindability data. Pictures of some biomasses are as Figure 2 shown, and the average particle size of the biomass particles obtained by image processing is as Figure 3 shown, and the gray values of some biomass pictures are as Figure 4 shown. The results are shown in Table 1.

[0110] Table 1

[0111]

[0112]

[0113] Select 10 types of biomass particles as samples to establish a database. The selected samples are all as-received basis. Obtain the pictures of these biomasses respectively, and after image processing, obtain their average particle size and picture gray value, and measure their moisture content data. The results are shown in Table 2.

[0114] Table 2

[0115] biomass mass moisture content W particle image diameter PZ gray value G peanut straw 8.87％ 3.11 105 corn cob 12.45％ 3.19 90.35 bagasse 15.43％ 10.09 77.52 peanut shell 7.21％ 1.97 142.26 corn straw 22％ 3.22 114.86 rice husk 10.09％ 2 83.21 traditional Chinese medicine residue 40％ 1.56 50.21 wood chip 15％ 3.14 86.26 coffee husk 8.62％ 2.03 135.25 cotton stalk 9.04％ 3.16 140.45

[0116] Raman tests were conducted on 10 selected types of biomass, and the Raman spectra obtained are as Figure 5 shown. Table 3 shows the conditions for Raman tests when using sample substances to make samples for the biomass-Raman laser database. Baseline drifts occurred in the Raman spectra of some biomass, as Figure 6 shown, and baseline correction is required.

[0117] Table 3

[0118] laser wavelength laser power eyepiece magnification scanning time scanning range 532.16 nm 5 mW X50 10s 500 - 3400 cm-1

[0119] After completing the above steps, the second-order Raman peaks were deconvoluted. The number of deconvoluted peaks was 3, and one example is as Figure 7 shown. The Raman spectrum parameters obtained after deconvolution include the total area S 1 of the first-order Raman peaks, the total area S 2 of the second-order Raman peaks, and the peak areas I -1 , I -1 , etc. at wave numbers such as 2940 cm 2940 , 3200 cm 3200 within the second-order Raman spectrum range, as well as their combinations I 2940 / S 1 , I 3200 / S 1 , I 3200 / S 2 , I 2940 / S 2 , I 2940 / I 3200 , (I 2940 + I 3200 ) / S 1 , (I 2940 + I 3200 ) / S 2 , S 2 / S 1 , etc. Among them, (I 2940 + I 3200 ) represents the sum of I 2940 and I 3200 .

[0120] After the deconvolution calculation, a mapping relationship was established between the Raman spectral characteristic parameters of the selected sample biomass and its chemical composition parameters, thus successfully constructing a biomass-Raman database. In this specific application example, (I2940 + I3200) / S2 was used as the Raman characteristic parameter of the biomass. The carbon-hydrogen element ratio and Raman characteristic parameters of the sample biomass are shown in Table 4.

[0121] Table 4

[0122] biomass carbon-hydrogen ratio C / H (I2940 + I3200) / S2 peanut straw 7 1.08 corn cob 8.26 3.8 bagasse 9.67 6.5 peanut shell 7.02 1.12 corn straw 7.79 2.7 rice husk 7.45 1.93 traditional Chinese medicine residue 7.81 2.61 wood chip 7.8 2.88 coffee husk 6.57 0.28 cotton stalk 7.23 1.5

[0123] (2) Establish the Hardgrove grindability prediction model, moisture content prediction model, and carbon-hydrogen element ratio prediction model

[0124] (2.1) Establish the prediction model of Hardgrove grindability with respect to {PZ, G, (I2940 + I3200) / S2}

[0125] The construction method is as follows: The BP neural network algorithm is adopted, and three groups of data, namely PZ, G, and (I2940 + I3200) / S2, are used as inputs, and the Hardgrove grindability index of biomass is used as the output. One set of hidden layer and output layer is set, and the structure of the BP neural network is shown in Figure 8(a). The model construction is shown in Figure 8(b).

[0126] During the model training process, first, the particle size data PZ, gray level data G, and Raman characteristic data (I2940 + I3200) / S2 of biomass are standardized. The data set is divided into a training set and a test set according to 8:2. 10 hidden layer neurons are set, the number of iterations is set to 1000 times, and the learning rate is set to 0.01. After training, the grindability prediction model is obtained. The fitting effect of the model on the test set is as Figure 9 shown. According to the fitting effect, it can be known that the Hardgrove grindability index calculated by fitting is basically the same as the actual Hardgrove grindability index of biomass.

[0127] (2.2) Establish the prediction model of the moisture content W of biomass particles with respect to {PZ, G, (I2940 + I3200) / S2}

[0128] The construction method is as follows: First, the particle size characteristics of the particles and the gray level value characteristics of the pictures are coupled to obtain the picture feature ImageFeature, abbreviated as IF. The two groups of data, IF and RF, are used as inputs and sent into the BP neural network model, and the mass moisture content W is used as the output. The prediction model construction is as Figure 10 shown, and the fitting effect of the model on the test set is as Figure 11 shown. According to the fitting results, it can be known that the mass moisture content of biomass calculated by fitting is basically the same as the actual mass moisture content.

[0129] (2.3) Establish the prediction model of the carbon-hydrogen element ratio C / H of biomass particles with respect to {PZ, G, (I2940 + I3200) / S2}

[0130] The construction method is as follows: First, linear fitting is adopted, with PZ and G as inputs and the image feature IF as the output. The equation of the linear fitting is IF = 4.9 - 0.27×PZ + 0.01×G; Then, the two sets of data of IF and RF of the biomass are used as inputs and C / H as the output for linear fitting, and the fitting equation obtained is C / H = 6.80 + 0.47×((I2940 + I3200) / S2) - 0.05×IF. The prediction model is constructed as Figure 12 shown, and the fitting effect on C / H is as Figure 13 shown. According to the fitting results, it can be known that the calculated carbon-hydrogen ratio of the biomass by fitting is basically consistent with the actual carbon-hydrogen ratio.

[0131] (3) Obtain the image, Raman spectrum and physicochemical property data of the biomass to be measured, and use three prediction models to identify the biomass to be measured

[0132] (3.1) Select the biomass to be detected and process it in the same method as in step (2) to obtain its average particle size PZ, picture gray level G and Raman characteristic parameters as shown in Table 5. It should be noted that peak areas such as I 2940 , I 3200 , S 1 , S 2 and other parameters are signal intensities and have no definite physical units. Just keep the calculation on the same scale. At the same time, conduct industrial elemental composition analysis on the biomass, and the results are shown in Table 6.

[0133] Table 5

[0134]

[0135]

[0136] Table 6

[0137]

[0138] (3.2) Feed the obtained characteristic parameters into the coupling model constructed in step (2) to predict the three indicators of the grindability index, mass moisture content and carbon-hydrogen element ratio of the biomass respectively. The obtained prediction results are as Figure 14 shown. The prediction results are in good agreement with the characteristic parameters of peanut shells in the database, so it can be determined that the biomass to be measured is peanut shells.

[0139] Next, a device for identifying biomass based on the coupling of machine vision and Raman laser provided by the present invention will be described. The device for identifying biomass based on the coupling of machine vision and Raman laser described below can be correspondingly referred to the method for identifying biomass based on the coupling of machine vision and Raman laser described above.

[0140] Figure 15It is a schematic structural diagram of a biomass identification device based on the coupling of machine vision and Raman laser provided by an embodiment of the present invention. As Figure 15 shown, it includes: an acquisition module 1511, an extraction module 152, a construction module 153, and an identification module 154, where:

[0141] The acquisition module 151 is used to obtain images, Raman spectra, and physical and chemical property data of various biomasses;

[0142] The extraction module 152 is used to extract image feature parameters of the image data to obtain a first feature data set; and extract Raman feature parameters of the Raman spectrum to obtain a second feature data set;

[0143] The construction module 153 is used to establish a Hardgrove grindability prediction model, a moisture content prediction model, and a carbon-hydrogen element ratio prediction model based on the first feature data set and the second feature data set respectively;

[0144] The identification module 154 is used to obtain images, Raman spectra, and physical and chemical property data of the biomass to be measured, and identify the biomass to be measured using the three prediction models.

[0145] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying biomass based on machine vision and Raman laser coupling, characterized in that: The following steps are involved: Obtain images, Raman spectra and physicochemical property data of various biomasses; Extracting image feature parameters of the image data to obtain a first feature data set; and extracting Raman characteristic parameters of the Raman spectrum to obtain a second characteristic data set; Based on the first characteristic data set and the second characteristic data set, a Hastelloy grindability prediction model, a moisture content prediction model and a carbon-hydrogen ratio prediction model are established respectively; The images, Raman spectra and physical and chemical property data of the biomass to be tested are obtained, and the biomass to be tested is identified using three prediction models.

2. The method for identifying biomass based on machine vision and Raman laser coupling according to claim 1, characterized in that: The images are of biomass after it has been ground into particles.

3. The method for identifying biomass based on machine vision and Raman laser coupling according to claim 1, characterized in that: The image characteristic parameters include average particle size and grayscale.

4. The method for identifying biomass based on machine vision and Raman laser coupling according to claim 1, characterized in that: The physical and chemical property data include Hardgrove grindability index, water content and carbon-hydrogen ratio.

5. The method for identifying biomass based on machine vision and Raman laser coupling according to claim 1, characterized in that: The method for obtaining the Raman characteristic parameters includes: Obtain the total area S1 of the first-order Raman peak and the total area S2 of the second-order Raman peak through Raman spectroscopy; and the peak area of ​​each characteristic peak within the second-order Raman spectrum; The area ratio that can represent the characteristics of biomass is selected as the Raman characteristic parameter.

6. The method for identifying biomass based on machine vision and Raman laser coupling according to claim 1, characterized in that: The method for establishing the Hastelloy grindability prediction model comprises: The first characteristic data set and the second characteristic data set are used as input, and the Hardgrove grindability index corresponding to the biomass is used as output; The neural network is used for training to obtain the Hastelloy grindability prediction model.

7. The method for identifying biomass based on machine vision and Raman laser coupling according to claim 1, characterized in that: The method for establishing the moisture content prediction model comprises: The particle size feature and the grayscale feature in the first feature data set are coupled to obtain an image coupling feature; The image coupling feature and the second feature dataset are used as input, and the mass moisture content is used as output; The neural network is used for training to obtain the moisture content prediction model.

8. The method for identifying biomass based on machine vision and Raman laser coupling according to claim 1, characterized in that: The method for establishing the carbon-hydrogen element ratio prediction model comprises: The particle size feature and the grayscale feature in the first feature data set are coupled to obtain an image coupling feature; The particle size feature and grayscale feature are taken as input, and the image coupling feature is taken as output, and the first equation is obtained by fitting; The image coupling feature and the second feature data set are taken as input, the carbon-hydrogen element ratio prediction is taken as output, and the second equation is obtained by fitting; Substituting the first equation into the second equation, we can get the carbon-hydrogen ratio prediction model.

9. The method for identifying biomass based on machine vision and Raman laser coupling according to claim 7 or 8, characterized in that: The method of coupling the particle size feature and the grayscale feature in the first feature data set to obtain the image coupling feature includes: Feature fusion: weighted sum and normalization of particle size features and grayscale features to obtain a fused feature set; Depth information coupling: Calculate the depth information of each pixel in the image data, and form a binary array with the corresponding fusion feature set to obtain the image coupling feature.

10. A device for identifying biomass based on machine vision and Raman laser coupling, characterized in that: include: Acquisition module, used to obtain images, Raman spectra and physicochemical property data of various biomasses; An extraction module, used to extract image feature parameters of the image data to obtain a first feature data set; and extracting Raman characteristic parameters of the Raman spectrum to obtain a second characteristic data set; A construction module, for establishing a Hastelloy grindability prediction model, a moisture content prediction model, and a carbon-hydrogen ratio prediction model, respectively, based on the first characteristic data set and the second characteristic data set; The identification module is used to obtain the image, Raman spectrum and physical and chemical property data of the biomass to be tested, and use three prediction models to identify the biomass to be tested.

Images