Biomass charcoal identity discrimination method, system and device based on support vector machine

Abstract

This invention discloses a method, system, and device for biochar identification based on support vector machines, belonging to the field of biochar detection technology. This invention collects the main physicochemical properties of waste biomass from different sources and biochar prepared at different carbonization temperatures, and uses these collected physicochemical properties to construct a biochar identification model based on support vector machines. This model can effectively perform the task of identifying biochar with many different properties and improve the accuracy of biochar identification.

Description

Technical Field

[0001] This invention relates to the field of biochar detection technology, and in particular to a biochar identification method, system and equipment based on support vector machine. Background Technology

[0002] Biochar is a carbon-rich, porous solid material produced by thermal decomposition of biomass under oxygen-limited and relatively low-temperature (300℃~700℃) conditions. It features highly carboxylated and aromatic structures, large porosity and specific surface area, and extremely low solubility. It contains nutrients required by plants and has strong adsorption, antioxidant, and anti-biodegradation capabilities, giving it potential advantages in in-situ carbon storage, flue gas purification, pollutant fixation, wastewater treatment, and improving soil fertility.

[0003] Biochar technology has matured, but its social benefits depend not only on the abundance of biomass resources, but also on the carbon sequestration potential and economic feasibility of biochar. Converting waste biomass into biochar is a green development path that combines waste recycling with carbon sequestration and emission reduction.

[0004] Due to differences in raw material types, technical methods, and pyrolysis processes, biochar exhibits significant variations in its physicochemical properties, such as structure, composition, pore volume, and specific surface area, resulting in diverse environmental effects. As the application fields of biochar continue to expand, it is crucial to classify biochar with different properties, such as carbon storage value, fertilizer value, lime equivalent value, and particle size, to standardize product quality, facilitate the selection of suitable biochar for application, promote the diversification, standardization, and serialization of biochar products, and contribute to the sustainable development of the biochar industry. Summary of the Invention

[0005] The purpose of this invention is to provide a biochar identification method, system, and device based on support vector machine. By constructing a biochar identification model based on support vector machine, the invention can complete the identification of biochar with different properties and improve the accuracy of biochar identification.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] A biochar identification method based on support vector machine includes:

[0008] The physicochemical properties of multiple biochar samples are used to construct an input variable matrix, and each biochar sample is coded with an identity category label; the multiple biochar samples include waste biomass from different sources and biochar prepared from waste biomass from different sources at different carbonization temperatures;

[0009] Principal component analysis was used to perform principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and the principal component matrix.

[0010] The principal component matrix and the identity category labels constitute the feature dataset of the biochar samples;

[0011] Based on the identity category label, an SVM classifier is established between any two types of biochar samples in the feature dataset of the biochar samples, and all the established SVM classifiers constitute a biochar identity discrimination model.

[0012] The physicochemical properties of individual biochar samples to be predicted are obtained, and the principal components are obtained through loading coefficient matrix transformation.

[0013] The principal component is input into all SVM classifiers in the biochar identity discrimination model to obtain the identity category prediction result of each SVM classifier.

[0014] A one-to-one classification strategy was used to statistically predict the identity category with the most votes, and this prediction was used as the biochar identity category of the individual biochar sample to be predicted.

[0015] Optionally, the step of constructing an input variable matrix from the physicochemical properties of multiple collected biochar samples and encoding the identity category label of each biochar sample specifically includes:

[0016] Carbon storage value, fertility value, pH, and particle size distribution were selected as initial physicochemical property parameters.

[0017] Specifically define the indexes for each initial physicochemical property parameter, and use all the indexes as the physicochemical property parameters of the biochar samples;

[0018] The experiment measured the physicochemical properties of each biochar sample and compiled the physicochemical properties into an input vector for each biochar sample in the form of row vectors. Then, the input vectors of each biochar sample were arranged in row order to form an input variable matrix.

[0019] Each biochar sample is tagged with a digital symbol to identify its category.

[0020] Optionally, the step of constructing an input variable matrix from the physicochemical properties of multiple collected biochar samples and encoding the identity category label of each biochar sample further includes:

[0021] Calculate the mean and standard deviation of the column vectors of each physicochemical property parameter in the input variable matrix;

[0022] Based on the mean and standard deviation of the column vectors of each physicochemical property parameter, the formula is used. The column vectors of each physicochemical property parameter in the input variable matrix are standardized to obtain the standardized input variable matrix; where, Let x represent the standardized column vector of the l-th physicochemical property parameters. l This represents the column vector of the l-th physicochemical property parameters. and s l Let represent the mean and standard deviation of the column vector of the l-th physicochemical property parameter, respectively, where l = 1, 2, ..., p, and p represents the number of physicochemical property parameters in the input variable matrix.

[0023] Optionally, the step of performing principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and the principal component matrix specifically includes:

[0024] Based on the input variable matrix X and the standardized input variable matrix Using formula Calculate the correlation matrix R;

[0025] Calculate the eigenvalues ​​and eigencolumn vectors of the correlation matrix R;

[0026] Based on the eigenvalues ​​and eigencolumn vectors of the correlation matrix R, using the formula... Calculate the contribution rate of each principal feature; where γ q Let λ be the contribution rate of the q-th principal feature. q Let λ be the q-th eigenvalue. i Let p' be the i-th eigenvalue, and p' be the number of eigenvalues.

[0027] Sort the contribution rates of each principal feature from largest to smallest. Starting with the number of principal features h=1, increase the number of principal features one by one, and then use the formula... Calculate the corresponding cumulative principal component contribution rate until the cumulative principal component contribution rate increases to greater than or equal to 95%, then determine the number of cumulative characteristic principal components m at this time as the number of principal component physicochemical property parameters; where p″ is the number of characteristic principal components, and the value of p″ is equal to the value of p'.

[0028] The loading coefficient column vectors corresponding to each feature principal element are used to form a loading coefficient matrix U = [u1, u2, ..., u]. m ];

[0029] Based on the standardized input variable matrix X and the loading coefficient matrix U = [u1, u2, ..., u] m Using the formula Determine the principal component matrix T.

[0030] Optionally, based on the identity category label, an SVM classifier is established between any two classes of biochar samples in the feature dataset of the biochar samples, and all established SVM classifiers constitute a biochar identity discrimination model, specifically including:

[0031] The objective function for determining the optimal hyperplane that partitions any two classes of biochar samples in the feature dataset of the biochar samples is: In the formula, F min Let w be the objective function, and ξ be the weight coefficient vector. j Let ξ be the slack variable controlling the classification error for the j-th biochar sample. j ≥0, c is the error penalty factor, || || is the norm;

[0032] The inner product operation of the feature space samples is performed using the Gaussian kernel function;

[0033] A one-to-one classification strategy is adopted to establish an SVM classifier between any two classes of biochar samples in the feature dataset of the biochar samples.

[0034] The error penalty factor c and the bandwidth of the Gaussian kernel function in the objective function are determined as the parameters to be optimized.

[0035] A gridded approach was used to optimize the parameters, resulting in multiple optimized SVM classifiers. All optimized SVM classifiers were then combined to form a biochar identification model.

[0036] A biochar identification system based on support vector machine includes:

[0037] The sample data acquisition module is used to construct an input variable matrix from the physicochemical properties of multiple collected biochar samples and to encode the identity category label of each biochar sample; the multiple biochar samples include waste biomass from different sources and biochar prepared from waste biomass from different sources at different carbonization temperatures;

[0038] The principal component analysis module is used to perform principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and the principal component matrix.

[0039] The sample set composition module is used to construct a feature dataset of biochar samples by combining the principal component matrix and the identity category labels.

[0040] The discrimination model building module is used to build an SVM classifier between any two types of biochar samples in the feature dataset of the biochar samples based on the identity category label, and to form a biochar identity discrimination model by combining all the built SVM classifiers.

[0041] The module for obtaining parameters to be predicted is used to obtain the physicochemical properties of individual biochar samples to be predicted, and to obtain the principal components through the transformation of the loading coefficient matrix.

[0042] The prediction module is used to input the principal component into all SVM classifiers in the biochar identity discrimination model and obtain the identity category prediction result of each SVM classifier.

[0043] The identity category determination module is used to statistically predict the identity category with the most votes using a one-to-one classification strategy, and use this prediction as the biochar identity category of the individual biochar sample to be predicted.

[0044] A biochar identification device based on support vector machine, comprising:

[0045] A memory for storing computer software programs; the computer software programs are used to implement the biochar identification method based on support vector machines as described above.

[0046] A processor, connected to the memory, is used to retrieve and execute the computer software program.

[0047] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0048] This invention discloses a biochar identification method, system, and device based on support vector machine. It collects the main physicochemical properties of waste biomass from different sources and biochar prepared at different carbonization temperatures, and uses the collected physicochemical properties to construct a biochar identification model based on support vector machine. This model can effectively complete the task of identifying biochar with many different properties and improve the accuracy of biochar identification. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 The architecture diagram of an existing support vector machine;

[0051] Figure 2 A flowchart of a biochar identification method based on support vector machine provided in an embodiment of the present invention;

[0052] Figure 3 A simplified schematic diagram of a biochar identification method based on support vector machine provided in an embodiment of the present invention;

[0053] Figure 4 A three-dimensional diagram of the error penalty coefficient c and kernel function bandwidth σ optimization process of the SVM classifier provided in this embodiment of the invention;

[0054] Figure 5 A two-dimensional diagram illustrating the optimization process of the error penalty coefficient c and kernel function bandwidth σ of the SVM classifier provided in this embodiment of the invention.

[0055] Figure 6 This is a diagram showing the classification results of sample data for biochar identification based on support vector machine, as provided in an embodiment of the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] The purpose of this invention is to provide a biochar identification method, system, and device based on support vector machine. By constructing a biochar identification model based on support vector machine, the invention can complete the identification of biochar with different properties and improve the accuracy of biochar identification.

[0058] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0059] Support Vector Machine (SVM) is a generalized linear classifier that classifies data samples under supervised learning. When samples are linearly separable, the optimal hyperplane for classification between the two classes is found in the original space. When samples are linearly inseparable, slack variables are added, and a nonlinear mapping is used to map the samples from the low-dimensional input space to a high-dimensional feature space. Then, the optimal hyperplane for linear separability is found in this high-dimensional feature space. Figure 1 The architecture diagram of Support Vector Machine (SVM) is shown. Because its loss function incorporates a regularization term to optimize structural risk when calculating empirical risk, it is a sparse and robust classifier with strong versatility and good performance on small samples, often making it the preferred method among many classification algorithms. Therefore, this invention provides a biochar identification method based on SVM, constructing a biochar identification model based on SVM, which has advantages such as strong specificity, good repeatability, high discrimination accuracy, and reliable results.

[0060] like Figure 2 As shown in the embodiment of the present invention, a biochar identification method based on support vector machine is provided, comprising:

[0061] Step 1: Construct an input variable matrix from the physicochemical properties of multiple biochar samples and encode the identity category label of each biochar sample; the multiple biochar samples include waste biomass from different sources and biochar prepared from waste biomass from different sources at different carbonization temperatures.

[0062] Step 1 corresponds to Figure 3 The "Biochar Sample Data Construction" section organizes the physicochemical properties and other data collected experimentally from waste biomass and its carbonized products, and thereby establishes the input vector of the sample data and the identity category labels of waste biomass and its carbonized products. The specific implementation process is as follows:

[0063] Step 1.1: Screening the physicochemical properties of waste biomass and its carbonized products: Based on the raw material source of biochar and the application value of carbonized biochar, select four types of physicochemical properties: carbon storage value, fertilizer value, pH, and particle size distribution.

[0064] Step 1.2, define specific physicochemical properties: For carbon storage value, hydrogen (H) content (%) and organic carbon concentration (Corg) (%) are selected; for fertility value, the contents (%) of nitrogen (N), phosphorus (P), potassium (K), sulfur (S), calcium (Ca), and magnesium (Mg) are selected; for acidity / alkalinity, pH value is selected; for particle size distribution, specific surface area (BET) (m²) is selected. 2 / g) and pore size Vpore (cm) 3 / g), the number of physicochemical property parameters p = 11.

[0065] Step 1.3, Construct the input vector: Encode the physicochemical property parameters defined in Step 1.2 sequentially as x1, x2, ..., x 11 The input vector of each sample individual is composed of rows and columns, namely x = [x1, x2, ..., x]. 11 The dimension of the input variables is the number of physicochemical property parameters.

[0066] Step 1.4, Experimental Input Variable Matrix: Different types of waste biomass raw materials and biochar prepared at three carbonization temperatures (low temperature (350℃), medium temperature (500℃), and high temperature (650℃)) were selected. The physicochemical characteristic parameters p = 11 defined in Step 1.2 were experimentally measured. The input vector x for each sample was then compiled according to Step 1.3, and the input vectors of each sample were arranged in row order to obtain the input variable matrix X of the biochar sample data. The number of samples is called the sample size n, where n = 24.

[0067] Step 1.5: Encode the identity category labels of waste biomass and its carbonized biochar. The sources of waste biomass include six categories: corn stalks, rice stalks, perishable waste, pecan shells, cow dung, and pig dung, i.e., ω = 6. The identity category information of waste biomass is classified according to its source, while the identity category information of biochar is attributed to the identity category information of its raw materials. Therefore, the identity categories of waste biomass and biochar produced at three different carbonization temperatures are labeled with the numerical symbols 1, 2, 3, 4, 5, and 6, respectively. Based on the row numbers of the input variable matrix X of the biochar sample data obtained in step S14, the identity category label of each individual sample is encoded, forming the identity category label column vector y of the biochar sample data.

[0068] Step 1.6, Construct biochar sample data: Combine the input variable matrix X and the identity category label column vector y to obtain the biochar sample data {X,y}. The dimensions of X and y are n×p and n×1, respectively.

[0069] To reduce the influence of the dimensions and orders of magnitude of various physicochemical properties on the discriminant model, standardization processing is implemented on the input variable matrix of the sample data. Figure 3 The "input variable matrix standardization process" in the text.

[0070] The standardization method for the input variable matrix is ​​as follows:

[0071] Step S1: Calculate the mean and standard deviation of each physicochemical property parameter in the input variable matrix: Take the input variable matrix X containing the sample data from step 1.4, and calculate the mean of each physicochemical property parameter column by column. and standard deviation s l (l=1,2,…,p).

[0072] Step S2, Input variable matrix standardization: Based on the input variable matrix X, implement the standardization of the column vectors x of each physicochemical property parameter according to equation (1). l The input variable matrix is ​​standardized and denoted as X.

[0073]

[0074] In the formula, Let x represent the standardized column vector of the l-th physicochemical property parameters. l This represents the column vector of the l-th physicochemical property parameters. and s l Let represent the mean and standard deviation of the column vector of the l-th physicochemical property parameter, respectively, where l = 1, 2, ..., p, and p represents the number of physicochemical property parameters in the input variable matrix.

[0075] Step 2: Use principal component analysis to perform principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and principal component matrix.

[0076] Step 2 corresponds to Figure 3 "Input variable PCA feature extraction" in the text.

[0077] For example, the method for determining the principal component matrix is ​​as follows:

[0078] Step 2.1: Obtain the standardized input variable matrix from step S2. Calculate the correlation matrix

[0079] Step 2.2, calculate the contribution rate of each extractable PCA (Principal Component Analysis) eigenvalue: Based on the correlation matrix R, calculate each eigenvalue λ. l and the corresponding load factor column vector u l (l=1,2,…,p), and then calculate the contribution rate γ of each PCA feature principal component using equation (2). q .

[0080]

[0081] In the formula, γ q Let λ be the contribution rate of the q-th principal feature. q Let λ be the q-th eigenvalue. i Let p' be the i-th eigenvalue, and p' be the number of eigenvalues. P' is numerically equal to the number of physicochemical property parameters in the input variable matrix.

[0082] Step 2.3, Optimize the number of selected PCA feature principal components: After sorting the contribution rates of each PCA feature principal component from largest to smallest, denoted as γ. h Then, starting from the number of PCA feature principal components h=1, the number of principal components is increased one by one, and the corresponding cumulative principal component contribution rate Ω is calculated by equation (3). m When Ω m When the percentage increases to 95% or higher, determine the number of PCA feature principal components m (m = 2 in this example).

[0083]

[0084] In the formula, p″ represents the number of principal components, which is numerically equal to the number of physicochemical property parameters p in the input variable matrix, and also equal to the number of eigenvalues ​​p'.

[0085] Step 2.4, calculate the PCA feature principal component matrix of the input variable matrix: based on the contribution rates γ of the m PCA feature principal components used in step 2.3. hFirst, extract the corresponding eigenvalues ​​λ from step 2.2. h and load factor column vector u h (h = 1, 2, ..., m), and form the load factor matrix U = [u1, u2, ..., u] using these m load factor column vectors. m Finally, the PCA feature principal matrix is ​​calculated using equation (4).

[0086]

[0087] Step 3: Combine the principal component matrix and the identity category labels to form the feature dataset of the biochar samples.

[0088] Step 4: Based on the identity category label, establish an SVM classifier between any two types of biochar samples in the feature dataset of the biochar samples, and combine all the established SVM classifiers to form a biochar identity discrimination model.

[0089] Step 4 corresponds to Figure 3 "Optimization and Construction of SVM Classification Model" in the text.

[0090] The method for constructing an optimized SVM classifier is as follows:

[0091] Step 4.1, SVM optimizes the objective function of the classifier: Let the nonlinear mapping of m-dimensional PCA feature principal component t be... The optimal hyperplane for partitioning between any two classes based on the principal component matrix T of S34 and the identity category label column vector y of S15. (j = 1, 2, ..., 24, w is the weight coefficient vector, b is the bias constant term, ξ) j The objective function (≥0 is the relaxation variable controlling classification error) is: (c is the error penalty factor).

[0092] Step 4.2, perform inner product operation on the feature space samples using a kernel function: the SVM classifier uses a Gaussian kernel function. (i,j=1,2,…,24, σ is the kernel function bandwidth), the inner product operation of the feature space samples is performed through the kernel function.

[0093] Step 4.3 employs a one-versus-one classification strategy: an SVM classifier is constructed between any two classes of biochar samples. For biochar samples with ω classes, the method for identifying their identities is determined by... It consists of several SVM classifiers.

[0094] Step 4.4, Parameter Optimization in SVM Classifiers: The Gaussian kernel function bandwidth σ in Step 4.2 and the error penalty factor c in Step S4.1 have a significant impact on the performance of the d SVM classifiers in Step 4.3. Therefore, a gridded approach is used to find the optimal c and σ. First, the optimization interval of c is set to [1, 10] with a step size of 1, and the optimization interval of σ is set to [0.1, 2.0] with a step size of 0.1. The different levels of c and σ form a total of 200 permutations and combinations. Then, the accuracy η of each SVM classifier in judging the sample data is calculated using Equation (5) as the optimization objective to select the optimal c and σ (where n is the accuracy η). k (The number of correctly identified samples of the k-th class).

[0095]

[0096] Figure 4 This is a three-dimensional diagram of the error penalty coefficient c and kernel function bandwidth σ optimization process for the SVM classifier. Figure 5 This is a two-dimensional graph showing the error penalty coefficient c and kernel function bandwidth σ optimization process of the SVM classifier.

[0097] Step 4.5, Biochar identification model: Take out the best c and σ found in step 4.4, and establish a biochar identification model composed of d SVM optimized classifiers.

[0098] Step 5: Obtain the physicochemical properties of the individual biochar samples to be predicted, and obtain the principal components through loading coefficient matrix transformation.

[0099] Step 6: Input the principal component into all SVM classifiers in the biochar identity discrimination model to obtain the identity category prediction result of each SVM classifier.

[0100] Step 7: Use a one-to-one classification strategy to statistically analyze the identity category prediction results with the most votes, and use it as the biochar identity category of the individual biochar sample to be predicted.

[0101] Steps 6 and 7 correspond to Figure 3 The article discusses "determining the identity of biochar based on the SVM classification model".

[0102] Figure 6 This is a diagram showing the classification results of sample data for biochar identification based on support vector machine. Figure 6 In this context, Predicted class represents the predicted class, False DiscoveryRate represents the false discovery rate, PositivePredictiveValue represents the positive prediction value, and True class represents the true class.

[0103] The biochar identification method based on support vector machine provided by this invention uses experimental data of the physicochemical properties of biochar to identify biochar. Specifically, for different types of waste biomass and biochar prepared at different carbonization temperatures, the main physicochemical properties of these waste biomass and biochar are sampled and measured. A biochar identification model is then established based on support vector machine, which can effectively complete the task of biochar identification and has the advantages of strong specificity, good repeatability, high identification accuracy, and reliable results.

[0104] This invention also provides a biochar identification system based on support vector machines, comprising:

[0105] The sample data acquisition module is used to construct an input variable matrix from the physicochemical properties of multiple collected biochar samples and to encode the identity category label of each biochar sample; the multiple biochar samples include waste biomass from different sources and biochar prepared from waste biomass from different sources at different carbonization temperatures;

[0106] The principal component analysis module is used to perform principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and the principal component matrix.

[0107] The sample set composition module is used to construct a feature dataset of biochar samples by combining the principal component matrix and the identity category labels.

[0108] The discrimination model building module is used to build an SVM classifier between any two types of biochar samples in the biochar sample set based on the identity category label, and to form a biochar identity discrimination model by combining all the built SVM classifiers.

[0109] The module for obtaining parameters to be predicted is used to obtain the physicochemical properties of individual biochar samples to be predicted, and to obtain the principal components through the transformation of the loading coefficient matrix.

[0110] The prediction module is used to input the principal component into all SVM classifiers in the biochar identity discrimination model and obtain the identity category prediction result of each SVM classifier.

[0111] The identity category determination module is used to statistically predict the identity category with the most votes using a one-to-one classification strategy, and use this prediction as the biochar identity category of the individual biochar sample to be predicted.

[0112] Furthermore, embodiments of the present invention also provide a biochar identification device based on support vector machines, comprising:

[0113] A memory for storing computer software programs; the computer software programs are used to implement the aforementioned biochar identification method based on support vector machines.

[0114] A processor, connected to the memory, is used to retrieve and execute the computer software program.

[0115] The processor includes:

[0116] The sample data acquisition module is used to construct an input variable matrix from the physicochemical properties of multiple collected biochar samples and to encode the identity category label of each biochar sample; the multiple biochar samples include waste biomass from different sources and biochar prepared from waste biomass from different sources at different carbonization temperatures;

[0117] The principal component analysis module is used to perform principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and the principal component matrix.

[0118] The sample set composition module is used to construct a feature dataset of biochar samples by combining the principal component matrix and the identity category labels.

[0119] The discrimination model building module is used to build an SVM classifier between any two types of biochar samples in the biochar sample set based on the identity category label, and to form a biochar identity discrimination model by combining all the built SVM classifiers.

[0120] The module for obtaining parameters to be predicted is used to obtain the physicochemical properties of individual biochar samples to be predicted, and to obtain the principal components through the transformation of the loading coefficient matrix.

[0121] The prediction module is used to input the principal component into all SVM classifiers in the biochar identity discrimination model and obtain the identity category prediction result of each SVM classifier.

[0122] The identity category determination module is used to statistically predict the identity category with the most votes using a one-to-one classification strategy, and use this prediction as the biochar identity category of the individual biochar sample to be predicted.

[0123] For example, the memory is a computer-readable storage medium.

[0124] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0125] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A biochar identification method based on support vector machine, characterized in that, include: The physicochemical properties of multiple collected biochar samples are used to construct an input variable matrix, and the identity category label of each biochar sample is encoded. The multiple biochar samples include waste biomass from different sources and biochar prepared from waste biomass from different sources at different carbonization temperatures; the biochar properties in the multiple biochar samples are different; the sources of waste biomass include 6 categories, and its identity category information is classified according to the source of waste biomass, while the identity category information of biochar is attributed to the identity category information of its raw materials. Thus, the identity categories of waste biomass and biochar prepared at 3 different carbonization temperatures are labeled with the numerical symbols 1, 2, 3, 4, 5, and 6, respectively; based on the raw material source of biochar and the application value of carbonized biochar, four types of physicochemical characteristic parameters are selected: carbon storage value, fertility value, pH, and particle size distribution. Principal component analysis was used to perform principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and the principal component matrix. The principal component matrix and the identity category labels constitute the feature dataset of the biochar samples; Based on the identity category label, an SVM classifier is established between any two types of biochar samples in the feature dataset of the biochar samples, and all the established SVM classifiers constitute a biochar identity discrimination model. The physicochemical properties of individual biochar samples to be predicted are obtained, and the principal components are obtained through loading coefficient matrix transformation. The principal component is input into all SVM classifiers in the biochar identity discrimination model to obtain the identity category prediction result of each SVM classifier. A one-to-one classification strategy was used to statistically predict the identity category with the most votes, and this prediction was used as the biochar identity category of the individual biochar sample to be predicted.

2. The biochar identification method based on support vector machine according to claim 1, characterized in that, The process involves constructing an input variable matrix from the physicochemical properties of multiple collected biochar samples and encoding the identity category label for each biochar sample, specifically including: Carbon storage value, fertility value, pH, and particle size distribution were selected as initial physicochemical property parameters. Specifically define the indexes for each initial physicochemical property parameter, and use all the indexes as the physicochemical property parameters of the biochar samples; The experiment measured the physicochemical properties of each biochar sample and compiled the physicochemical properties into an input vector for each biochar sample in the form of row vectors. Then, the input vectors of each biochar sample were arranged in row order to form an input variable matrix. Each biochar sample is tagged with a digital symbol to identify its category.

3. The biochar identification method based on support vector machine according to claim 1, characterized in that, The process involves constructing an input variable matrix from the physicochemical properties of multiple collected biochar samples and encoding the identity category label for each biochar sample. This is followed by: Calculate the mean and standard deviation of the column vectors of each physicochemical property parameter in the input variable matrix; Based on the mean and standard deviation of the column vectors of each physicochemical property parameter, the formula is used. The column vectors of each physicochemical property parameter in the input variable matrix are standardized to obtain the standardized input variable matrix; where, Represents the standardized first l A column vector of physicochemical property parameters Indicates the first l A column vector of physicochemical property parameters and They represent the first l Mean and standard deviation of the column vector of individual physicochemical property parameters ， ，p This indicates the number of physicochemical property parameters in the input variable matrix.

4. The biochar identification method based on support vector machine according to claim 3, characterized in that, The step of performing principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and the principal component matrix specifically includes: Based on the input variable matrix and the standardized input variable matrix Using the formula Calculate the correlation matrix ; Calculate the correlation matrix eigenvalues ​​and eigencolumn vectors; According to the relevant matrix The eigenvalues ​​and eigencolumn vectors are obtained using the formula. Calculate the contribution rate of each principal feature; where, For the first q Contribution rate of each principal feature, For the first q 1 eigenvalue, For the first i 1 eigenvalue, p ' represents the number of eigenvalues; Sort the contribution rates of each feature principal component from largest to smallest, based on the number of feature principal components. Starting from this point, the number of key features is gradually increased, and then the formula is used to... Calculate the corresponding cumulative principal component contribution rate until the cumulative principal component contribution rate increases to greater than or equal to 95%, then determine the cumulative number of feature principal components at this point. m The number of physicochemical property parameters of the main component; where, The number of characteristic principal components, The value and p The values ​​of ' are equal; The loading coefficient column vectors corresponding to each feature principal element are used to form a loading coefficient matrix. ; Based on the standardized input variable matrix and load factor matrix Using the formula Determine the principal component matrix T .

5. The biochar identification method based on support vector machine according to claim 1, characterized in that, Based on the identity category label, an SVM classifier is established between any two classes of biochar samples in the feature dataset of the biochar samples, and all established SVM classifiers constitute a biochar identity discrimination model, specifically including: The objective function for determining the optimal hyperplane that partitions any two classes of biochar samples in the feature dataset of the biochar samples is: In the formula, Let be the objective function. For weight coefficient vectors, For the first j The slack variable controlling for classification error for each biochar sample , As the error penalty factor, It is a norm; The inner product operation of the feature space samples is performed using the Gaussian kernel function; A one-to-one classification strategy is adopted to establish an SVM classifier between any two classes of biochar samples in the feature dataset of the biochar samples. Determine the error penalty factor in the objective function The bandwidth of the Gaussian kernel function is the parameter to be optimized. A gridded approach was used to optimize the parameters, resulting in multiple optimized SVM classifiers. All optimized SVM classifiers were then combined to form a biochar identification model.

6. A biochar identification system based on support vector machine, characterized in that, include: The sample data acquisition module is used to construct an input variable matrix from the physicochemical properties of multiple collected biochar samples and encode the identity category label of each biochar sample. The multiple biochar samples include waste biomass from different sources and biochar prepared from waste biomass from different sources at different carbonization temperatures. The biochar properties in the multiple biochar samples are different. The sources of waste biomass include six categories, and its identity category information is classified according to its source. The identity category information of biochar is attributed to the identity category information of its raw materials. Therefore, the identity categories of waste biomass and biochar prepared at three different carbonization temperatures are labeled with the numerical symbols 1, 2, 3, 4, 5, and 6, respectively. Based on the raw material source of the biochar and the application value of the carbonized biochar, four types of physicochemical property parameters are selected: carbon storage value, fertility value, pH, and particle size distribution. The principal component analysis module is used to perform principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and the principal component matrix. The sample set construction module is used to construct a feature dataset of biochar samples by combining the principal component matrix and the identity category labels. The discrimination model building module is used to build an SVM classifier between any two types of biochar samples in the feature dataset of the biochar samples based on the identity category label, and to form a biochar identity discrimination model by combining all the built SVM classifiers. The module for obtaining parameters to be predicted is used to obtain the physicochemical properties of individual biochar samples to be predicted, and to obtain the principal components through the transformation of the loading coefficient matrix. The prediction module is used to input the principal component into all SVM classifiers in the biochar identity discrimination model and obtain the identity category prediction result of each SVM classifier. The identity category determination module is used to statistically predict the identity category with the most votes using a one-to-one classification strategy, and use this prediction as the biochar identity category of the individual biochar sample to be predicted.

7. A biochar identification device based on support vector machine, characterized in that, include: A memory for storing computer software programs; the computer software programs are used to implement the biochar identification method based on support vector machines as described in any one of claims 1-5; A processor, connected to the memory, is used to retrieve and execute the computer software program.

8. The biochar identification device based on support vector machine according to claim 7, characterized in that, The processor includes: The sample data acquisition module is used to construct an input variable matrix from the physicochemical properties of multiple collected biochar samples and to encode the identity category label of each biochar sample; the multiple biochar samples include waste biomass from different sources and biochar prepared from waste biomass from different sources at different carbonization temperatures; The principal component analysis module is used to perform principal component analysis on the physicochemical property parameters in the input variable matrix to obtain the loading coefficient matrix and the principal component matrix. The sample set construction module is used to construct a feature dataset of biochar samples by combining the principal component matrix and the identity category labels. The discrimination model building module is used to build an SVM classifier between any two types of biochar samples in the feature dataset of the biochar samples based on the identity category label, and to form a biochar identity discrimination model by combining all the built SVM classifiers. The module for obtaining parameters to be predicted is used to obtain the physicochemical properties of individual biochar samples to be predicted, and to obtain the principal components through the transformation of the loading coefficient matrix. The prediction module is used to input the principal component into all SVM classifiers in the biochar identity discrimination model and obtain the identity category prediction result of each SVM classifier. The identity category determination module is used to statistically predict the identity category with the most votes using a one-to-one classification strategy, and use this prediction as the biochar identity category of the individual biochar sample to be predicted.

9. The biochar identification device based on support vector machine according to claim 7, characterized in that, The memory is a computer-readable storage medium.

Images