Glass classification method, device, equipment and medium based on classification model

Abstract

The application relates to a glass classification method and device based on a classification model, equipment and a medium. The method comprises the following steps: automatically extracting chemical composition discrimination features of glass sample data by constructing a machine learning-based classification model, constructing a linear discrimination function according to the discrimination features and the discrimination units in the classification model, preliminarily classifying the glass sample data based on the linear discrimination function, further using a sub-classification unit in the classification model to perform detailed sub-classification on the preliminary classification result by using a K-means clustering algorithm, and obtaining a sub-classification result. By using the method, the classification model can be learned and trained through large-scale data, different chemical composition features of the glass sample data can be extracted based on the classification model, the glass sample data can be preliminarily classified and sub-classified, and the accuracy and reliability of glass classification are improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a glass classification method, apparatus, device and medium based on a classification model. Background Technology

[0002] The main raw material for ancient glass was quartz sand, whose primary chemical component was silicon dioxide (SiO2). Because pure quartz sand has a high melting point, fluxes were added during the refining process to lower the melting temperature. Commonly used fluxes in ancient times included wood ash, natural caustic soda, saltpeter, and lead ore, with limestone added as a stabilizer. After calcination, the limestone was converted into calcium oxide (CaO). Due to the different fluxes used, the main chemical composition of ancient glass varied. For example, lead-barium glass, with lead ore added as a flux during firing, had a high content of lead oxide (PbO) and barium oxide (BaO); high-potassium glass was made using substances with high potassium content, such as wood ash, as a flux.

[0003] Currently, the commonly used method for glass classification is chemical analysis. This technique involves analyzing the chemical composition of glass samples to determine their chemical composition, thereby classifying them. However, ancient glass is highly susceptible to weathering due to its burial environment. During weathering, a significant amount of internal elements exchange with environmental elements, causing changes in the proportions of their components. This affects the accuracy of classification results obtained through chemical analysis. Therefore, it is necessary to propose a highly accurate classification method for ancient glass. Summary of the Invention

[0004] Therefore, it is necessary to provide a glass classification method, apparatus, equipment, and medium based on a classification model to address the aforementioned technical problems.

[0005] A glass classification method based on a classification model, the method comprising:

[0006] Obtain the glass sample dataset and divide it into training and test sets;

[0007] The glass sample data in the training set is input into a pre-built classification model for training until the preset training termination condition is met, resulting in a trained classification model. The classification model includes a feature extraction unit, a discriminant unit, and a sub-classification unit connected in sequence. The feature extraction unit extracts chemical composition discrimination features from the glass sample data. The discriminant unit calculates the discrimination coefficients of the chemical composition discrimination features using the Fisher discriminant method and constructs a linear discriminant function based on these coefficients to classify the glass sample data, obtaining a preliminary classification result. The sub-classification unit uses the K-means clustering algorithm to subdivide the preliminary classification result into subclasses, obtaining the subclass division result.

[0008] The glass sample data to be classified in the test set is input into the trained classification model to classify the glass type, and the output is the type to which the glass sample data belongs.

[0009] In one embodiment, the chemical composition discrimination features include the content of potassium oxide, lead oxide, barium oxide, strontium oxide, and silicon dioxide in the glass sample data.

[0010] In one embodiment, the discrimination unit is further configured to calculate the discrimination coefficients of high-potassium glass and lead-barium glass based on the Fisher discrimination method, and to construct linear discrimination functions for high-potassium glass and lead-barium glass based on the high-potassium glass and lead-barium glass, respectively, expressed as follows:

[0011]

[0012]

[0013] Among them, F 高钾 Represents the linear discriminant function of high-potassium glass, F 铅钡 This represents the linear discriminant function for lead-barium glass. X PbO X BaO X SrO These represent the contents of potassium oxide, lead oxide, barium oxide, strontium oxide, and silicon dioxide, respectively, in the chemical composition discrimination characteristics. C PbO C BaO C SrO The discriminant coefficients for high-potassium glass, representing potassium oxide, lead oxide, barium oxide, strontium oxide, and silicon dioxide content, are respectively. W PbO W BaO W SrO The discrimination coefficients of lead-barium glass represent the contents of potassium oxide, lead oxide, barium oxide, strontium oxide, and silicon dioxide, and L1 and L2 represent the first and second set constants, respectively.

[0014] According to F 高钾 and F 铅钡 The glass sample data were classified to obtain preliminary classification results, where, when F 铅钡 <F 高钾 When F is used, the glass sample data is classified as high-potassium glass samples. 铅钡 ≥F 高钾 At that time, the glass sample data were classified as lead-barium glass samples.

[0015] In one embodiment, the subclassification unit is also used to perform low variance filtering on the chemical composition discrimination features of high potassium glass samples and lead-barium glass samples to obtain the subclass classification features of high potassium glass samples and lead-barium glass samples.

[0016] The K-means clustering algorithm was used to cluster the subclassing features of high-potassium glass samples and lead-barium glass samples, respectively, to obtain high-potassium glass subclass samples and lead-barium glass subclass samples.

[0017] In one embodiment, low-variance filtering is applied to the chemical composition discrimination features of high-potassium glass samples and lead-barium glass samples to obtain subclass classification features for high-potassium glass samples and lead-barium glass samples, including:

[0018] Low variance filtering was applied to the chemical composition discrimination features of high potassium glass samples to obtain the variance of the content of different chemical components in the chemical composition discrimination features of high potassium glass samples and arranged in descending order. The content of the two chemical components with the highest variance was selected as the subclass classification features of high potassium glass samples.

[0019] Low-variance filtering was applied to the chemical composition discrimination features of lead-barium glass samples to obtain the variance of the content of different chemical components in the chemical composition discrimination features of lead-barium glass samples. The variances were then arranged in descending order, and the contents of the two chemical components with the highest variances were selected as the subclass classification features of lead-barium glass samples.

[0020] In one embodiment, the subclassing features of high-potassium glass samples and lead-barium glass samples are clustered according to the K-means clustering algorithm to obtain high-potassium glass subclass samples and lead-barium glass subclass samples, including:

[0021] The high-potassium glass samples were clustered based on the subclassing characteristics of K-means clustering, and the number of clusters was set to 3, resulting in three subclasses of high-potassium glass samples.

[0022] The lead-barium glass samples were clustered based on the subclassing characteristics of K-means clustering, and the number of clusters was set to 4, resulting in four subclasses of lead-barium glass samples.

[0023] In one embodiment, the preset training termination condition includes at least one of the following:

[0024] The accuracy of the classification model has reached the preset accuracy threshold;

[0025] The number of times the training classification model is executed exceeds the preset loop count threshold;

[0026] The classification model converged.

[0027] A glass classification device based on a classification model, the device comprising:

[0028] The data acquisition module is used to acquire the glass sample dataset and divide the glass sample dataset into a training set and a test set.

[0029] The model training module is used to input glass sample data from the training set into a pre-built classification model for training until a preset training termination condition is met, resulting in a trained classification model. The classification model includes a feature extraction unit, a discriminant unit, and a sub-classification unit connected in sequence. The feature extraction unit extracts chemical composition discrimination features from the glass sample data. The discriminant unit calculates the discrimination coefficients of the chemical composition discrimination features using the Fisher discriminant method and constructs a linear discriminant function based on these coefficients to classify the glass sample data, obtaining a preliminary classification result. The sub-classification unit uses the K-means clustering algorithm to subdivide the preliminary classification result into subclasses, obtaining the subclass division result.

[0030] The glass classification module is used to input the glass sample data to be classified from the test set into the trained classification model to classify the glass type, and output the type to which the glass sample data belongs.

[0031] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program performing the following steps:

[0032] Obtain the glass sample dataset and divide it into training and test sets;

[0033] The glass sample data in the training set is input into a pre-built classification model for training until the preset training termination condition is met, resulting in a trained classification model. The classification model includes a feature extraction unit, a discriminant unit, and a sub-classification unit connected in sequence. The feature extraction unit extracts chemical composition discrimination features from the glass sample data. The discriminant unit calculates the discrimination coefficients of the chemical composition discrimination features using the Fisher discriminant method and constructs a linear discriminant function based on these coefficients to classify the glass sample data, obtaining a preliminary classification result. The sub-classification unit uses the K-means clustering algorithm to subdivide the preliminary classification result into subclasses, obtaining the subclass division result.

[0034] The glass sample data to be classified in the test set is input into the trained classification model to classify the glass type, and the output is the type to which the glass sample data belongs.

[0035] A computer-readable storage medium having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0036] Obtain the glass sample dataset and divide it into training and test sets;

[0037] The glass sample data in the training set is input into a pre-built classification model for training until the preset training termination condition is met, resulting in a trained classification model. The classification model includes a feature extraction unit, a discriminant unit, and a sub-classification unit connected in sequence. The feature extraction unit extracts chemical composition discrimination features from the glass sample data. The discriminant unit calculates the discrimination coefficients of the chemical composition discrimination features using the Fisher discriminant method and constructs a linear discriminant function based on these coefficients to classify the glass sample data, obtaining a preliminary classification result. The sub-classification unit uses the K-means clustering algorithm to subdivide the preliminary classification result into subclasses, obtaining the subclass division result.

[0038] The glass sample data to be classified in the test set is input into the trained classification model to classify the glass type, and the output is the type to which the glass sample data belongs.

[0039] The aforementioned glass classification method, apparatus, computer equipment, and storage medium based on a classification model automatically extract chemical composition discrimination features from glass sample data by constructing a machine learning-based classification model. A linear discriminant function is then constructed based on these chemical composition discrimination features using the discriminant units in the classification model. This linear discriminant function is used to perform preliminary classification of the glass sample data. Furthermore, the K-means clustering algorithm is employed to refine the preliminary classification results using the sub-classification units in the classification model, resulting in detailed sub-class divisions. This method improves the accuracy and reliability of glass classification by learning and training the classification model on large-scale data, extracting different chemical composition features from the glass sample data, and performing preliminary classification and sub-class division. Attached Figure Description

[0040] Figure 1 This is a flowchart illustrating a glass classification method based on a classification model in one embodiment;

[0041] Figure 2 This is a structural block diagram of a glass classification device based on a classification model in one embodiment;

[0042] Figure 3 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0044] In one embodiment, such as Figure 1 As shown, a glass classification method based on a classification model is provided, including the following steps:

[0045] Step S1: Obtain the glass sample dataset and divide the glass sample dataset into a training set and a test set.

[0046] It can be understood that the glass sample dataset is a dataset of labeled data. The glass sample dataset is divided into a training set and a test set. The data in the training set is used to train and optimize the parameters of the classification model, and the test set is used to test the feasibility of the classification model.

[0047] Step S2 involves inputting the glass sample data from the training set into a pre-built classification model for training until a preset training termination condition is met, resulting in a trained classification model. The classification model comprises a feature extraction unit, a discriminant unit, and a sub-classification unit connected sequentially. The feature extraction unit extracts chemical composition discrimination features from the glass sample data. The discriminant unit calculates the discrimination coefficients of the chemical composition discrimination features using the Fisher discriminant method and constructs a linear discriminant function based on these coefficients to classify the glass sample data, obtaining a preliminary classification result. The sub-classification unit uses the K-means clustering algorithm to subdivide the preliminary classification result into subclasses, obtaining the subclass division result.

[0048] Step S3: Input the glass sample data to be classified in the test set into the trained classification model to classify the glass type, and output the type to which the glass sample data belongs.

[0049] In one embodiment, the chemical composition discrimination features include the content of potassium oxide, lead oxide, barium oxide, strontium oxide, and silicon dioxide in the glass sample data.

[0050] Specifically, this application uses the rank-sum test to select the chemical composition discrimination features between high-potassium glass and lead-barium glass. The rank-sum test is a non-parametric test that can determine whether the distribution functions of two populations are consistent without needing to determine the distribution type of the population. Moreover, it is applicable to unpaired samples. The distribution types of the two data of the same chemical composition of high-potassium glass and lead-barium glass are unknown and are unpaired samples. It is suitable to use the rank-sum test to determine whether their distribution functions are consistent, and then find the chemical composition that can distinguish the two types of glass.

[0051] In one embodiment, the discrimination unit is further configured to calculate the discrimination coefficients of high-potassium glass and lead-barium glass based on the Fisher discrimination method, and to construct linear discrimination functions for high-potassium glass and lead-barium glass based on the high-potassium glass and lead-barium glass, respectively, expressed as follows:

[0052]

[0053]

[0054] Among them, F 高钾 Represents the linear discriminant function of high-potassium glass, F 铅钡 This represents the linear discriminant function for lead-barium glass. X PbO X BaO X SrO These represent the contents of potassium oxide, lead oxide, barium oxide, strontium oxide, and silicon dioxide, respectively, in the chemical composition discrimination characteristics. C PbO C BaO C SrO The discriminant coefficients for high-potassium glass, representing potassium oxide, lead oxide, barium oxide, strontium oxide, and silicon dioxide content, are respectively. W PbO W BaO W SrO The discrimination coefficients of lead-barium glass represent the contents of potassium oxide, lead oxide, barium oxide, strontium oxide, and silicon dioxide, and L1 and L2 represent the first and second set constants, respectively.

[0055] According to F 高钾 and F 铅钡 The glass sample data were classified to obtain preliminary classification results, where, when F 铅钡 <F 高钾 When F is used, the glass sample data is classified as high-potassium glass samples. 铅钡 ≥F 高钾 At that time, the glass sample data were classified as lead-barium glass samples.

[0056] Fisher's discriminant analysis is a linear discriminant method that projects all population sample points onto a one-dimensional space, ensuring that sample points from different populations are separated as much as possible within that space. Then, it calculates the distance between the projections of the glass sample data's observations onto this one-dimensional space and the projections of each population's mean onto the same one-dimensional space, classifying the glass sample data according to the closest population. Since glass sample data is high-dimensional, Fisher's discriminant analysis is suitable for determining whether it belongs to high-potassium glass or lead-barium glass.

[0057] In one embodiment, the subclassification unit is also used to perform low variance filtering on the chemical composition discrimination features of high potassium glass samples and lead-barium glass samples to obtain the subclass classification features of high potassium glass samples and lead-barium glass samples.

[0058] The K-means clustering algorithm was used to cluster the subclassing features of high-potassium glass samples and lead-barium glass samples, respectively, to obtain high-potassium glass subclass samples and lead-barium glass subclass samples.

[0059] In one embodiment, low-variance filtering is applied to the chemical composition discrimination features of high-potassium glass samples and lead-barium glass samples to obtain subclass classification features for high-potassium glass samples and lead-barium glass samples, including:

[0060] Low-variance filtering was applied to the chemical composition discrimination features of high-potassium glass samples to obtain the variance of the contents of different chemical components in the chemical composition discrimination features of high-potassium glass samples. The variances were then arranged in descending order, and the contents of the two chemical components with the highest variances were selected as the subclass classification features of high-potassium glass samples. Specifically, the subclass classification features of high-potassium glass samples were silica content and potassium oxide content.

[0061] Low-variance filtering was applied to the chemical composition discrimination features of lead-barium glass samples to obtain the variance of the content of different chemical components in the chemical composition discrimination features of lead-barium glass samples. The variances were then arranged in descending order, and the contents of the two chemical components with the highest variances were selected as the subclass classification features of lead-barium glass samples. Specifically, the subclass classification features of lead-barium glass samples were silicon dioxide content and lead oxide content.

[0062] Low variance filtering is a common data dimensionality reduction method that can filter out variables that carry a large amount of information. It has the advantages of being simple in principle and easy to implement. The principle of low variance filtering is to calculate the variance of each variable and remove variables with small variances. For high-dimensional datasets, variables with large variances represent that the values ​​of that variable vary greatly among different samples, and can be further divided based on these variables.

[0063] In one embodiment, the subclassing features of high-potassium glass samples and lead-barium glass samples are clustered according to the K-means clustering algorithm to obtain high-potassium glass subclass samples and lead-barium glass subclass samples, including:

[0064] The high-potassium glass samples were clustered based on K-means clustering characteristics, with a cluster size of 3, resulting in three subclasses of high-potassium glass samples. Specifically, these three subclasses include: high-silicon potassium-rich glass, low-silicon potassium-rich glass, and silicon-containing high-potassium glass. High-silicon potassium-rich glass corresponds to high-potassium glass with high silica content and low potassium oxide content; low-silicon potassium-rich glass corresponds to high-potassium glass with low silica content and high potassium oxide content; and silicon-containing high-potassium glass corresponds to high-potassium glass with neither high nor low silica content nor high nor low potassium oxide content.

[0065] The lead-barium glass samples were clustered based on K-means clustering to determine their subclass characteristics, with a set number of clusters of four, resulting in four subclasses of lead-barium glass samples. Specifically, these four subclasses are: high-silicon rare-lead-barium glass, low-silicon lead-rich-lead-barium glass, low-silicon lead-barium glass, and silicon-containing lead-barium glass. High-silicon rare-lead-barium glass corresponds to lead-barium glass with high silica content and low lead oxide content; low-silicon lead-barium glass corresponds to lead-barium glass with low silica content and moderate lead oxide content; and silicon-containing lead-barium glass corresponds to lead-barium glass with moderate silica and lead oxide content. These four categories of glass are classifications of lead-barium glass based on silica and lead oxide content.

[0066] K-means clustering is understood to cluster data by finding the centroids of each cluster as prototypes and grouping points adjacent to each centroid into different clusters. It is suitable for data with distinct inter-cluster characteristics. Since low-variance filtering has already been used to filter out sub-class features carrying significant information and exhibiting large differences between samples, the dataset used for clustering only contains these sub-class features, resulting in clear data characteristics suitable for K-means clustering and achieving good clustering results. Furthermore, when using K-means clustering to cluster the high-potassium glass dataset, the number of clusters needs to be manually set based on the silhouette coefficient—the K-means clustering performance evaluation coefficient. A larger silhouette coefficient indicates better clustering results. Experiments in this application revealed that when subclassing high-potassium glass samples, K-means clustering with 3 clusters yielded the largest silhouette coefficient; when subclassing lead-barium glass samples, K-means clustering with 4 clusters yielded the largest silhouette coefficient.

[0067] In one embodiment, the preset training termination condition includes at least one of the following:

[0068] The accuracy of the classification model has reached the preset accuracy threshold;

[0069] The number of times the training classification model is executed exceeds the preset loop count threshold;

[0070] The classification model converged.

[0071] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0072] In one embodiment, such as Figure 2 As shown, a glass classification device based on a classification model is provided, comprising:

[0073] Data acquisition module 201 is used to acquire glass sample dataset and divide the glass sample dataset into training set and test set;

[0074] Model training module 202,

[0075] This method is used to input glass sample data from the training set into a pre-built classification model for training until a preset training termination condition is met, resulting in a trained classification model. The classification model includes a feature extraction unit, a discriminant unit, and a sub-classification unit connected in sequence. The feature extraction unit extracts chemical composition discrimination features from the glass sample data. The discriminant unit calculates the discrimination coefficients of the chemical composition discrimination features using the Fisher discriminant method and constructs a linear discriminant function based on these coefficients to classify the glass sample data, obtaining a preliminary classification result. The sub-classification unit uses the K-means clustering algorithm to subdivide the preliminary classification result into subclasses, obtaining the subclass division result.

[0076] The glass classification module 203 is used to input the glass sample data to be classified in the test set into the trained classification model to classify the glass type, and output the type to which the glass sample data belongs.

[0077] Specific limitations regarding the classification model-based glass classification device can be found in the limitations of the classification model-based glass classification method described above, and will not be repeated here. Each module in the aforementioned classification model-based glass classification device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute them.

[0078] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 3As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a glass classification method based on a classification model. The display screen can be an LCD screen or an e-ink display screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0079] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0080] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to perform the following steps:

[0081] Obtain the glass sample dataset and divide it into training and test sets;

[0082] The glass sample data in the training set is input into a pre-built classification model for training until the preset training termination condition is met, resulting in a trained classification model. The classification model includes a feature extraction unit, a discriminant unit, and a sub-classification unit connected in sequence. The feature extraction unit extracts chemical composition discrimination features from the glass sample data. The discriminant unit calculates the discrimination coefficients of the chemical composition discrimination features using the Fisher discriminant method and constructs a linear discriminant function based on these coefficients to classify the glass sample data, obtaining a preliminary classification result. The sub-classification unit uses the K-means clustering algorithm to subdivide the preliminary classification result into subclasses, obtaining the subclass division result.

[0083] The glass sample data to be classified in the test set is input into the trained classification model to classify the glass type, and the output is the type to which the glass sample data belongs.

[0084] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0085] Obtain the glass sample dataset and divide it into training and test sets;

[0086] The glass sample data in the training set is input into a pre-built classification model for training until the preset training termination condition is met, resulting in a trained classification model. The classification model includes a feature extraction unit, a discriminant unit, and a sub-classification unit connected in sequence. The feature extraction unit extracts chemical composition discrimination features from the glass sample data. The discriminant unit calculates the discrimination coefficients of the chemical composition discrimination features using the Fisher discriminant method and constructs a linear discriminant function based on these coefficients to classify the glass sample data, obtaining a preliminary classification result. The sub-classification unit uses the K-means clustering algorithm to subdivide the preliminary classification result into subclasses, obtaining the subclass division result.

[0087] The glass sample data to be classified in the test set is input into the trained classification model to classify the glass type, and the output is the type to which the glass sample data belongs.

[0088] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0089] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0090] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A glass classification method based on a classification model, characterized by, The method comprises: obtaining a glass sample data set and dividing the glass sample data set into a training set and a test set; inputting the glass sample data in the training set into a pre-constructed classification model for training until a preset training end condition is reached, to obtain a trained classification model; wherein the classification model comprises a feature extraction unit, a discrimination unit and a sub-classification unit connected in sequence, the feature extraction unit is used to extract chemical composition discrimination features of the glass sample data; the discrimination unit is used to calculate discrimination coefficients of the chemical composition discrimination features according to a Fisher discrimination method, and construct a linear discrimination function according to the discrimination coefficients to classify the glass sample data, to obtain a preliminary classification result; the sub-classification unit is used to perform sub-class division on the preliminary classification result according to a K-means clustering algorithm, to obtain a sub-class division result; inputting the to-be-classified glass sample data in the test set into the trained classification model for glass type classification, to output a type to which the to-be-classified glass sample data belongs; the discrimination unit is further used to calculate high-potassium glass discrimination coefficients and lead-barium glass discrimination coefficients of the chemical composition discrimination features according to the Fisher discrimination method respectively, and construct high-potassium glass linear discrimination functions and lead-barium glass linear discrimination functions according to the high-potassium glass discrimination coefficients and the lead-barium glass discrimination coefficients, respectively, wherein, represents a high potassium glass linear discriminant function, represents a lead barium glass linear discriminant function, respectively represent the potassium oxide content, the lead oxide content, the barium oxide content, the strontium oxide content, and the silicon dioxide content in the chemical composition discriminant feature, respectively represent high potassium glass discriminant coefficients of the potassium oxide content, the lead oxide content, the barium oxide content, the strontium oxide content, and the silicon dioxide content, represent lead barium glass discriminant coefficients of the potassium oxide content, the lead oxide content, the barium oxide content, the strontium oxide content, and the silicon dioxide content, and respectively represent a first setting constant and a second setting constant; According to the method and Classifying the glass sample data to obtain a preliminary classification result, wherein when the glass sample data is classified as high potassium glass sample, and when the glass sample data is classified as lead-barium glass sample.

2. The method of claim 1, wherein, the chemical composition discrimination features include potassium oxide content, lead oxide content, barium oxide content, strontium oxide content and silicon dioxide content contained in the glass sample data.

3. The method of claim 1, wherein, The sub-classification unit is further used to perform low-variance filtering on the chemical composition discrimination features of the high-potassium glass sample and the lead-barium glass sample, to obtain sub-class division features of the high-potassium glass sample and sub-class division features of the lead-barium glass sample; the sub-class division features of the high-potassium glass sample and the sub-class division features of the lead-barium glass sample are clustered according to the K-means clustering algorithm respectively, to obtain high-potassium glass sub-class samples and lead-barium glass sub-class samples.

4. The method of claim 3, wherein, The sub-class division features of the high-potassium glass sample and the sub-class division features of the lead-barium glass sample are obtained by performing low-variance filtering on the chemical composition discrimination features of the high-potassium glass sample and the lead-barium glass sample, comprising: the variances of different chemical composition contents in the chemical composition discrimination features of the high-potassium glass sample are obtained by performing low-variance filtering on the chemical composition discrimination features of the high-potassium glass sample, and are arranged in descending order, and the chemical composition contents with the top two variances are selected as the sub-class division features of the high-potassium glass sample; the variances of different chemical composition contents in the chemical composition discrimination features of the lead-barium glass sample are obtained by performing low-variance filtering on the chemical composition discrimination features of the lead-barium glass sample, and are arranged in descending order, and the chemical composition contents with the top two variances are selected as the sub-class division features of the lead-barium glass sample.

5. The method of claim 3, wherein, the sub-class division features of the high-potassium glass sample and the sub-class division features of the lead-barium glass sample are clustered according to the K-means clustering algorithm respectively, to obtain high-potassium glass sub-class samples and lead-barium glass sub-class samples, comprising: According to K-means clustering, the sub-class division features of the high-potassium glass samples are clustered, and the number of clusters is set to 3, to obtain three types of high-potassium glass sub-class samples; According to K-means clustering, the sub-class division features of the lead-barium glass samples are clustered, and the number of clusters is set to 4, to obtain four types of lead-barium glass sub-class samples.

6. The method of claim 1, wherein, The preset training end condition includes at least one of the following: The accuracy of the classification model reaches a preset accuracy threshold; The number of times of cyclically training the classification model exceeds a preset cycle number threshold; The classification effect of the classification model converges.

7. A glass classification apparatus based on a classification model, characterized by, The device comprises: A data acquisition module configured to acquire a glass sample data set and divide the glass sample data set into a training set and a test set; A model training module configured to input glass sample data in the training set into a pre-constructed classification model for training until a preset training end condition is reached, to obtain a trained classification model; wherein the classification model comprises a feature extraction unit, a discrimination unit and a sub-classification unit connected in sequence, the feature extraction unit is configured to extract chemical composition discrimination features of the glass sample data; the discrimination unit is configured to calculate discrimination coefficients of the chemical composition discrimination features according to a Fisher discrimination method, and construct a linear discrimination function according to the discrimination coefficients to classify the glass sample data, to obtain a preliminary classification result; the sub-classification unit is configured to perform sub-class division on the preliminary classification result according to a K-means clustering algorithm, to obtain a sub-class division result; A glass classification module configured to input glass sample data to be classified in the test set into the trained classification model for glass type classification, to output a type to which the glass sample data to be classified belongs; The discrimination unit is further configured to calculate high-potassium glass discrimination coefficients and lead-barium glass discrimination coefficients of the chemical composition discrimination features according to the Fisher discrimination method, respectively, and construct high-potassium glass linear discrimination functions and lead-barium glass linear discrimination functions according to the high-potassium glass discrimination coefficients and the lead-barium glass discrimination coefficients, respectively, represented as wherein, represents a high potassium glass linear discriminant function, represents a lead barium glass linear discriminant function, respectively represent a potassium oxide content, a lead oxide content, a barium oxide content, a strontium oxide content, and a silicon dioxide content in the chemical composition discriminant feature, respectively represent high potassium glass discriminant coefficients of the potassium oxide content, the lead oxide content, the barium oxide content, the strontium oxide content, and the silicon dioxide content, represent lead barium glass discriminant coefficients of the potassium oxide content, the lead oxide content, the barium oxide content, the strontium oxide content, and the silicon dioxide content, and respectively represent a first setting constant and a second setting constant; According to the method and classifying the glass sample data to obtain a preliminary classification result, wherein when the glass sample data is classified as high potassium glass sample, and when the glass sample data is classified as lead-barium glass sample.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 6.

Images