Coal quality evaluation method and system based on spectrum and image feature fusion
Through the coal quality evaluation method of spectral and image features fusion, deep learning and support vector machine model are used to solve the problem that the existing technology cannot integrate the physical and image features of coal species, and the precise evaluation and classification support of coal powder data are achieved.
Patent Information
- Application Number
- CN202411083046.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-11-22
AI Technical Summary
The prior art cannot perform fusion analysis and evaluation based on the physical characteristics and image characteristics of coal species, which affects the accurate feature evaluation representation of coal powder data.
Using the coal quality evaluation method based on the fusion of spectral and image features, the coal powder feature data is preprocessed, and a self-attention mechanism model with deep learning combined with support vector machine is constructed to extract and classify the physical characteristics of coal species. Fusion analysis and image features.
It improves the accuracy of coal powder evaluation, provides classification support for the utilization of coal powder, ensures the accuracy and consistency of data, is suitable for processing image spectral data, and realizes accurate analysis and classification of coal quality.
Smart Images

Figure CN119000670A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of coal quality evaluation, and in particular relates to a coal quality evaluation method and system based on fusion of spectrum and image features. Background Art
[0002] Coal quality evaluation refers to the analysis and evaluation of various indicators of coking coal, including physical and chemical indicators such as ash, volatile matter, sulfur, moisture, etc. In the process of coal production and use, coal quality evaluation needs to comprehensively consider these indicators to determine the final use and value of coal.
[0003] Chinese patent CN110739029B discloses a coal quality evaluation method and device, the method comprising: obtaining identity information of the type of coal to be evaluated, the identity information comprising: coal type name, mining time, mining site; detecting whether there is target identity information consistent with the identity information in the database; if there is no target identity information consistent with the identity information, obtaining industrial analysis index data of the type of coal to be evaluated; inputting the industrial analysis index data into the evaluation formula corresponding to the type of coal name to obtain an evaluation value; generating evaluation information of the type of coal to be evaluated according to the evaluation value; however, the existing method cannot perform fusion analysis and evaluation based on the physical characteristics and image characteristics of the coal type, thereby affecting the accurate feature evaluation representation of the coal powder data. To address the above problems, we propose a coal quality evaluation method and system based on the fusion of spectral and image features. Summary of the invention
[0004] The purpose of the present invention is to address the deficiencies of the prior art and provide a coal quality evaluation method and system based on the fusion of spectral and image features, thereby solving the problem that the prior methods are unable to perform fusion analysis and evaluation based on the physical characteristics and image characteristics of the coal type, thereby affecting the accurate feature evaluation representation of coal powder data.
[0005] Existing methods cannot perform fusion analysis and evaluation based on the physical characteristics and image characteristics of coal types, which affects the accurate feature evaluation and representation of coal powder data. In response to the above problems, we propose a coal quality evaluation method and system based on the fusion of spectral and image features. In short, the method includes preprocessing the coal powder feature data to obtain a preprocessing set, obtaining coal powder sample data, and dividing the coal powder sample data into a training set and a test set, pre-constructing a coal quality evaluation model based on the fusion of spectral and image features, training the coal quality evaluation model with the training set, and testing the accuracy of the coal quality evaluation model with the test set, outputting the trained coal quality evaluation model, and finally the coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs the coal quality evaluation result. In an embodiment of the present invention, by pre-constructing a coal quality evaluation model based on the fusion of spectral and image features, training the coal quality evaluation model with the training set, and testing the accuracy of the coal quality evaluation model with the test set, the fusion analysis and evaluation of the physical characteristics and image characteristics of coal types is realized, the accuracy of coal powder evaluation is guaranteed, and classification support is provided for coal powder utilization.
[0006] The present invention is implemented in this way: a coal quality evaluation method based on spectral and image feature fusion, the coal quality evaluation method based on spectral and image feature fusion comprises:
[0007] Collecting coal powder characteristic data, preprocessing the coal powder characteristic data, and obtaining a preprocessing set;
[0008] Acquire pulverized coal sample data, divide the pulverized coal sample data into a training set and a test set, and preprocess the training set and the test set respectively;
[0009] Pre-constructing a coal quality evaluation model based on spectral and image feature fusion, training the coal quality evaluation model using a training set, testing the accuracy of the coal quality evaluation model using a test set, and outputting the trained coal quality evaluation model;
[0010] A preprocessing set is obtained, and the coal quality evaluation model is executed with the preprocessing set as input. The coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs a coal quality evaluation result.
[0011] Preferably, the method for preprocessing the coal powder characteristic data specifically includes:
[0012] Acquire coal powder characteristic data, perform data cleaning and missing value processing on the coal powder characteristic data, and obtain a cleaned data set;
[0013] Loading the cleaned data set, and using the softmax function to standardize the cleaned data set to obtain a standard data set;
[0014] Obtain a standard data set, process the data values in the standard data set using Min-max standardization and Z-score standardization respectively, calculate the covariance matrix of the standard data set based on the Min-max standardization results and the Z-score standardization results, perform eigendecomposition on the covariance matrix to obtain a eigendecomposition set, output the eigendecomposition set and the standard data set, and integrate the eigendecomposition set and the standard data set to obtain a preprocessing set.
[0015] Preferably, the method for pre-constructing a coal quality evaluation model based on spectral and image feature fusion specifically comprises:
[0016] A self-attention mechanism model is constructed based on deep learning combined with support vector machine, and the self-attention mechanism model is used as the initial model of the coal quality evaluation model, wherein the self-attention mechanism model includes an input layer, a feature extraction layer, a feature fusion layer, a classification recognition network and an output layer;
[0017] Build a classification recognition network, classify and label the acquired training set, and define the network architecture of the classifier and discriminator of the classification recognition network;
[0018] The performance of the classification and recognition network is determined based on the chaotic attractor method. After the classification and recognition network is determined to have converged, the construction of the classification and recognition network is completed.
[0019] Select the cross entropy loss function as the loss function of the coal quality evaluation model, define the optimizer of the coal quality evaluation model, and fix the model hyperparameters except the learning rate;
[0020] Among them, the loss function expression of the coal quality evaluation model is as follows:
[0021]
[0022] Among them, x i ,y i They represent the probability distribution of the true value and predicted value of the sample, n is the number of samples, and α is the gradient coefficient of the loss function.
[0023] Preferably, the method of training the coal quality evaluation model using a training set and testing the accuracy of the coal quality evaluation model using a test set specifically includes:
[0024] Obtain a training set, use the training set to perform multiple rounds of iterative training on the coal quality evaluation model, obtain an improved coal quality evaluation model, and perform ablation experiments on the input layer, feature extraction layer, feature fusion layer, classification recognition network, and output layer of the coal quality evaluation model based on preset hyperparameters;
[0025] The training set is enhanced based on the Mosaic-Mixup image enhancement method to obtain the training enhancement set. The coal quality evaluation model is trained again for multiple rounds of iterations until convergence using the training enhancement set as input.
[0026] Preset the prediction threshold to determine whether the trained coal quality evaluation model meets the prediction threshold. If it meets the prediction threshold, output the coal quality evaluation model;
[0027] If the prediction threshold is not met, the model parameters are adjusted based on the MSE mean square error, and the training set is enhanced based on the Mosaic-Mixup image enhancement method to obtain the training enhancement set. The coal quality evaluation model is trained again for multiple rounds of iterations until convergence using the training enhancement set as input.
[0028] Preferably, the method of training the coal quality evaluation model with a training set and testing the accuracy of the coal quality evaluation model with a test set further includes:
[0029] Obtain the test set, use the MSE mean square error to calculate the accuracy error between the model's prediction results and the actual results, and determine whether the accuracy error of the coal quality evaluation model meets the preset accuracy;
[0030] If the preset accuracy is met, the trained coal quality evaluation model is verified and the trained coal quality evaluation model is output;
[0031] If the preset accuracy is not met, the optimizer Adam is used to adjust the model parameters.
[0032] Preferably, the classification recognition network consists of 5 stages, each stage includes a convolution group consisting of n skip connections and a 3*3 convolution layer, and the function of the 3*3 convolution layer is to change the dimension of the output feature map.
[0033] Preferably, the coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs the coal quality evaluation result, which specifically includes:
[0034] Get the preprocessed set and input the preprocessed set into the input layer;
[0035] The feature extraction layer extracts image features and spectral features from the feature data of the input layer and performs dimension reduction and normalization processing;
[0036] The image features and spectral features given by the feature fusion layer are used for feature fusion, and invalid feature information is filtered to obtain a fusion result set of standard image features and spectral features;
[0037] Load the classification recognition network, which converts all the retained fusion result set information into a standard classification recognition vector;
[0038] The classification recognition vector is captured through the output layer to obtain the coal quality evaluation result.
[0039] The present invention also provides a coal quality evaluation system based on spectral and image feature fusion, and the coal quality evaluation system based on spectral and image feature fusion specifically includes:
[0040] A data preprocessing module is used to collect coal powder characteristic data, preprocess the coal powder characteristic data, and obtain a preprocessing set;
[0041] The sample processing module is used to obtain the coal powder sample data, divide the coal powder sample data into a training set and a test set, and pre-process the training set and the test set respectively;
[0042] A model building module, which is based on a coal quality evaluation model that is fused with spectral and image features, uses a training set to train the coal quality evaluation model, and uses a test set to test the accuracy of the coal quality evaluation model, and outputs the trained coal quality evaluation model;
[0043] The result output module is used to obtain a preprocessing set, take the preprocessing set as input, execute the coal quality evaluation model, and the coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs the coal quality evaluation result.
[0044] Preferably, the data preprocessing module specifically includes:
[0045] A data cleaning unit is used to obtain coal powder characteristic data, perform data cleaning and missing value processing on the coal powder characteristic data, and obtain a cleaned data set;
[0046] A standardization unit, used for loading the cleaned data set, and using a softmax function to standardize the cleaned data set to obtain a standard data set;
[0047] The data integration unit is used to obtain a standard data set, process the data values in the standard data set using Min-max standardization and Z-score standardization respectively, calculate the covariance matrix of the standard data set based on the Min-max standardization results and the Z-score standardization results, perform eigendecomposition on the covariance matrix to obtain a eigendecomposition set, output the eigendecomposition set and the standard data set, and integrate the eigendecomposition set and the standard data set to obtain a preprocessing set.
[0048] Preferably, the result output module specifically includes:
[0049] The feature extraction unit extracts image features and spectral features of the feature data of the input layer based on the feature extraction layer and performs dimension reduction and normalization processing;
[0050] The feature fusion unit uses the image features and spectral features given by the feature fusion layer to perform feature fusion and filter out invalid feature information to obtain a fusion result set of standard image features and spectral features;
[0051] A classification recognition unit is used to load a classification recognition network, which converts all the retained fusion result set information into a standard classification recognition vector;
[0052] The evaluation generation unit captures the classification recognition vector through the output layer to obtain the coal quality evaluation result.
[0053] Compared with the prior art, the embodiments of the present application have the following beneficial effects:
[0054] In an embodiment of the present invention, a coal quality evaluation model based on the fusion of spectral and image features is pre-constructed, the coal quality evaluation model is trained using a training set, and the accuracy of the coal quality evaluation model is tested using a test set, thereby achieving fusion analysis and evaluation of the physical characteristics and image characteristics of coal types, ensuring the accuracy of coal powder evaluation, and providing classification support for coal powder utilization.
[0055] In the embodiment of the present invention, the coal powder characteristic data is preprocessed to ensure the accuracy, completeness and consistency of the data, thereby improving the accuracy of data analysis and modeling, and facilitating the construction of a coal quality evaluation model and the classification and evaluation of data.
[0056] In an embodiment of the present invention, a coal quality evaluation model and a training method thereof are provided. The coal quality evaluation model consists of an input layer, a feature extraction layer, a feature fusion layer, a classification and recognition network, and an output layer. The coal quality evaluation model is constructed based on deep learning combined with a support vector machine, thereby achieving accurate analysis and evaluation of coal quality. The model can automatically extract multi-source heterogeneous features from data, is very suitable for processing image spectral data, and effectively fuses image spectral data to achieve accurate classification and identification of coal quality. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 It is a schematic diagram of the implementation process of the coal quality evaluation method based on spectral and image feature fusion provided by the present invention.
[0058] Figure 2 A schematic diagram of the implementation flow of the method for preprocessing coal powder characteristic data is shown.
[0059] Figure 3 A schematic diagram of the implementation process of the method for pre-constructing a coal quality evaluation model based on spectral and image feature fusion is shown.
[0060] Figure 4 It is a schematic diagram of the classification recognition network structure provided by the present invention.
[0061] Figure 5 It is a schematic diagram of the stage structure composition in the classification recognition network provided by the present invention.
[0062] Figure 6 The figure shows a schematic diagram of a process for training a coal quality evaluation model using a training set and testing the accuracy of the coal quality evaluation model using a test set.
[0063] Figure 7 It is a schematic diagram of the architecture of the coal quality evaluation model provided by the present invention.
[0064] Figure 8 The figure shows a schematic diagram of the implementation process of the coal quality evaluation model performing feature fusion analysis on the preprocessing set and outputting the coal quality evaluation results.
[0065] Fig. 9 A schematic diagram showing the process of feature fusion in the feature fusion layer.
[0066] Fig.10 It is a structural schematic diagram of a coal quality evaluation system based on spectral and image feature fusion provided by the present invention. DETAILED DESCRIPTION
[0067] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of this application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.
[0068] Existing methods cannot perform fusion analysis and evaluation based on the physical characteristics and image characteristics of coal types, which affects the accurate feature evaluation and representation of coal powder data. In response to the above problems, we propose a coal quality evaluation method and system based on the fusion of spectral and image features. In short, the method includes preprocessing the coal powder feature data to obtain a preprocessing set, obtaining coal powder sample data, and dividing the coal powder sample data into a training set and a test set, pre-constructing a coal quality evaluation model based on the fusion of spectral and image features, training the coal quality evaluation model with the training set, and testing the accuracy of the coal quality evaluation model with the test set, outputting the trained coal quality evaluation model, and finally the coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs the coal quality evaluation result. In an embodiment of the present invention, by pre-constructing a coal quality evaluation model based on the fusion of spectral and image features, training the coal quality evaluation model with the training set, and testing the accuracy of the coal quality evaluation model with the test set, the fusion analysis and evaluation of the physical characteristics and image characteristics of coal types is realized, the accuracy of coal powder evaluation is guaranteed, and classification support is provided for coal powder utilization.
[0069] The embodiment of the present invention provides a coal quality evaluation method based on spectral and image feature fusion. Figure 1 The figure shows a schematic diagram of the implementation process of the coal quality evaluation method based on the fusion of spectrum and image features. The coal quality evaluation method based on the fusion of spectrum and image features specifically includes:
[0070] Step S10, collecting coal powder characteristic data, preprocessing the coal powder characteristic data, and obtaining a preprocessing set;
[0071] It should be noted that the coal powder characteristic data includes but is not limited to coal powder moisture, ash content, volatile matter, fixed carbon content, calorific value, sulfur content, coal ash fusion temperature data, and also includes coal powder spectral characteristics and appearance image data.
[0072] Step S20, obtaining pulverized coal sample data, dividing the pulverized coal sample data into a training set and a test set, and preprocessing the training set and the test set respectively;
[0073] In this embodiment, the ratio of the training set to the test set may be 3:1 or 7:3.
[0074] Step S30, pre-constructing a coal quality evaluation model based on spectral and image feature fusion, training the coal quality evaluation model using a training set, testing the accuracy of the coal quality evaluation model using a test set, and outputting the trained coal quality evaluation model;
[0075] Step S40, obtaining a preprocessing set, taking the preprocessing set as input, executing the coal quality evaluation model, the coal quality evaluation model performs feature fusion analysis on the preprocessing set, and outputs a coal quality evaluation result.
[0076] In an embodiment of the present invention, a coal quality evaluation model based on the fusion of spectral and image features is pre-constructed, the coal quality evaluation model is trained using a training set, and the accuracy of the coal quality evaluation model is tested using a test set, thereby achieving fusion analysis and evaluation of the physical characteristics and image characteristics of coal types, ensuring the accuracy of coal powder evaluation, and providing classification support for coal powder utilization.
[0077] The embodiment of the present invention provides a method for preprocessing coal powder characteristic data. Figure 2 The schematic diagram of the implementation process of the method for preprocessing the coal powder characteristic data is shown. The method for preprocessing the coal powder characteristic data specifically includes:
[0078] Step S101, obtaining coal powder characteristic data, performing data cleaning and missing value processing on the coal powder characteristic data, and obtaining a cleaned data set;
[0079] It should be noted that data cleaning of coal powder characteristic data includes but is not limited to missing value processing, outlier processing, format and data type conversion, so as to ensure that multi-source heterogeneous data can be used for model construction and training, and the methods of missing value processing include but are not limited to deletion, interpolation, using default values, extrapolation and anonymization.
[0080] Step S102, loading the cleaned data set, and using a softmax function to standardize the cleaned data set to obtain a standard data set;
[0081] Step S103, obtain a standard data set, and process the data values in the standard data set using Min-max standardization and Z-score standardization respectively, calculate the covariance matrix of the standard data set based on the Min-max standardization results and the Z-score standardization results, perform eigendecomposition on the covariance matrix to obtain a eigendecomposition set, output the eigendecomposition set and the standard data set, and integrate the eigendecomposition set and the standard data set to obtain a preprocessing set.
[0082] In the embodiment of the present invention, the coal powder characteristic data is preprocessed to ensure the accuracy, completeness and consistency of the data, thereby improving the accuracy of data analysis and modeling, and facilitating the construction of a coal quality evaluation model and the classification and evaluation of data.
[0083] The embodiment of the present invention provides a method for pre-constructing a coal quality evaluation model based on spectral and image feature fusion. Figure 3 The present invention shows a schematic diagram of a method for pre-constructing a coal quality evaluation model based on spectral and image feature fusion, wherein the method for pre-constructing a coal quality evaluation model based on spectral and image feature fusion specifically includes:
[0084] Step S201, constructing a self-attention mechanism model based on deep learning combined with support vector machine, and using the self-attention mechanism model as the initial model of the coal quality evaluation model, wherein the self-attention mechanism model includes an input layer, a feature extraction layer, a feature fusion layer, a classification recognition network and an output layer;
[0085] Step S202, constructing a classification recognition network, performing feature classification marking on the acquired training set, and defining the network architecture of the classifier and discriminator of the classification recognition network;
[0086] It should be noted that the classification recognition network classifies the foreground and background by adding a classifier to the feature extraction subnetwork to output a 2D feature vector. When the network reaches convergence through training, the classifier is removed. At this time, the data feature distribution will be implicitly learned in the parameters of the classification recognition network. The classification recognition network structure is as follows Figure 4 shown.
[0087] Step S203, judging the performance of the classification recognition network based on the chaotic attractor method, and completing the construction of the classification recognition network after judging that the classification recognition network has converged;
[0088] Step S204, selecting a cross entropy loss function as the loss function of the coal quality evaluation model, defining an optimizer of the coal quality evaluation model, and fixing model hyperparameters except the learning rate;
[0089] Among them, the loss function expression of the coal quality evaluation model is as follows:
[0090]
[0091] Among them, x i ,y i They respectively represent the probability distribution of the true value and the predicted value of the sample, n is the number of samples, α is the gradient coefficient of the loss function, and in the embodiment of the present invention, the gradient coefficient can be 0.2-0.25.
[0092] In this embodiment, the classification recognition network is composed of 5 stages, each of which includes a convolution group consisting of n skip connections and a 3*3 convolution layer. The function of the 3*3 convolution layer is to change the dimension of the output feature map. Each stage of the classification recognition network is composed of the following structure: Figure 5 , mainly consists of a convolution group of n skip connections and a 3×3 convolution layer. The main function of the skip connection is to extract features and suppress overfitting of the algorithm.
[0093] The embodiment of the present invention provides a method for training a coal quality evaluation model using a training set and testing the accuracy of the coal quality evaluation model using a test set. Figure 6 The figure shows a schematic flow chart of using a training set to train a coal quality evaluation model and using a test set to test the accuracy of the coal quality evaluation model. The method of using a training set to train a coal quality evaluation model and using a test set to test the accuracy of the coal quality evaluation model specifically includes:
[0094] Step S301, obtaining a training set, using the training set to perform multiple rounds of iterative training on the coal quality evaluation model to obtain an improved coal quality evaluation model, and performing ablation experiments on the input layer, feature extraction layer, feature fusion layer, classification recognition network, and output layer of the coal quality evaluation model based on preset hyperparameters;
[0095] Step S302, enhancing the training set based on the Mosaic-Mixup image enhancement method to obtain a training enhancement set, using the training enhancement set as input, and performing multiple rounds of iterative training on the coal quality evaluation model again until convergence;
[0096] Step S303, preset a prediction threshold, and determine whether the trained coal quality evaluation model meets the prediction threshold; in this embodiment, the prediction threshold can be set to 0.3-0.35.
[0097] Step S304: if the prediction threshold is met, output the coal quality evaluation model;
[0098] If the prediction threshold is not met, the model parameters are adjusted based on the MSE mean square error, and the training set is enhanced based on the Mosaic-Mixup image enhancement method to obtain the training enhancement set. The coal quality evaluation model is trained again for multiple rounds of iterations until convergence using the training enhancement set as input.
[0099] Step S305, obtaining a test set, using the MSE mean square error to calculate the accuracy error between the prediction result of the model and the actual result, and judging whether the accuracy error of the coal quality evaluation model meets the preset accuracy;
[0100] Step S306: If the preset accuracy is met, the trained coal quality evaluation model is verified and the trained coal quality evaluation model is output. Figure 7 As shown, a schematic diagram of the architecture of the coal quality evaluation model is shown.
[0101] If the preset accuracy is not met, the optimizer Adam is used to adjust the model parameters.
[0102] In an embodiment of the present invention, a coal quality evaluation model and a training method thereof are provided. The coal quality evaluation model consists of an input layer, a feature extraction layer, a feature fusion layer, a classification and recognition network, and an output layer. The coal quality evaluation model is constructed based on deep learning combined with a support vector machine, thereby achieving accurate analysis and evaluation of coal quality. The model can automatically extract multi-source heterogeneous features from data, is very suitable for processing image spectral data, and effectively fuses image spectral data to achieve accurate classification and identification of coal quality.
[0103] The embodiment of the present invention provides a coal quality evaluation model to perform feature fusion analysis on the preprocessing set and output a coal quality evaluation result. Figure 8 The figure shows a schematic diagram of the implementation process of the coal quality evaluation model performing feature fusion analysis on the preprocessing set and outputting the coal quality evaluation result. The coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs the coal quality evaluation result, which specifically includes:
[0104] Step S401, obtaining a preprocessing set, and inputting the preprocessing set into an input layer;
[0105] Step S402, the feature extraction layer extracts image features and spectral features from the feature data of the input layer and performs dimension reduction and normalization processing;
[0106] Step S403, performing feature fusion using the image features and spectral features given by the feature fusion layer, and filtering invalid feature information to obtain a fusion result set of standard image features and spectral features;
[0107] like Fig. 9 As shown in the figure, a schematic diagram of the feature fusion process of the feature fusion layer is shown, where H×W×C are the height, width and depth of the output feature maps of the feature extraction network stage3, stage4 and stage5 respectively. The fusion process keeps the feature map size unchanged. The rounded rectangle in the figure represents the convolution group used, and the rectangle represents the input and output feature map size.
[0108] Step S404, loading the classification recognition network, the classification recognition network converts all the retained fusion result set information into a standard classification recognition vector;
[0109] Step S405, capturing the classification recognition vector through the output layer to obtain the coal quality evaluation result.
[0110] The embodiment of the present invention provides a coal quality evaluation system based on spectral and image feature fusion. Fig.10 The structural schematic diagram of the coal quality evaluation system based on spectral and image feature fusion is shown. The coal quality evaluation system based on spectral and image feature fusion specifically includes:
[0111] The data preprocessing module 100 is used to collect coal powder characteristic data, preprocess the coal powder characteristic data, and obtain a preprocessing set;
[0112] The sample processing module 200 is used to obtain the pulverized coal sample data, divide the pulverized coal sample data into a training set and a test set, and pre-process the training set and the test set respectively;
[0113] A model building module 300, which is based on a coal quality evaluation model fused with spectral and image features, uses a training set to train the coal quality evaluation model, and uses a test set to test the accuracy of the coal quality evaluation model, and outputs the trained coal quality evaluation model;
[0114] The result output module 400 is used to obtain a preprocessing set, take the preprocessing set as input, execute the coal quality evaluation model, and the coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs the coal quality evaluation result.
[0115] It should be noted that the data preprocessing module 100, the sample processing module 200, the model building module 300, and the result output module 400 can be connected by 5G communication or DTU communication.
[0116] In this embodiment, the data preprocessing module 100 specifically includes:
[0117] A data cleaning unit 110 is used to obtain pulverized coal characteristic data, perform data cleaning and missing value processing on the pulverized coal characteristic data, and obtain a cleaned data set;
[0118] A standardization unit 120 is used to load the cleaned data set and use a softmax function to standardize the cleaned data set to obtain a standard data set;
[0119] The data integration unit 130 is used to obtain a standard data set, process the data values in the standard data set using Min-max standardization and Z-score standardization, calculate the covariance matrix of the standard data set based on the Min-max standardization results and the Z-score standardization results, perform eigendecomposition on the covariance matrix to obtain a eigendecomposition set, output the eigendecomposition set and the standard data set, and integrate the eigendecomposition set and the standard data set to obtain a preprocessing set.
[0120] In this embodiment, the result output module 400 specifically includes:
[0121] The feature extraction unit 410 extracts image features and spectral features of the feature data of the input layer based on the feature extraction layer and performs dimension reduction and normalization processing;
[0122] The feature fusion unit 420 performs feature fusion using the image features and spectral features given by the feature fusion layer, and filters invalid feature information to obtain a fusion result set of standard image features and spectral features;
[0123] A classification recognition unit 430 is used to load a classification recognition network, which converts all the retained fusion result set information into a standard classification recognition vector;
[0124] The evaluation generation unit 440 captures the classification recognition vector through the output layer to obtain the coal quality evaluation result.
[0125] Another aspect of the embodiments of the present invention further provides a computer-readable storage medium, which stores computer program instructions, and the computer program instructions can be executed by a processor. When the computer program instructions are executed, the method of any one of the above embodiments is implemented.
[0126] In another aspect of the embodiments of the present invention, a computer device is provided. The computer device includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the method of any one of the above embodiments is implemented.
[0127] It should be noted that the memory, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as the program instructions / modules corresponding to the coal quality evaluation method based on spectral and image feature fusion in the embodiment of the present application. The memory may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the use of the coal quality evaluation method based on spectral and image feature fusion, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the local module via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0128] Finally, it should be noted that the computer-readable storage medium (e.g., memory) herein may be a volatile memory or a nonvolatile memory, or may include both a volatile memory and a nonvolatile memory. As an example and not by way of limitation, a nonvolatile memory may include a read-only memory (ROM), a programmable ROM (PROM), an electrically programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), or a flash memory. A volatile memory may include a random access memory (RAM), which may act as an external cache memory. As an example and not by way of limitation, RAM may be obtained in a variety of forms, such as synchronous RAM (DRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The storage devices of the disclosed aspects are intended to include, but are not limited to, these and other suitable types of memory.
[0129] The various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure herein may be implemented or executed using the following components designed to perform the functions herein: a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination of these components. A general purpose processor may be a microprocessor, but alternatively, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP, and / or any other such configuration.
[0130] In summary, the present invention provides a coal quality evaluation method and system based on the fusion of spectral and image features. In an embodiment of the present invention, a coal quality evaluation model based on the fusion of spectral and image features is pre-constructed, the coal quality evaluation model is trained with a training set, and the accuracy of the coal quality evaluation model is tested with a test set, thereby realizing the fusion analysis and evaluation of the physical characteristics and image characteristics of the coal types, ensuring the accuracy of coal powder evaluation, and providing classification support for coal powder utilization.
[0131] It should be noted that, for the above-mentioned embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described order of actions, because according to the present invention, some steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.
[0132] In the several embodiments provided in this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic, and the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.
[0133] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the scope of protection of the invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on these embodiments, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in this field can still combine, add, delete or make other adjustments to the features in the various embodiments of the present invention according to the circumstances without conflict, without making creative work, so as to obtain different other technical solutions that do not deviate from the concept of the present invention in essence, and these technical solutions also belong to the scope of protection of the present invention.
Claims
1. A coal quality evaluation method based on spectral and image feature fusion, characterized in that: The coal quality evaluation method based on spectral and image feature fusion includes: Collecting coal powder characteristic data, preprocessing the coal powder characteristic data, and obtaining a preprocessing set; Acquire pulverized coal sample data, divide the pulverized coal sample data into a training set and a test set, and preprocess the training set and the test set respectively; Pre-constructing a coal quality evaluation model based on spectral and image feature fusion, training the coal quality evaluation model using a training set, testing the accuracy of the coal quality evaluation model using a test set, and outputting the trained coal quality evaluation model; A preprocessing set is obtained, and the coal quality evaluation model is executed with the preprocessing set as input. The coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs a coal quality evaluation result.
2. The coal quality evaluation method based on spectral and image feature fusion according to claim 1, characterized in that: The method for preprocessing coal powder characteristic data specifically includes: Acquire coal powder characteristic data, perform data cleaning and missing value processing on the coal powder characteristic data, and obtain a cleaned data set; Loading the cleaned data set, and using the softmax function to standardize the cleaned data set to obtain a standard data set; Obtain a standard data set, process the data values in the standard data set using Min-max standardization and Z-score standardization respectively, calculate the covariance matrix of the standard data set based on the Min-max standardization results and the Z-score standardization results, perform eigendecomposition on the covariance matrix to obtain a eigendecomposition set, output the eigendecomposition set and the standard data set, and integrate the eigendecomposition set and the standard data set to obtain a preprocessing set.
3. The coal quality evaluation method based on spectral and image feature fusion according to claim 2, characterized in that: The method for pre-constructing a coal quality evaluation model based on spectral and image feature fusion specifically includes: A self-attention mechanism model is constructed based on deep learning combined with support vector machine, and the self-attention mechanism model is used as the initial model of the coal quality evaluation model, wherein the self-attention mechanism model includes an input layer, a feature extraction layer, a feature fusion layer, a classification recognition network and an output layer; Build a classification recognition network, classify and label the acquired training set, and define the network architecture of the classifier and discriminator of the classification recognition network; The performance of the classification and recognition network is determined based on the chaotic attractor method. After the classification and recognition network is determined to have converged, the construction of the classification and recognition network is completed. Select the cross entropy loss function as the loss function of the coal quality evaluation model, define the optimizer of the coal quality evaluation model, and fix the model hyperparameters except the learning rate; Among them, the loss function expression of the coal quality evaluation model is as follows: Among them, x i ,y i They represent the probability distribution of the true value and predicted value of the sample, n is the number of samples, and α is the gradient coefficient of the loss function.
4. The coal quality evaluation method based on spectral and image feature fusion according to claim 1, characterized in that: The method of training the coal quality evaluation model with a training set and testing the accuracy of the coal quality evaluation model with a test set specifically includes: Obtain a training set, use the training set to perform multiple rounds of iterative training on the coal quality evaluation model, obtain an improved coal quality evaluation model, and perform ablation experiments on the input layer, feature extraction layer, feature fusion layer, classification recognition network, and output layer of the coal quality evaluation model based on preset hyperparameters; The training set is enhanced based on the Mosaic-Mixup image enhancement method to obtain the training enhancement set. The coal quality evaluation model is trained again for multiple rounds of iterations until convergence using the training enhancement set as input. Preset the prediction threshold to determine whether the trained coal quality evaluation model meets the prediction threshold. If it meets the prediction threshold, output the coal quality evaluation model; If the prediction threshold is not met, the model parameters are adjusted based on the MSE mean square error, and the training set is enhanced based on the Mosaic-Mixup image enhancement method to obtain the training enhancement set. The coal quality evaluation model is trained again for multiple rounds of iterations until convergence using the training enhancement set as input.
5. The method for evaluating coal quality based on fusion of spectrum and image features as claimed in claim 4, characterized in that: The method of training the coal quality evaluation model with a training set and testing the accuracy of the coal quality evaluation model with a test set specifically includes: Obtain the test set, use the MSE mean square error to calculate the accuracy error between the model's prediction results and the actual results, and determine whether the accuracy error of the coal quality evaluation model meets the preset accuracy; If the preset accuracy is met, the trained coal quality evaluation model is verified and the trained coal quality evaluation model is output; If the preset accuracy is not met, the optimizer Adam is used to adjust the model parameters.
6. The method for evaluating coal quality based on spectral and image feature fusion according to claim 3, characterized in that: The classification recognition network consists of 5 stages, each of which includes a convolution group consisting of n skip connections and a 3*3 convolution layer. The function of the 3*3 convolution layer is to change the dimension of the output feature map.
7. The method for evaluating coal quality based on fusion of spectrum and image features as claimed in claim 5, characterized in that: The coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs the coal quality evaluation results, specifically including: Get the preprocessed set and input the preprocessed set into the input layer; The feature extraction layer extracts image features and spectral features from the feature data of the input layer and performs dimension reduction and normalization processing; The image features and spectral features given by the feature fusion layer are used for feature fusion, and invalid feature information is filtered to obtain a fusion result set of standard image features and spectral features; Load the classification recognition network, which converts all the retained fusion result set information into a standard classification recognition vector; The classification recognition vector is captured through the output layer to obtain the coal quality evaluation result.
8. A coal quality evaluation system based on fusion of spectrum and image features, used to implement the coal quality evaluation method based on fusion of spectrum and image features as claimed in any one of claims 1 to 7, characterized in that: The coal quality evaluation system based on spectral and image feature fusion specifically includes: A data preprocessing module is used to collect coal powder characteristic data, preprocess the coal powder characteristic data, and obtain a preprocessing set; The sample processing module is used to obtain the coal powder sample data, divide the coal powder sample data into a training set and a test set, and pre-process the training set and the test set respectively; A model building module, which is based on a coal quality evaluation model that is fused with spectral and image features, uses a training set to train the coal quality evaluation model, and uses a test set to test the accuracy of the coal quality evaluation model, and outputs the trained coal quality evaluation model; The result output module is used to obtain a preprocessing set, take the preprocessing set as input, execute the coal quality evaluation model, and the coal quality evaluation model performs feature fusion analysis on the preprocessing set and outputs the coal quality evaluation result.
9. The coal quality evaluation system based on spectral and image feature fusion as claimed in claim 8, characterized in that: The data preprocessing module specifically includes: A data cleaning unit is used to obtain coal powder characteristic data, perform data cleaning and missing value processing on the coal powder characteristic data, and obtain a cleaned data set; A standardization unit, used for loading the cleaned data set, and using a softmax function to standardize the cleaned data set to obtain a standard data set; The data integration unit is used to obtain a standard data set, process the data values in the standard data set using Min-max standardization and Z-score standardization respectively, calculate the covariance matrix of the standard data set based on the Min-max standardization results and the Z-score standardization results, perform eigendecomposition on the covariance matrix to obtain a eigendecomposition set, output the eigendecomposition set and the standard data set, and integrate the eigendecomposition set and the standard data set to obtain a preprocessing set.
10. The coal quality evaluation system based on spectral and image feature fusion according to claim 9, characterized in that: The result output module specifically includes: The feature extraction unit extracts image features and spectral features of the feature data of the input layer based on the feature extraction layer and performs dimension reduction and normalization processing; The feature fusion unit uses the image features and spectral features given by the feature fusion layer to perform feature fusion and filter out invalid feature information to obtain a fusion result set of standard image features and spectral features; A classification recognition unit is used to load a classification recognition network, which converts all the retained fusion result set information into a standard classification recognition vector; The evaluation generation unit captures the classification recognition vector through the output layer to obtain the coal quality evaluation result.
Citation Information
Cited By
Coal rock identification method and system based on image and XRF spectral information fusion
CN119251599A
Mobile platform coal quality remote detection system and method
CN119322054A
Deep learning-based intelligent identification system for polluted coal sample
CN119851252A
Sensitivity analysis-based asphalt paving segregation detection method, system and device
CN120823448A
Multi-channel coal quality online detection system and method based on signal enhancement
CN120927592A
