Coal multi-granularity hierarchical sampling and comprehensive detection system

Abstract

The invention discloses a coal multi-granularity hierarchical sampling and comprehensive detection system, and relates to the technical field of coal detection, the system comprises a granularity layering and sampling point selection unit, a weight setting and density optimization unit, a characteristic prediction and sample classification unit, a characteristic extraction and detection strategy optimization unit and a correlation analysis and early warning mechanism unit; the weight setting and density optimization unit is used for setting a corresponding weight for each selected coal sampling point on the basis of historical detection data, and optimizing the sampling density by utilizing the set weight to obtain coal sample data; and the characteristic prediction and sample classification unit is used for predicting the coal quality characteristics of the unsampled region by using a Gaussian process regression algorithm, and dividing the obtained coal sample data into a plurality of coal categories through a fuzzy clustering algorithm. According to the method, granularity layering of the coal is realized through an image processing technology, sampling points are accurately selected, and the precision and representativeness of coal sampling are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of coal detection, and in particular to a coal multi-granularity level sampling and comprehensive detection system. Background Art

[0002] As a traditional and important energy source, coal occupies an indispensable position in the global energy structure. Its quality is not only directly related to power generation efficiency and industrial production costs, but also has a profound impact on environmental protection effectiveness. With the development of the global economy and technological advancement, the requirements for coal quality are increasing, and coal for different purposes needs to meet strict standards and specifications. Therefore, ensuring the quality of coal products has become one of the key tasks in the coal industry.

[0003] In order to ensure that coal meets the established quality standards, accurate sampling and comprehensive testing are essential steps. At present, the field of coal testing mainly relies on manual sampling and traditional laboratory analysis; however, this approach has a series of significant limitations: (1) Insufficient sampling representativeness: Due to the uneven size of coal particles, the randomness of manual sampling and the subjective judgment of the operator may result in the sample not accurately reflecting the true characteristics of the entire coal pile or batch, causing deviation in the test results.

[0004] (2) Large errors: Manual operations inevitably introduce human factors and increase the risk of sampling errors.

[0005] (3) Inefficiency: The traditional manual sampling and laboratory testing process is cumbersome, time-consuming, and difficult to quickly respond to market changes and production needs.

[0006] (4) Lack of real-time and automation: Traditional detection methods usually do not have real-time monitoring capabilities, cannot provide immediate feedback, and have a low degree of automation, which limits the speed and accuracy of data processing.

[0007] (5) Difficulty in multi-size level detection: Traditional detection methods are difficult to conduct comprehensive and systematic analysis of coal of different particle size levels, which limits the in-depth understanding of the physical and chemical properties of coal.

[0008] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0009] In view of the problems in the related technology, the present invention proposes a coal multi-granularity level sampling and comprehensive detection system to overcome the above-mentioned technical problems existing in the existing related technology.

[0010] To this end, the specific technical solution adopted by the present invention is as follows: A coal multi-granularity level sampling and comprehensive detection system, the system includes: a granularity stratification and sampling point selection unit, a weight setting and density optimization unit, a characteristic prediction and sample classification unit, a feature extraction and detection strategy optimization unit and a correlation analysis and early warning mechanism unit; The particle size stratification and sampling point selection unit is used to use image processing technology to stratify the coal particle size and select coal sampling points in each particle size layer; The weight setting and density optimization unit is used to set a corresponding weight for each selected coal sampling point based on historical detection data, and optimize the sampling density using the set weight to obtain coal sample data; The characteristic prediction and sample classification unit is used to predict the coal quality characteristics of the unsampled area using the Gaussian process regression algorithm, and to classify the acquired coal sample data into several coal categories using the fuzzy clustering algorithm; Feature extraction and detection strategy optimization unit, used to perform principal component analysis on each coal category, extract coal quality characteristics, and select the optimal coal quality detection strategy based on comprehensive performance; The correlation analysis and early warning mechanism unit is used to explore the correlation between coal particle size levels and coal quality characteristics according to the optimal coal quality detection strategy, perform anomaly detection on the correlation results, and define the coal quality early warning mechanism based on the anomaly detection results.

[0011] Optionally, the granularity stratification and sampling point selection unit includes: an image acquisition and preprocessing module, an image segmentation module, a granularity analysis module, and a feature extraction and sampling point selection module; An image acquisition and preprocessing module is used to acquire coal image data using a camera and preprocess the image data; An image segmentation module, used to divide the pre-processed image data into a number of image blocks using an image segmentation algorithm; The particle size analysis module is used to perform particle size analysis on each image block using a convolutional neural network algorithm to identify the size distribution of coal particles and divide the image blocks into different particle size levels based on the identification results; The feature extraction and sampling point selection module is used to extract the feature data of coal particles in each particle size level, and based on the extracted feature data, select coal sampling points in each particle size level using a sampling point selection algorithm.

[0012] Optionally, the image acquisition and preprocessing module includes: an image acquisition submodule, a grayscale and binarization processing submodule, a mean filter denoising submodule and a standardization processing submodule; The image acquisition submodule is used to collect image data of coal using a camera according to coal detection requirements; The grayscale and binarization processing submodule is used to perform grayscale processing on the collected image data to obtain grayscale image data, and convert the grayscale image data into binary image data using a threshold segmentation algorithm; The mean filter denoising submodule is used to apply the mean filter algorithm to the binarized image data for denoising; The standardization processing submodule is used to perform standardization processing on the binary image data after denoising.

[0013] Optionally, in each particle size level, characteristic data of coal particles are extracted, and based on the extracted characteristic data, a sampling point selection algorithm is used to select coal sampling points in each particle size level, including: S11. In each particle size level, extracting characteristic data of coal particles using a feature extraction algorithm; S12, based on the extracted feature data, randomly generating a number of initial populations including a set of sampling points in each granularity level; S13, defining a fitness function for each granularity level, and using the fitness function to calculate the fitness value of each sampling point set in the initial population; S14, based on the calculated fitness value, using the roulette wheel selection algorithm to select the best individual from the initial population as the parent generation, and performing a crossover operation on the selected parent generation individual to generate a child generation individual containing a new sampling point set; S15, perform mutation operation on the offspring individuals of each granularity level, and replace the parent individuals with the offspring individuals to obtain a new population; S16. Repeat steps S14 to S15 until a preset number of iterations is reached, and select a set of sampling points with the highest fitness value from the final population as coal sampling points of the corresponding particle size level.

[0014] Optionally, the formula for the feature extraction algorithm is: ; In the formula, δ i represents the characteristic data of the i-th coal particle; i represents the number of characteristic data; n represents the index value of the number of items; m represents the index value of the binomial coefficient; l represents the binomial coefficient; μ i-m，m represents the normalized central moment.

[0015] Optionally, the weight setting and density optimization unit includes: a historical data and preprocessing module, a correlation analysis and weight setting module, a sampling density optimization module, and a sampling point adjustment and sample acquisition module; The historical data and preprocessing module is used to collect historical detection data from each coal sampling point and preprocess the historical detection data; The correlation analysis and weight setting module is used to analyze the correlation between the pre-processed historical detection data and the feature data using statistical analysis algorithms and machine learning algorithms, and to set corresponding weights for each sampling point based on the analysis results; The sampling density optimization module is used to optimize the sampling density by combining the set weights with the selected coal sampling points using the simulated annealing algorithm; The sampling point adjustment and sample acquisition module is used to adjust the sampling points in each granularity level based on the optimized sampling density, and use the adjusted sampling points to sample coal and obtain coal sample data.

[0016] Optionally, the Gaussian process regression algorithm is used to predict the coal quality characteristics of the unsampled area, and the obtained coal sample data is divided into several coal categories by a fuzzy clustering algorithm, including: S21. According to the Gaussian process regression algorithm, a kernel function is selected, a Gaussian process regression model is constructed based on the selected kernel function, and the Gaussian process regression model is trained using historical detection data; S22, based on the spatial relationship between the acquired coal sample data and the unsampled area, combined with the trained Gaussian process regression model, predict the coal quality characteristics of the unsampled area and output the prediction results; S23, based on the prediction results, the cluster centers are initialized using the fuzzy clustering algorithm, and the membership degree of each coal sample data to each cluster center is calculated to obtain a preliminary classification result; S24, optimizing the preliminary classification result by repeating step S23 until a preset number of iterations is reached; S25. From the finally determined membership matrix, the category to which each coal sample data belongs is obtained, and the coal sample data is divided into several coal categories according to the category to which it belongs, and the coal classification result is output.

[0017] Optionally, the membership degree is calculated as: ; In the formula, f ab represents the degree of membership of coal sample data a to category b; c represents the total number of categories; k represents the index value; x a represents the characteristic vector of coal sample data a; x b represents the center vector of category b; x k represents the central vector of category k; m represents the fuzzy index.

[0018] Optionally, the comprehensive performance includes: detection accuracy, detection time and detection cost.

[0019] Optionally, the correlation analysis and early warning mechanism unit includes: a data collection module, a correlation mining module, a quantitative impact module, an anomaly detection execution module and a coal quality early warning definition module; A data collection module is used to collect coal sample data of different particle size levels and corresponding coal quality characteristic data according to the optimal coal quality detection strategy; The correlation mining module is used to mine the correlation between coal particle size levels and coal quality characteristics using the Pearson correlation coefficient; A quantitative impact module is used to quantify the impact of particle size levels on coal quality characteristics based on the correlation analysis results using a pre-built multivariate regression model; An anomaly detection execution module is used to perform anomaly detection on the prediction results based on the quantification results using the isolation forest algorithm, and obtain an anomaly detection result in combination with a preset threshold; The coal quality early warning definition module is used to define the coal quality early warning mechanism based on the abnormal detection results.

[0020] The beneficial effects of the present invention are: 1. The present invention realizes coal particle size stratification through image processing technology and accurately selects sampling points, which significantly improves the accuracy and representativeness of coal sampling, thereby providing high-quality basic samples for subsequent coal quality testing; in addition, the Gaussian process regression algorithm is used to predict the coal quality characteristics of unsampled areas, which not only expands the data coverage, but also realizes a comprehensive understanding and precise control of coal quality.

[0021] 2. The present invention extracts coal quality characteristics through principal component analysis, which can not only accurately identify different types of coal, but also reduce data dimensions and simplify subsequent analysis processes; combined with a comprehensive performance evaluation of detection accuracy, time and cost, it significantly improves the overall detection efficiency and the effectiveness of resource utilization.

[0022] 3. The present invention can timely discover potential coal quality problems by deeply exploring the correlation between coal particle size levels and coal quality characteristics, define a coal quality early warning mechanism based on abnormal detection results, provide timely decision support for coal production and use, and effectively avoid economic losses and safety risks caused by coal quality problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0024] Figure 1It is a principle block diagram of a coal multi-granularity level sampling and comprehensive detection system according to an embodiment of the present invention.

[0025] In the figure: 1. Granularity stratification and sampling point selection unit; 2. Weight setting and density optimization unit; 3. Characteristic prediction and sample classification unit; 4. Feature extraction and detection strategy optimization unit; 5. Correlation analysis and early warning mechanism unit. DETAILED DESCRIPTION

[0026] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments. They can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention.

[0027] According to an embodiment of the present invention, a coal multi-granularity level sampling and comprehensive detection system is provided.

[0028] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to the coal multi-granularity level sampling and comprehensive detection system of the embodiment of the present invention, the system includes: a granularity stratification and sampling point selection unit 1, a weight setting and density optimization unit 2, a characteristic prediction and sample classification unit 3, a feature extraction and detection strategy optimization unit 4 and a correlation analysis and early warning mechanism unit 5; The particle size stratification and sampling point selection unit 1 is used to use image processing technology to stratify the coal particle size and select coal sampling points in each particle size layer.

[0029] Preferably, the particle size stratification and sampling point selection unit 1 comprises: an image acquisition and preprocessing module, an image segmentation module, a particle size analysis module and a feature extraction and sampling point selection module; The image acquisition and preprocessing module is used to acquire the image data of coal using a camera and preprocess the image data.

[0030] Preferably, the image acquisition and preprocessing module includes: an image acquisition submodule, a grayscale and binarization processing submodule, a mean filter denoising submodule and a standardization processing submodule; The image acquisition submodule is used to collect image data of coal using a camera according to coal detection requirements; The grayscale and binarization processing submodule is used to perform grayscale processing on the collected image data to obtain grayscale image data, and convert the grayscale image data into binary image data using a threshold segmentation algorithm; The mean filter denoising submodule is used to apply the mean filter algorithm to the binarized image data for denoising; The standardization processing submodule is used to perform standardization processing on the binary image data after denoising.

[0031] The image segmentation module is used to divide the preprocessed image data into a number of image blocks using an image segmentation algorithm.

[0032] The particle size analysis module is used to perform particle size analysis on each image block using a convolutional neural network algorithm, identify the size distribution of coal particles, and divide the image blocks into different particle size levels based on the identification results.

[0033] It should be noted that, in a specific embodiment, it is assumed that there is an image data set containing 100 coal samples, each sample is cropped into multiple image blocks, and the corresponding granularity levels are labeled. ResNet50 is selected as the convolutional neural network model, and implemented using Python and TensorFlow frameworks. After training and verification, a model with stable performance is obtained; the model is used to perform granularity analysis on each image block, and the image blocks are divided into three granularity levels: fine, medium and coarse; the number of image blocks in each granularity level is counted, and the granularity distribution map of the coal sample is obtained.

[0034] The feature extraction and sampling point selection module is used to extract the feature data of coal particles in each particle size level, and based on the extracted feature data, select coal sampling points in each particle size level using a sampling point selection algorithm.

[0035] Preferably, in each particle size level, characteristic data of coal particles are extracted, and based on the extracted characteristic data, a sampling point selection algorithm is used to select coal sampling points in each particle size level, including: S11. In each particle size level, feature extraction algorithm is used to extract feature data of coal particles.

[0036] Preferably, the formula of the feature extraction algorithm is: ; In the formula, δ i represents the characteristic data of the i-th coal particle; i represents the number of characteristic data; n represents the index value of the number of items; m represents the index value of the binomial coefficient; l represents the binomial coefficient; μ i-m，m represents the normalized central moment.

[0037] S12. Based on the extracted feature data, a number of initial populations including a set of sampling points are randomly generated in each granularity level.

[0038] S13. Define a fitness function for each granularity level, and use the fitness function to calculate the fitness value of each sampling point set in the initial population.

[0039] S14. Based on the calculated fitness value, the best individual is selected from the initial population as the parent using the roulette wheel selection algorithm, and a crossover operation is performed on the selected parent individual to generate a child individual containing a new sampling point set.

[0040] S15. Perform mutation operation on the offspring individuals at each granularity level, and replace the parent individuals with the offspring individuals to obtain a new population.

[0041] S16. Repeat steps S14 to S15 until a preset number of iterations is reached, and select a set of sampling points with the highest fitness value from the final population as coal sampling points of the corresponding particle size level.

[0042] The weight setting and density optimization unit 2 is used to set a corresponding weight for each selected coal sampling point based on historical detection data, and optimize the sampling density using the set weight to obtain coal sample data.

[0043] Preferably, the weight setting and density optimization unit 2 includes: a historical data and preprocessing module, a correlation analysis and weight setting module, a sampling density optimization module, and a sampling point adjustment and sample acquisition module; The historical data and preprocessing module is used to collect historical detection data from each coal sampling point and preprocess the historical detection data.

[0044] The correlation analysis and weight setting module is used to use statistical analysis algorithms and machine learning algorithms to analyze the correlation between preprocessed historical detection data and feature data, and based on the analysis results, set corresponding weights for each sampling point.

[0045] The sampling density optimization module is used to combine the set weights with the selected coal sampling points and optimize the sampling density using the simulated annealing algorithm.

[0046] The sampling point adjustment and sample acquisition module is used to adjust the sampling points in each granularity level based on the optimized sampling density, and use the adjusted sampling points to sample coal and obtain coal sample data.

[0047] It should be noted that the coal sample data includes physical property data and chemical property data.

[0048] Physical property data: such as coal particle size distribution (i.e. the proportion of particles of different sizes), density, hardness, moisture content, ash content, volatile matter content, etc. These data reflect the basic physical properties and composition of coal.

[0049] Chemical property data: such as the calorific value of coal (i.e. the energy released when burned), sulfur content, phosphorus content, nitrogen content, heavy metal content, etc. These data are crucial for evaluating the combustion performance, environmental impact and possible uses of coal.

[0050] The characteristic prediction and sample classification unit 3 is used to predict the coal quality characteristics of the unsampled area by using the Gaussian process regression algorithm, and to classify the acquired coal sample data into several coal categories by using the fuzzy clustering algorithm.

[0051] Preferably, the Gaussian process regression algorithm is used to predict the coal quality characteristics of the unsampled area, and the obtained coal sample data is divided into several coal categories by a fuzzy clustering algorithm, including: S21. According to the Gaussian process regression algorithm, a kernel function is selected, a Gaussian process regression model is constructed based on the selected kernel function, and the Gaussian process regression model is trained using historical detection data.

[0052] S22. Based on the spatial relationship between the acquired coal sample data and the unsampled area, combined with the trained Gaussian process regression model, the coal quality characteristics of the unsampled area are predicted and the prediction results are output.

[0053] S23. Based on the prediction results, the cluster centers are initialized using the fuzzy clustering algorithm, and the membership of each coal sample data to each cluster center is calculated to obtain a preliminary classification result.

[0054] Preferably, the calculation formula of the membership degree is: ; In the formula, f ab represents the degree of membership of coal sample data a to category b; c represents the total number of categories; k represents the index value; x a represents the characteristic vector of coal sample data a; x b represents the center vector of category b; x k represents the central vector of category k; m represents the fuzzy index.

[0055] S24, optimizing the preliminary classification result by repeating step S23 until a preset number of iterations is reached.

[0056] S25. From the finally determined membership matrix, the category to which each coal sample data belongs is obtained, and the coal sample data is divided into several coal categories according to the category to which it belongs, and the coal classification result is output.

[0057] It should be noted that coal classification categories include coal indicators, coal uses, geological origins and production areas, market standards, etc.

[0058] The feature extraction and detection strategy optimization unit 4 is used to perform principal component analysis on each coal category, extract coal quality characteristics, and select the optimal coal quality detection strategy based on comprehensive performance.

[0059] Preferably, the comprehensive performance includes: detection accuracy, detection time and detection cost.

[0060] It should be noted that the coal quality testing strategy is as follows: (1) Detection method based on physical properties Density measurement: The density of coal is measured directly by floating and sinking experiments or densitometer.

[0061] Ash determination: Determination of mineral content in coal using high temperature ignition method.

[0062] Sulfur content determination: Coulometric titration or X-ray fluorescence spectrometer (XRF) is used to measure the sulfur content in coal.

[0063] (2) Detection methods based on chemical properties Elemental analysis: Use an elemental analyzer to determine the proportions of carbon, hydrogen, oxygen, nitrogen and other elements in coal.

[0064] Volatile matter determination: The volatile components in coal are determined by heating in a muffle furnace.

[0065] Fixed carbon determination: Fixed carbon content is calculated based on total carbon minus volatile matter and ash.

[0066] (3) Detection method based on calorific value Calorific value determination: The lower heating value (LHV) or higher heating value (HHV) of coal is measured using a bomb calorimeter.

[0067] (4) Detection methods based on optics and spectroscopy Infrared spectroscopy analysis: The functional group structure of coal was analyzed by Fourier transform infrared spectroscopy (FTIR).

[0068] Raman spectroscopy: used to characterize the microstructure and composition of coal.

[0069] (5) Detection methods based on image processing and machine vision Particle size distribution analysis: Use image processing technology to analyze the size distribution of coal particles.

[0070] Color and Texture Analysis: Evaluate the color and surface texture characteristics of coal using machine vision systems.

[0071] (6) Detection method based on multimodal data fusion Comprehensive sensor network: Combine multiple sensors (such as temperature, humidity, and gas sensors) to obtain comprehensive data.

[0072] Big Data Platform: Build a cloud-based big data platform to store and analyze historical and current data, and provide customized reports and recommendations.

[0073] (7) Detection method based on model prediction Gaussian Process Regression (GPR): used to predict coal quality changes in unsampled areas.

[0074] Machine learning model: Apply random forest, gradient boosted tree (GBDT), support vector machine (SVM) and other algorithms to establish coal quality prediction model.

[0075] In addition, for each coal category, the following steps are performed to select the optimal coal quality detection strategy: Step 1: Principal Component Analysis (PCA): PCA is performed on samples of each coal category to extract the principal components that best represent the coal quality characteristics, reducing the data dimension while retaining key information.

[0076] Step 2: Comprehensive performance evaluation: Test accuracy: Evaluates how accurately different test methods measure specific coal characteristics.

[0077] Detection time: Compare the time required for various methods and choose a fast and efficient detection method.

[0078] Testing cost: Consider equipment purchase, maintenance costs, and the cost of each test, and choose an economically feasible method.

[0079] Step 3: Strategy selection: Based on the PCA results and comprehensive performance evaluation, the most suitable combination of detection methods is selected for each coal category to ensure that the detection time and cost are shortened as much as possible while meeting the accuracy requirements.

[0080] Step 4: Continuous Improvement: Regularly review and evaluate the effectiveness of the selected strategies, and continuously optimize and adjust the detection strategies as new technologies develop and new data accumulate.

[0081] The correlation analysis and early warning mechanism unit 5 is used to mine the correlation between coal particle size levels and coal quality characteristics according to the optimal coal quality detection strategy, perform anomaly detection on the correlation results, and define a coal quality early warning mechanism based on the anomaly detection results.

[0082] Preferably, the correlation analysis and early warning mechanism unit 5 includes: a data collection module, a correlation mining module, a quantitative impact module, an anomaly detection execution module and a coal quality early warning definition module; A data collection module is used to collect coal sample data of different particle size levels and corresponding coal quality characteristic data according to the optimal coal quality detection strategy; The correlation mining module is used to mine the correlation between coal particle size levels and coal quality characteristics using the Pearson correlation coefficient; A quantitative impact module is used to quantify the impact of particle size levels on coal quality characteristics based on the correlation analysis results using a pre-built multivariate regression model; An anomaly detection execution module is used to perform anomaly detection on the prediction results based on the quantification results using the isolation forest algorithm, and obtain an anomaly detection result in combination with a preset threshold; The coal quality early warning definition module is used to define the coal quality early warning mechanism based on the abnormal detection results.

[0083] In summary, with the help of the above technical solution of the present invention, the coal particle size stratification is realized through image processing technology, and the sampling points are accurately selected, which significantly improves the accuracy and representativeness of coal sampling, thereby providing high-quality basic samples for subsequent coal quality detection; in addition, the coal quality characteristics of unsampled areas are predicted using the Gaussian process regression algorithm, which not only expands the data coverage, but also realizes a comprehensive understanding and precise control of coal quality. Extracting coal quality characteristics through principal component analysis can not only accurately identify different coal categories, but also reduce data dimensions and simplify subsequent analysis processes; combined with the comprehensive performance evaluation of detection accuracy, time and cost, the overall detection efficiency and the effectiveness of resource utilization are significantly improved. By deeply exploring the correlation between coal particle size levels and coal quality characteristics, potential coal quality problems can be discovered in a timely manner, and a coal quality early warning mechanism can be defined based on abnormal detection results to provide timely decision support for coal production and use, effectively avoiding economic losses and safety risks caused by coal quality problems.

[0084] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A coal multi-granularity level sampling and comprehensive detection system, characterized in that: The system includes: granularity stratification and sampling point selection unit, weight setting and density optimization unit, characteristic prediction and sample classification unit, feature extraction and detection strategy optimization unit and correlation analysis and early warning mechanism unit; The particle size stratification and sampling point selection unit is used to use image processing technology to stratify the coal particle size and select coal sampling points in each particle size level; The weight setting and density optimization unit is used to set a corresponding weight for each selected coal sampling point based on historical detection data, and optimize the sampling density using the set weight to obtain coal sample data; The characteristic prediction and sample classification unit is used to predict the coal quality characteristics of the unsampled area by using the Gaussian process regression algorithm, and to classify the acquired coal sample data into several coal categories by using the fuzzy clustering algorithm; The feature extraction and detection strategy optimization unit is used to perform principal component analysis on each coal category, extract coal quality features, and select the optimal coal quality detection strategy based on comprehensive performance; The correlation analysis and early warning mechanism unit is used to mine the correlation between coal particle size levels and coal quality characteristics according to the optimal coal quality detection strategy, perform anomaly detection on the correlation results, and define a coal quality early warning mechanism based on the anomaly detection results.

2. A coal multi-granularity level sampling and comprehensive detection system according to claim 1, characterized in that: The granularity stratification and sampling point selection unit includes: an image acquisition and preprocessing module, an image segmentation module, a granularity analysis module and a feature extraction and sampling point selection module; The image acquisition and preprocessing module is used to acquire image data of coal using a camera and preprocess the image data; The image segmentation module is used to divide the pre-processed image data into a number of image blocks using an image segmentation algorithm; The particle size analysis module is used to perform particle size analysis on each image block using a convolutional neural network algorithm to identify the size distribution of coal particles and divide the image blocks into different particle size levels based on the identification results; The feature extraction and sampling point selection module is used to extract the feature data of coal particles in each particle size level, and select coal sampling points in each particle size level using a sampling point selection algorithm based on the extracted feature data.

3. A coal multi-granularity level sampling and comprehensive detection system according to claim 2, characterized in that: The image acquisition and preprocessing module includes: an image acquisition submodule, a grayscale and binarization processing submodule, a mean filter denoising submodule and a standardization processing submodule; The image acquisition submodule is used to collect image data of coal using a camera according to coal detection requirements; The grayscale and binarization processing submodule is used to perform grayscale processing on the collected image data to obtain grayscale image data, and convert the grayscale image data into binary image data using a threshold segmentation algorithm; The mean filter denoising submodule is used to apply a mean filter algorithm to the binarized image data to perform denoising processing; The standardization processing submodule is used to perform standardization processing on the binary image data after denoising processing.

4. A coal multi-granularity level sampling and comprehensive detection system according to claim 3, characterized in that: The method of extracting characteristic data of coal particles in each particle size level and selecting coal sampling points in each particle size level by using a sampling point selection algorithm based on the extracted characteristic data includes: S11. In each particle size level, extracting characteristic data of coal particles using a feature extraction algorithm; S12, based on the extracted feature data, randomly generating a number of initial populations including a set of sampling points in each granularity level; S13, defining a fitness function for each granularity level, and using the fitness function to calculate the fitness value of each sampling point set in the initial population; S14, based on the calculated fitness value, using the roulette wheel selection algorithm to select the best individual from the initial population as the parent generation, and performing a crossover operation on the selected parent generation individual to generate a child generation individual containing a new sampling point set; S15, perform mutation operation on the offspring individuals of each granularity level, and replace the parent individuals with the offspring individuals to obtain a new population; S16. Repeat steps S14 to S15 until a preset number of iterations is reached, and select a set of sampling points with the highest fitness value from the final population as coal sampling points of the corresponding particle size level.

5. A coal multi-granularity level sampling and comprehensive detection system according to claim 4, characterized in that: The formula of the feature extraction algorithm is: ； In the formula, δ i represents the characteristic data of the i-th coal particle; i represents the number of characteristic data; n represents the index value of the number of items; m represents the index value of the binomial coefficient; l represents the binomial coefficient; μ i-m，m表示归一化中心矩。 6. A coal multi-granularity level sampling and comprehensive detection system according to claim 1, characterized in that: The weight setting and density optimization unit includes: a historical data and preprocessing module, a correlation analysis and weight setting module, a sampling density optimization module, and a sampling point adjustment and sample acquisition module; The historical data and preprocessing module is used to collect historical detection data from each coal sampling point and preprocess the historical detection data; The correlation analysis and weight setting module is used to analyze the correlation between the pre-processed historical detection data and the feature data by using a statistical analysis algorithm and a machine learning algorithm, and to set a corresponding weight for each sampling point based on the analysis result; The sampling density optimization module is used to optimize the sampling density by combining the set weights with the selected coal sampling points using a simulated annealing algorithm; The sampling point adjustment and sample acquisition module is used to adjust the sampling points in each granularity level based on the optimized sampling density, and use the adjusted sampling points to perform coal sampling to obtain coal sample data.

7. A coal multi-granularity level sampling and comprehensive detection system according to claim 1, characterized in that: The Gaussian process regression algorithm is used to predict the coal quality characteristics of the unsampled area, and the obtained coal sample data is divided into several coal categories by the fuzzy clustering algorithm, including: S21. According to the Gaussian process regression algorithm, a kernel function is selected, a Gaussian process regression model is constructed based on the selected kernel function, and the Gaussian process regression model is trained using historical detection data; S22, based on the spatial relationship between the acquired coal sample data and the unsampled area, combined with the trained Gaussian process regression model, predict the coal quality characteristics of the unsampled area and output the prediction results; S23, based on the prediction results, the cluster centers are initialized using the fuzzy clustering algorithm, and the membership degree of each coal sample data to each cluster center is calculated to obtain a preliminary classification result; S24, optimizing the preliminary classification result by repeating step S23 until a preset number of iterations is reached; S25. From the finally determined membership matrix, the category to which each coal sample data belongs is obtained, and the coal sample data is divided into several coal categories according to the category to which it belongs, and the coal classification result is output.

8. A coal multi-granularity level sampling and comprehensive detection system according to claim 7, characterized in that: The calculation formula of the membership degree is: ； In the formula, f ab represents the degree of membership of coal sample data a to category b; c represents the total number of categories; k represents the index value; x a represents the characteristic vector of coal sample data a; x b represents the center vector of category b; x k represents the central vector of category k; m represents the fuzzy index.

9. A coal multi-granularity level sampling and comprehensive detection system according to claim 1, characterized in that: The comprehensive performance includes: detection accuracy, detection time and detection cost.

10. A coal multi-granularity level sampling and comprehensive detection system according to claim 1, characterized in that: The correlation analysis and early warning mechanism unit includes: a data collection module, a correlation mining module, a quantitative impact module, an anomaly detection execution module and a coal quality early warning definition module; The data collection module is used to collect coal sample data of different particle size levels and corresponding coal quality characteristic data according to the optimal coal quality detection strategy; The correlation mining module is used to mine the correlation between coal particle size levels and coal quality characteristics using the Pearson correlation coefficient; The quantitative impact module is used to quantify the impact of particle size levels on coal quality characteristics based on the correlation analysis results using a pre-built multivariate regression model; The anomaly detection execution module is used to perform anomaly detection on the prediction result based on the quantification result using the isolation forest algorithm, and obtain the anomaly detection result in combination with a preset threshold; The coal quality early warning definition module is used to define a coal quality early warning mechanism according to abnormal detection results.

Images