Coal-rock-gangue separation method and system based on multi-source data fusion analysis

Abstract

The present invention discloses a coal-rock-gangue sorting method and system based on multi-source data fusion analysis, which belongs to the field of coal-rock-gangue sorting, performs the same load loading test on rock blocks and coal blocks, monitors the strain of coal blocks and rock blocks, monitors the heat released during loading, tests different types of coal, rock, and gangue, analyzes the spectral absorption peak characteristics, associates different strains, released heat, and different absorption peak characteristics with corresponding types of coal, rock blocks, and gangue, respectively establishes strain multi-classification models, heat multi-classification models, and spectral multi-classification models, respectively uses the above models to sort and identify the mixture of coal, rock, and gangue, and determines the final sorting result based on the dynamic weighted fusion of three sorting results. Combined with the apparent characteristics, physical properties, and composition differences of coal, rock, and gangue, deep-level decision fusion is carried out to solve the uncertainty problem caused by a single sorting method, and ensure the accuracy and reliability of the sorting results.

Description

Technical Field

[0001] The present invention belongs to the field of coal-rock-gangue separation, and in particular relates to a coal-rock-gangue separation method and system based on multi-source data fusion analysis. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] In coal mining, efficient and accurate identification of coal, rock and gangue is an important prerequisite for improving economic benefits and reducing equipment loss. In the actual coal mining process, the complex situation of coal, rock and gangue mixing often occurs. The mixture of coal, rock and gangue will reduce the quality of coal. This is because rocks and gangue are not combustible. If rocks and gangue are not separated, more ash will be produced when coal is burned, which will reduce the calorific value and affect the combustion efficiency. In addition, the presence of rocks and gangue will greatly increase the transportation cost. Therefore, how to accurately and efficiently separate coal, rock and gangue is a crucial step in the coal mining and processing process.

[0004] The current coal gangue sorting methods and existing problems mainly include: ① Manual sorting, sorting coal gangue according to color, shape, weight, etc., relying on manual experience for qualitative judgment, with low sorting efficiency and insufficient sorting; ② Heavy medium coal gangue sorting using the difference in density between coal and gangue, but the heavy medium coal gangue sorting method has a cumbersome process. There are many types of coal, and it is difficult to ensure that the density of coal, rock and gangue are significantly different, so the accuracy and efficiency of coal-rock-gangue sorting are low; ③ The difference in the reflection and absorption of gamma rays and X-rays by coal and gangue can also be used for coal-rock-gangue identification, but gamma rays and X-rays release strong radiation, which is extremely harmful to the human body and is prone to danger; ④ Obtain machine vision data such as images of coal, rocks and gangue, and use machine learning to identify image grayscale, texture, shape and other features to achieve coal-rock-gangue identification, but image recognition has a fatal flaw, that is, there must be a complete and clear coal rock image before identification. When the coal gangue is on the conveyor belt, there will be a lot of dust and dim light, which greatly increases the difficulty of identification. Summary of the invention

[0005] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a coal-rock-gangue sorting method and system based on multi-source data fusion analysis, which jointly utilizes the differences in the apparent characteristics, physical properties, and composition of different coals and rocks to conduct deep-level decision fusion, thereby achieving high-precision sorting of coal, rock and gangue, improving the technical level of coal, rock and gangue sorting, and promoting the development of the coal industry.

[0006] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0007] In a first aspect, the present invention provides a coal-rock-gangue separation method based on multi-source data fusion analysis, comprising:

[0008] Collect different types of coal, rock, gangue and process them into rock and coal blocks of the same size;

[0009] The same load loading test is performed on the rock block and the coal block, the strain of the coal block and the rock block is monitored, different strains are correlated with the corresponding types of coal, rock block and gangue, and a strain multi-classification model is established;

[0010] Monitor the heat released by the coal and rock blocks during the load loading process, correlate the heat released under the same size and the same load loading conditions with the corresponding types of coal, rock, and gangue, and establish a heat multi-classification model;

[0011] Test and identify the spectral absorption peaks of different types of coal, rock, and gangue, associate the different absorption peak characteristics with the corresponding types of coal, rock, and gangue, and establish a spectral multi-classification model;

[0012] The strain multi-classification model, thermal multi-classification model and spectral multi-classification model are used respectively to sort and identify the mixture of coal, rock and gangue, and the sorting results of the strain multi-classification model, thermal multi-classification model and spectral multi-classification model are integrated to determine the final sorting result.

[0013] According to a further technical solution, the strain multi-classification model is trained using a neural network, the input parameters are the total strain of the test block and the strain values ​​corresponding to different moments, and the output parameters are the probabilities of the corresponding types of coal, rock or gangue.

[0014] According to a further technical solution, the heat multi-classification model is trained using a neural network, the input parameters are the total heat release, heat release rate and average temperature of test blocks of the same size and under the same load conditions, and the output parameters are the probabilities of the corresponding types of coal, rock or gangue.

[0015] According to a further technical solution, the spectral multi-classification model is trained using a neural network, the input parameters are the absorption peak characteristics of different test blocks, and the output parameters are the probabilities of the corresponding types of coal, rock or gangue.

[0016] A further technical solution is to integrate the sorting results of the strain multi-classification model, the heat multi-classification model and the spectral multi-classification model to determine the final sorting result as follows: weighted average the three sorting results, select the category with the highest probability after the weighted average as the final sorting result, and transport the sorted coal and rock separately.

[0017] In a further technical solution, the weighting is a dynamic weighting of the uncertainty of the fusion model prediction, the uncertainty is quantified by the entropy value, and the weight is assigned according to the entropy value; the calculation formula of the entropy value is:

[0018]

[0019] in, is the category probability distribution predicted by the model, C is the number of categories, The model predicts the probability that the sample belongs to category i.

[0020] A further technical solution is to perform weighted averaging according to the weights, and select the category with the highest probability after weighted averaging as the final sorting result. The calculation formula is as follows:

[0021]

[0022] in, is the entropy value of the m-th model for the current sample, M is the total number of classification models, k represents the k-th classification model, is the weight of the mth model, is the entropy value of the k-th model for the current sample;

[0023] Assume that the predicted probability of model m is , then the final weighted output probability is .

[0024] In a second aspect, the present invention provides a coal-rock-gangue separation system based on multi-source data fusion analysis, comprising:

[0025] A sample collection and processing module, which is configured to: collect different types of coal, rock, and gangue, and process them into rock and coal blocks of the same size;

[0026] A strain multi-classification model building module is configured to: perform the same load loading test on the rock block and the coal block, monitor the strain of the coal block and the rock block, associate different strains with corresponding types of coal, rock block and gangue, and establish a strain multi-classification model;

[0027] A heat multi-classification model building module is configured to: monitor the heat released by the coal and rock blocks during the load loading process, associate the heat released under the same size and the same load loading conditions with the corresponding types of coal, rock, and gangue, and establish a heat multi-classification model;

[0028] A spectral multi-classification model building module is configured to: test and identify the spectral absorption peaks of different types of coal, rock, and gangue, associate different absorption peak features with corresponding types of coal, rock, and gangue, and establish a spectral multi-classification model;

[0029] The fusion sorting module is configured to sort and identify the mixture of coal, rock and gangue using the strain multi-classification model, the heat multi-classification model and the spectral multi-classification model respectively, and to fuse the sorting results of the strain multi-classification model, the heat multi-classification model and the spectral multi-classification model to determine the final sorting result.

[0030] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the coal-rock-gangue separation method based on multi-source data fusion analysis as described in the first aspect.

[0031] In a fourth aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the coal-rock-gangue separation method based on multi-source data fusion analysis as described in the first aspect are implemented.

[0032] One or more of the above technical solutions have the following beneficial effects:

[0033] The present invention combines the apparent characteristics, physical properties, and composition differences of coal, rock, and gangue. In terms of performance characteristics, considering the influence of dust in the sorting environment, non-contact strain is selected to replace images for sorting; in terms of physical properties, coal and rock have differences in thermodynamic properties, molecular structure, and mechanical properties, and the heat released during compression is different, so the physical property of the released temperature is selected to characterize; in terms of composition, the composition is the key to the most essential difference between the reaction objects. Coal, rock, and gangue have different material compositions, and the composition is quantitatively analyzed through the spectrum; and a deep-level decision-making fusion of multiple sorting methods can significantly improve the sorting accuracy of coal, rock, and gangue, solve the uncertainty problem caused by a single sorting method, and ensure the accuracy and reliability of the sorting results. The three bases of strain, heat, and spectrum are combined for sorting, and coal, rock, and gangue are sorted from three different angles: apparent characteristics, physical properties, and composition differences; and these three methods are easy to implement and simple in process compared to using density differences and manual sorting, and are more practical.

[0034] The present invention fuses strain, heat and spectrum multi-classification models. These three models are three classification models with different levels of characteristics, easy to implement and complementary, and the deep fusion effect is better. In addition, dynamic weighting of uncertainty based on entropy is performed during fusion, and the model weight is flexibly adjusted according to the uncertainty, which makes the model more adaptable and the classification more accurate.

[0035] The present invention can accurately sort coal gangue, increase the proportion of marketable coal, reduce unnecessary expenses in the processing and transportation process, further reduce production costs, and optimize resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0037] Figure 1 It is a flow chart of the coal-rock-gangue separation method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0038] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0039] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0040] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0041] Embodiment 1

[0042] like Figure 1 As shown, this embodiment discloses a coal-rock-gangue separation method based on multi-source data fusion analysis, which includes the following steps:

[0043] Step 1: Collect different types of coal, rock, and gangue, and process them into rock and coal blocks of the same size;

[0044] In this embodiment, different types of coal include fat coal, lignite, coking coal, gas fat coal, gas coal, and anthracite; different types of rocks include granite, diabase, sandstone, limestone, quartz sandstone, quartzite, plagioclase, and shale commonly found in coal areas; different types of gangue include lignite gangue and bituminous coal gangue.

[0045] Step 2: Perform the same load loading test on the rock block and the coal block, monitor the strain of the coal block and the rock block, associate different strains with the corresponding types of coal, rock block and gangue, and establish a strain multi-classification model;

[0046] Correlation analysis is to establish a nonlinear complex mapping relationship between strain values ​​and types of coal, rock and gangue.

[0047] In this embodiment, a three-dimensional non-contact strain monitor is used to monitor the strain of coal and rock blocks. Before the test, the three-dimensional non-contact strain monitor is calibrated, including checking the optical system and calibration points of the equipment, and making the instrument parallel and stable to the surface of the test block.

[0048] The load loading test specifically applies the same axial compression force to rock blocks and coal blocks of the same size, and uses a three-dimensional non-contact strain monitor to capture and record the total strain of the test block from the beginning to the end of loading and the corresponding strain value at each moment.

[0049] The strain multi-classification model is trained using a neural network. Specifically, the neural network includes an input layer, a hidden layer, and an output layer. The input parameters are the total strain of the test block and the strain values ​​corresponding to different times. The output parameters are the probabilities of the corresponding coal, rock, or gangue types.

[0050] The output layer consists of a fully connected layer and a Softmax layer. For a multi-classification task with C categories, the output of the fully connected layer is , represents the unnormalized score of category i, and then uses the Softmax layer to map the unnormalized score into a probability distribution. The probability of each category is Calculated by the following formula:

[0051]

[0052] in, is the probability that the input sample x belongs to category i, represents the sum of the exponentials of the unnormalized scores of all categories, represents the unnormalized score of category j.

[0053] Among them, the use of neural network for training includes: constructing a data set with the total strain obtained from the test and the strain values ​​at different times and their corresponding coal and rock gangue types, dividing the data set into a training set and a test set, using cross entropy loss as the loss function for training on the training set, and using the test set to evaluate the model's classification accuracy, precision, recall rate, F1 score and other indicators. After the evaluation indicators meet the classification accuracy, the strain multi-classification model is saved.

[0054] Step 3: monitor the heat released by the coal and rock blocks during the load loading process, correlate the heat released under the same size and the same load loading conditions with the corresponding types of coal, rock, and gangue, and establish a heat multi-classification model;

[0055] Correlation analysis is to establish a nonlinear complex mapping relationship between the released heat and the types of coal, rock and gangue.

[0056] In this embodiment, an infrared thermal imager is used to monitor the heat released during the load loading process. The infrared thermal imaging should have high resolution and sensitivity to ensure the accuracy of measuring temperature changes. Before starting the test, the infrared thermal imager is aimed at the test block to ensure that its field of view covers the entire surface of the test block.

[0057] The thermal imaging test specifically records the total amount of heat released, the heat release rate and the average temperature of each specimen during the loading process.

[0058] The heat multi-classification model is trained using a neural network. Specifically, the neural network includes an input layer, a hidden layer, and an output layer. The input parameters are the total heat release, heat release rate, and average temperature of test blocks of the same size and under the same load conditions. The output parameters are the probabilities of the corresponding types of coal, rock, or gangue.

[0059] The output layer consists of a fully connected layer and a Softmax layer. For a multi-classification task with C categories, the output of the fully connected layer is , represents the unnormalized score of category i, and then uses the Softmax layer to map the unnormalized score into a probability distribution. The probability of each category is Calculated by the following formula:

[0060]

[0061] in, is the probability that the input sample x belongs to category i, represents the sum of the exponentials of the unnormalized scores of all categories, C represents the number of categories, represents the unnormalized score of category j.

[0062] Among them, the use of neural networks for training includes: constructing a data set based on the total heat release, heat release rate and average temperature of test blocks of the same size and under the same load conditions obtained in the test and their corresponding coal gangue types, dividing the data set into a training set and a test set, using cross entropy loss as the loss function for training on the training set, and using the test set to evaluate the model's classification accuracy, precision, recall rate, F1 score and other indicators. After the evaluation indicators meet the classification accuracy, the heat multi-classification model is saved.

[0063] Step 4: Test and identify the spectral absorption peaks of different types of coal, rock, and gangue, associate the different absorption peak characteristics with the corresponding types of coal, rock, and gangue, and establish a spectral multi-classification model;

[0064] Correlation analysis is to establish a nonlinear and complex mapping relationship between absorption peak characteristics and types of coal, rock and gangue.

[0065] In this embodiment, near infrared spectroscopy is used to test different types of coal, rock, and gangue to analyze the absorption peak characteristics. The wavelength range of the near infrared spectrometer is generally 700nm-2500nm. Before the test begins, the near infrared spectrometer is calibrated, including zero point calibration and wavelength calibration.

[0066] The spectral test is specifically as follows: use near-infrared spectroscopy to test and record the spectral data of each test block, use baseline correction and smoothing filtering to process the original spectral data to subtract the background, remove the influence of noise and water vapor, and analyze the spectral data to identify the main absorption peaks and extract the wavelength, absorbance and other characteristic values ​​of each absorption peak.

[0067] The spectral multi-classification model is trained using a neural network. Specifically, the neural network includes an input layer, a hidden layer, and an output layer. The input parameters are the absorption peak characteristics of different test blocks, including characteristic values ​​such as wavelength position and absorbance, and the output parameters are the probabilities of the corresponding types of coal, rock, or gangue.

[0068] The output layer consists of a fully connected layer and a Softmax layer. For a multi-classification task with C categories, the output of the fully connected layer is , represents the unnormalized score of category i, and then uses the Softmax layer to map the unnormalized score into a probability distribution. The probability of each category is Calculated by the following formula:

[0069]

[0070] in, is the probability that the input sample x belongs to category i, represents the sum of the exponentials of the unnormalized scores of all categories, C represents the number of categories, represents the unnormalized score of category j.

[0071] Among them, the use of neural networks for training includes: constructing a data set based on the absorption peak characteristics of different test blocks obtained from the test, including characteristic values ​​such as wavelength position and absorbance and their corresponding coal gangue types, dividing the data set into a training set and a test set, using cross entropy loss as a loss function for training on the training set, and using the test set to evaluate the model's classification accuracy, precision, recall rate, F1 score and other indicators. After the evaluation indicators meet the classification accuracy, the spectral multi-classification model is saved.

[0072] The classification accuracy of the strain multi-classification model, the heat multi-classification model and the spectral multi-classification model refers to the proportion of correctly classified samples in the test set to the total samples; the classification accuracy is differentiated according to the actual requirements of the site. When the site has high requirements for the sorting accuracy of coal and rock gangue, the evaluation indicators should also be improved.

[0073] It should be noted that the strain multi-classification model, the heat multi-classification model and the spectrum multi-classification model use the same neural network structure, and the corresponding neural network can be flexibly selected according to actual conditions, which is not specifically limited in this embodiment.

[0074] Step 5: Use the strain multi-classification model, thermal multi-classification model and spectral multi-classification model to sort and identify the mixture of coal, rock and gangue respectively, and integrate the sorting results of the strain multi-classification model, thermal multi-classification model and spectral multi-classification model to determine the final sorting result.

[0075] In this embodiment, in the actual coal-rock-gangue sorting, a three-dimensional non-contact strain monitor, an infrared thermal imager and a near-infrared spectrometer are set up on the conveyor belt to test the coal-rock-gangue mixture to be sorted, and the strain multi-classification model, the heat multi-classification model and the mineral multi-classification model are used to sort and identify the coal, rock and gangue respectively, and the sorting results of the three multi-classification models are fused at the decision level as the final sorting result.

[0076] The actual coal-rock gangue sorting is as follows: before sorting, the mined coal-rock gangue mixture is cut into blocks of the same size using a cutting machine, and before being transported on a conveyor belt, the blocks are compressed using a compression testing machine;

[0077] A three-dimensional non-contact strain monitor is set up on the left side of the conveyor belt and aimed at the test block under the experimental compressor to capture and record the total strain of the test block from the beginning to the end of loading and the corresponding strain value at each moment; an infrared thermal imager is set up at the corresponding right position of the conveyor belt and aimed at the loaded test piece to record the total amount of heat released, heat release rate and average temperature of each test piece during the loading process; a near-infrared spectrometer is set up above the transmission belt to test and record the spectral data of each test block.

[0078] Sorting and identification are as follows: using the established strain multi-classification model, heat multi-classification model and spectral multi-classification model to separately predict the types of test blocks and output the corresponding prediction probabilities.

[0079] As in the example of this embodiment, the probability of outputting fat coal using the strain multi-classification model is 40%, the probability of granite is 40%, and the probability of lignite gangue is 20%; the probability of outputting fat coal using the heat multi-classification model is 40%, the probability of granite is 50%, and the probability of lignite gangue is 10%; the probability of outputting fat coal using the spectral multi-classification model is 24%, the probability of granite is 71%, and the probability of lignite gangue is 5%.

[0080] The final sorting result is determined by integrating the sorting results of the strain multi-classification model, the heat multi-classification model and the spectral multi-classification model. The three sorting results are weighted averaged, and the category with the highest probability after the weighted average is selected as the final sorting result. The sorted coal and rock are then transported separately.

[0081] In actual coal and rock gangue sorting, affected by on-site environment such as dust, humidity and temperature changes, etc., although these multi-classification models have been trained, there may still be uncertainties in practice. In order to reduce classification errors, the present invention integrates the uncertainties predicted by each multi-classification model to determine the final sorting result.

[0082] Specifically, fusion sorting is to dynamically weight the uncertainty of the fusion model prediction, quantify the uncertainty through entropy, and assign weights based on the entropy value. For a given sample, the class probability distribution predicted by the model is , C is the number of categories, The model predicts the probability that the sample belongs to category i. The entropy calculation formula is:

[0083] .

[0084] When the entropy value is low, it means that the model prediction is stable and has high confidence in the classification result of the current sample, and a higher weight is assigned in the dynamic weighting; when the entropy value is high, it means that the model prediction is unstable, and a lower weight is assigned in the dynamic weighting.

[0085] Perform weighted averaging based on the weights, and select the category with the highest probability after weighted averaging as the final sorting result. The calculation formula is as follows:

[0086]

[0087] in, is the entropy value of the m-th model for the current sample, M is the total number of classification models, k represents the k-th classification model, is the weight of the mth model, is the entropy value of the kth model for the current sample.

[0088] Assume that the predicted probability of model m is , then the final weighted output probability is .

[0089] As in the example of this embodiment, assuming that the number of classifications is 3, including fat coal, granite and lignite gangue, the predicted probability distribution of the strain multi-classification model is: , the calculated entropy value H1 is 0.610; the predicted probability distribution of the heat multi-classification model is: , the calculated entropy value H1 is 1.029; the predicted probability distribution of the spectral multi-classification model is: , the calculated entropy value H1 is 0.500. Then the weights of the three multi-classification models are:

[0090] .

[0091] The final fusion prediction is Finally, the sorting probability of fat coal is 0.6767, the sorting probability of granite is 0.2016, and the sorting probability of lignite gangue is 0.1217. The sorting probability of fat coal is the largest and is identified as fat coal.

[0092] The sorted coal is transmitted to the small conveyor belt corresponding to coal. The small conveyor belt of coal includes 6 types, including fat coal, lignite, coking coal, gas fat coal, gas coal and anthracite. The small conveyor belt of rock includes 8 types, including granite, diabase, sandstone, limestone, quartz sandstone, quartzite, plagioclase and shale. The small conveyor belt of gangue includes 2 types, including lignite gangue and bituminous coal gangue.

[0093] The main methods for coal gangue sorting include manual sorting (time-consuming, labor-intensive, and inaccurate qualitative judgment), density difference (complicated process), and machine vision (complete and clear coal gangue images are required, but there is a lot of dust in the actual sorting process). In response to the above problems, the present invention selects strain, heat, and spectrum as three bases for sorting, and uses the differences in apparent characteristics, physical properties, and composition of different coal rocks for sorting. These three bases are used to achieve coal gangue sorting from different angles. In terms of performance characteristics, considering the influence of dust in the sorting environment, non-contact strain is used instead of image sorting; in terms of physical properties, coal and rock have differences in thermodynamic properties, molecular structure, and mechanical properties. The heat released during compression is different, and the physical property of released temperature is selected to characterize; in terms of composition, the composition is the key to the most essential difference between the reaction objects. Coal, rock, and gangue have different material compositions, and the composition is quantitatively analyzed by spectrum. In-depth decision-making integration of multiple sorting methods can significantly improve the sorting accuracy of coal, rock and gangue, solve the uncertainty problem caused by a single sorting method, and ensure the accuracy and reliability of the sorting results.

[0094] The strain, heat and spectrum are combined for sorting, and coal gangue is sorted from three different angles: apparent characteristics, physical properties and composition differences. These three methods are easy to implement and simple in process, and more practical than using density differences and manual sorting. The strain multi-classification model, heat multi-classification model and spectrum multi-classification model are constructed based on three classification models with different levels of characteristics, easy to implement and complementary, with better deep fusion effect. In addition, dynamic weighting of uncertainty based on entropy is performed during fusion, and the model weights are flexibly adjusted according to the actual performance of the three classification models, which are more adaptable and more accurate in classification.

[0095] Embodiment 2

[0096] This embodiment provides a coal-rock-gangue separation system based on multi-source data fusion analysis, including:

[0097] A sample collection and processing module, which is configured to: collect different types of coal, rock, and gangue, and process them into rock and coal blocks of the same size;

[0098] A strain multi-classification model building module is configured to: perform the same load loading test on the rock block and the coal block, monitor the strain of the coal block and the rock block, associate different strains with corresponding types of coal, rock block and gangue, and establish a strain multi-classification model;

[0099] A heat multi-classification model building module is configured to: monitor the heat released by the coal and rock blocks during the load loading process, associate the heat released under the same size and the same load loading conditions with the corresponding types of coal, rock, and gangue, and establish a heat multi-classification model;

[0100] A spectral multi-classification model building module is configured to: test and identify the spectral absorption peaks of different types of coal, rock, and gangue, associate different absorption peak features with corresponding types of coal, rock, and gangue, and establish a spectral multi-classification model;

[0101] The fusion sorting module is configured to sort and identify the mixture of coal, rock and gangue using the strain multi-classification model, the heat multi-classification model and the spectral multi-classification model respectively, and to fuse the sorting results of the strain multi-classification model, the heat multi-classification model and the spectral multi-classification model to determine the final sorting result.

[0102] Embodiment 3

[0103] The purpose of this embodiment is to provide a computing device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method of embodiment 1 when executing the program.

[0104] Embodiment 4

[0105] The purpose of this embodiment is to provide a computer-readable storage medium, a computer-readable storage medium having a computer program stored thereon, and when the program is executed by a processor, the steps of the method of embodiment 1 are performed.

[0106] The steps involved in the apparatus of the above embodiments 3 and 4 correspond to the method embodiment 1, and the specific implementation method can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0107] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0108] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0109] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A coal-rock-gangue separation method based on multi-source data fusion analysis, characterized in that: include: Collect different types of coal, rock, gangue and process them into rock and coal blocks of the same size; The same load loading test is performed on the rock block and the coal block, the strain of the coal block and the rock block is monitored, different strains are correlated with the corresponding types of coal, rock block and gangue, and a strain multi-classification model is established; Monitor the heat released by the coal and rock blocks during the load loading process, correlate the heat released under the same size and the same load loading conditions with the corresponding types of coal, rock, and gangue, and establish a heat multi-classification model; Test and identify the spectral absorption peaks of different types of coal, rock, and gangue, associate the different absorption peak characteristics with the corresponding types of coal, rock, and gangue, and establish a spectral multi-classification model; The mixture of coal, rock and gangue is sorted and identified using strain multi-classification model, heat multi-classification model and spectrum multi-classification model respectively, and the sorting results of strain multi-classification model, heat multi-classification model and spectrum multi-classification model are integrated to determine the final sorting result; The final sorting result is determined by integrating the sorting results of the strain multi-classification model, the heat multi-classification model and the spectrum multi-classification model: weighted averaging the three sorting results, selecting the category with the highest probability after the weighted average as the final sorting result, and transporting the sorted coal and rock separately; The weighting is a dynamic weighting of the uncertainty of the fusion model prediction, the uncertainty is quantified by the entropy value, and the weight is assigned according to the entropy value; The calculation formula of the entropy value is: in, is the category probability distribution predicted by the model, C is the number of categories, The model predicts the probability that the sample belongs to category i.

2. The coal-rock-gangue separation method based on multi-source data fusion analysis according to claim 1, characterized in that: The strain multi-classification model is trained by a neural network, the input parameters are the total strain of the test block and the strain values ​​corresponding to different moments, and the output parameters are the probabilities of the corresponding types of coal, rock or gangue.

3. The coal-rock-gangue separation method based on multi-source data fusion analysis according to claim 1, characterized in that: The heat multi-classification model is trained using a neural network, with input parameters being the total heat release, heat release rate and average temperature of test blocks of the same size and under the same load conditions, and output parameters being the probability of the corresponding coal, rock or gangue types.

4. The coal-rock-gangue separation method based on multi-source data fusion analysis according to claim 1, characterized in that: The spectral multi-classification model is trained using a neural network, the input parameters are the absorption peak characteristics of different test blocks, and the output parameters are the probabilities of the corresponding types of coal, rock or gangue.

5. The coal-rock-gangue separation method based on multi-source data fusion analysis according to claim 1, characterized in that: Perform weighted averaging based on the weights, and select the category with the highest probability after weighted averaging as the final sorting result. The calculation formula is as follows: in, is the entropy value of the m-th model for the current sample, M is the total number of classification models, k represents the k-th classification model, is the weight of the mth model, is the entropy value of the k-th model for the current sample; Assume that the predicted probability of model m is , then the final weighted output probability is .

6. The coal-rock-gangue separation system based on multi-source data fusion analysis is characterized by: include: A sample collection and processing module, which is configured to: collect different types of coal, rock, and gangue, and process them into rock and coal blocks of the same size; A strain multi-classification model building module is configured to: perform the same load loading test on the rock block and the coal block, monitor the strain of the coal block and the rock block, associate different strains with corresponding types of coal, rock block and gangue, and establish a strain multi-classification model; A heat multi-classification model building module is configured to: monitor the heat released by the coal and rock blocks during the load loading process, associate the heat released under the same size and the same load loading conditions with the corresponding types of coal, rock, and gangue, and establish a heat multi-classification model; A spectral multi-classification model building module is configured to: test and identify the spectral absorption peaks of different types of coal, rock, and gangue, associate different absorption peak features with corresponding types of coal, rock, and gangue, and establish a spectral multi-classification model; A fusion sorting module is configured to: sort and identify the mixture of coal, rock and gangue using the strain multi-classification model, the heat multi-classification model and the spectrum multi-classification model respectively, and to fuse the sorting results of the strain multi-classification model, the heat multi-classification model and the spectrum multi-classification model to determine the final sorting result; The final sorting result is determined by integrating the sorting results of the strain multi-classification model, the heat multi-classification model and the spectrum multi-classification model: weighted averaging the three sorting results, selecting the category with the highest probability after the weighted average as the final sorting result, and transporting the sorted coal and rock separately; The weighting is a dynamic weighting of the uncertainty of the fusion model prediction, the uncertainty is quantified by the entropy value, and the weight is assigned according to the entropy value; The calculation formula of the entropy value is: in, is the category probability distribution predicted by the model, C is the number of categories, The model predicts the probability that the sample belongs to category i.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the coal-rock-gangue separation method based on multi-source data fusion analysis as described in any one of claims 1 to 5 are implemented.

8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the coal-rock-gangue separation method based on multi-source data fusion analysis as described in any one of claims 1-5 are implemented.

Images