A coal colloid layer thickness measurement system and method based on big data

Abstract

The present invention relates to the technical field of thickness measurement, and to a system and method for measuring the thickness of a coal colloid layer based on big data, including: obtaining coal chemical characteristic data of a coal sample to be detected at different temperatures; inputting the coal chemical characteristic data into a machine learning model to obtain the thickness of the coal sample to be detected; wherein the training process of the machine learning model includes: collecting coal chemical characteristic data of multiple coal samples at different temperatures; setting multiple neighborhood radii, clustering multiple coal samples, and obtaining clustering clusters under each neighborhood radius; evaluating the clustering clusters using an improved CH index to obtain corresponding clustering quality indicators, and using the cluster with the largest clustering quality indicator and its corresponding thickness as a training set; inputting the training set into a machine learning model for training to obtain a trained machine learning model. The present invention can improve the detection efficiency of the colloid layer of coal samples and the accuracy of the detection results.

Description

Technical Field

[0001] The present invention relates to the technical field of thickness measurement, and more specifically, to a system and method for measuring the thickness of a coal colloid layer based on big data. Background Art

[0002] Coal colloid layer measurement technology is a key means to evaluate the quality of coking coal. Its core lies in examining the softening, melting and solidification behavior of coal at high temperature, especially the process of forming colloid. Traditional coal colloid layer measurement methods, such as the Gieseler plasticity tester method, involve placing the coal sample in a specific container and forming a series of isothermal layers under unilateral heating conditions to observe the softening, colloid formation and solidification process of the coal sample. The measurement focuses on the maximum thickness, expansion pressure and volume change curve of the colloid. These data are helpful to evaluate the colloid characteristics of coal and then infer its coking performance. In recent years, with the advancement of science and technology, the application of computer vision and image processing technology in coal colloid layer measurement has been significantly improved, realizing the automation of colloid layer index measurement, and improving accuracy and efficiency. Among them, measuring the thickness of the colloid layer of an unknown coal sample is a very critical part of coal colloid layer measurement. It involves the accurate prediction of the maximum colloid layer thickness and required temperature of an unknown coal sample, and further controlling the coking conditions. The accuracy of the prediction is crucial for the analysis of the colloid layer characteristics and coking performance of unknown coal samples.

[0003] The patent application document with application publication number CN116973541A discloses a method for detecting the maximum thickness of the bituminous coal colloid layer, which specifically involves heating the coal sample and using a probe to detect the upper and lower surface heights of the coal sample to obtain the maximum thickness of the bituminous coal colloid layer. In the above scheme, the coal sample is directly heated and measured directly. It should be noted that different coal samples need to be measured, which makes the detection efficiency of the coal sample colloid layer low, and when measuring, the measurement conditions are relatively high, such as the heating rate, coal sample particle size, pressure, coal cup material and the thermal properties of the refractory material of the coal cup outer furnace bricks, etc., which significantly affect the measurement results; that is, when the measurement conditions are inaccurate, the accuracy rate is not high.

[0004] Therefore, it is particularly important to improve the detection efficiency of the colloid layer of coal samples and the accuracy of the detection results. Summary of the invention

[0005] In order to solve the problem raised in the above background technology that when the measurement conditions are inaccurate, the accuracy is not high, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for measuring the thickness of a coal colloid layer based on big data, comprising: obtaining coal chemical characteristic data at different temperatures of a coal sample to be tested;

[0007] The coal chemical characteristic data is input into the machine learning model to obtain the thickness of the coal sample to be detected; wherein, the training process of the machine learning model includes: collecting the coal chemical characteristic data of multiple coal samples at different temperatures; setting multiple neighborhood radii, clustering the multiple coal samples based on the neighborhood radius, and obtaining cluster clusters under each neighborhood radius; using the improved CH index to evaluate the cluster clusters under each neighborhood radius, and obtain the corresponding cluster quality index, and taking the cluster with the largest cluster quality index and its corresponding thickness as the training set; inputting the training set into the machine learning model for training to obtain a trained machine learning model; wherein, the improved CH index includes weighting the trace of the discrete degree matrix within the cluster cluster using the feature weight; the feature weight is:

[0008] ;

[0009] In the formula, Indicates The characteristic weight of each coal sample, Indicates The variance of the coal sample, Indicates The standard deviation of the coal samples, represents the set of variances, represents the set of standard deviations, represents the adjustment parameters, It means to take the maximum value in the set. Indicates taking the minimum value in a set.

[0010] The above method can significantly improve the accuracy of coal gelatin layer thickness prediction, and is applicable to various types of coal samples, with broad application prospects.

[0011] Furthermore, the process of obtaining the multiple neighborhood radii is as follows:

[0012] According to the Minkowski algorithm, different parameters P are selected to calculate the distance values ​​between the coal chemical characteristic data of any two coal samples, and then draw a distance distribution diagram after the distance values ​​are processed in ascending order; the inflection point in the distance distribution diagram is obtained, and the distance value corresponding to the inflection point is used as the neighborhood radius, so as to obtain multiple neighborhood radii under different parameters P.

[0013] By selecting different parameters P through the Minkowski algorithm, the distance values ​​between the coal chemical characteristic data of any two coal samples are calculated respectively, and the distance values ​​are processed in ascending order to draw the distance distribution diagram. The inflection point in the distance distribution diagram is detected by the difference method, and multiple neighborhood radii are determined. In addition, by selecting the neighborhood radius under different parameters P, it is possible to dynamically adapt to the characteristics of different data sets, thereby selecting a more appropriate neighborhood radius under different parameters P, and ensuring the stability and reliability of the subsequent clustering effect.

[0014] Furthermore, the inflection point in the distance distribution graph is obtained by using a difference method, a curvature method or an interpolation method.

[0015] Furthermore, the algorithm for clustering multiple coal samples based on the neighborhood radius is a DBSCAN algorithm.

[0016] The DBSCAN algorithm is used for cluster analysis. DBSCAN is a density clustering algorithm that can effectively identify clusters of any shape and is particularly suitable for data sets with noise. By setting the neighborhood radius and the minimum number of sample points, the DBSCAN algorithm can adaptively adjust to the characteristics of different data sets and ensure the stability and robustness of the clustering effect. This efficient clustering method significantly improves the quality of clustering analysis of coal sample data and provides a reliable data basis for the training of machine learning models.

[0017] Furthermore, the method further includes adjusting the clustering quality index to obtain a final clustering quality index, specifically, calculating a second evaluation index, and obtaining a final clustering quality index based on the clustering quality index and the second evaluation index, wherein the formula is as follows:

[0018] ;

[0019] In the formula, score represents the final clustering quality index, represents the weight, represents the improved CH index, represents the second evaluation index, and log represents the logarithmic function.

[0020] By combining the improved CH index and the second evaluation index and introducing weight adjustment, the final clustering quality index is calculated. It can comprehensively consider the internal consistency and separation of the clustering results and evaluate the clustering quality in multiple dimensions, thereby providing a more comprehensive and flexible clustering performance evaluation and improving the accuracy and reliability of the clustering results.

[0021] Furthermore, the calculation of the second evaluation index adopts the Rand index algorithm, the adjusted Rand index algorithm, the adjusted mutual information score or the Dunn index algorithm.

[0022] Furthermore, the machine learning model includes: a linear regression model, a support vector machine model and a random forest model.

[0023] Provides a variety of machine learning model options, including linear regression models, support vector machine models, and random forest models. Different models have different advantages and disadvantages. By flexibly selecting the most suitable model, the accuracy and applicability of the prediction can be improved. Multi-model selection enhances the robustness of the method. By combining the prediction results of multiple models, the robustness and reliability of the prediction can be improved, avoiding the errors and instability that may be caused by a single model.

[0024] In a second aspect, the present invention provides a coal colloid layer thickness measurement system based on big data, comprising a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, any of the above-mentioned methods for measuring the thickness of a coal colloid layer based on big data is implemented.

[0025] The beneficial effects of the present invention are:

[0026] The present invention collects coal chemical characteristic data at different temperatures of coal samples, sets multiple neighborhood radii for clustering, and uses the improved CH index for evaluation. The cluster with the largest clustering quality index and its corresponding thickness are selected as the training set, and the machine learning model is trained to predict the thickness of the coal sample. This method effectively integrates big data analysis and machine learning technology, optimizes the clustering process through the Minkowski algorithm and the DBSCAN clustering algorithm, and combines the improved CH index with the second evaluation index to make the clustering quality evaluation more accurate, thereby being able to accurately predict the thickness of the coal sample and improve the measurement accuracy and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0028] Figure 1 is a flow chart schematically showing a method for measuring the thickness of a coal colloid layer based on big data according to an embodiment of the present invention;

[0029] Figure 2 is a grayscale diagram schematically showing a distance value at an inflection point in a distance distribution diagram according to an embodiment of the present invention;

[0030] Figure 3 It is a structural block diagram schematically showing a coal colloid layer thickness measurement system based on big data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0032] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0033] An embodiment of a method for measuring the thickness of a coal colloid layer based on big data.

[0034] like Figure 1 As shown, a flow chart of a method for measuring the thickness of a coal colloid layer based on big data in an embodiment of the present invention comprises the following steps:

[0035] S1: Obtain the coal chemical characteristic data of the coal samples to be tested at different temperatures and build a machine learning model.

[0036] Among them, the coal sample to be tested refers to a coal sample that needs to be subjected to chemical characteristic analysis and thickness prediction; the coal chemical characteristic data in this embodiment may be one or more of volatile matter, ash content and sulfur content, which can reflect the intrinsic characteristics of the coal sample at the corresponding temperature; specifically, the coal chemical characteristics at different temperatures can be collected by using chemical instruments such as volatile matter meters, ash content meters and sulfur content meters.

[0037] The temperature range of the coal chemical characteristic data obtained above is 250° C.-600° C. In this embodiment, the coal chemical characteristic data of the coal sample to be tested at different temperatures are collected at set temperature intervals, wherein the set temperature intervals may be 5° C., 10° C. or 15° C., etc.

[0038] Exemplarily, when the temperature interval is set to 5°C, the coal sample to be tested is first heated to 250°C and the coal chemical characteristic data is collected once, then heated to 255°C and collected once, and so on, until the temperature reaches 600°C, to obtain the coal chemical characteristic data of the coal sample to be tested at different temperatures.

[0039] The machine learning model represents an algorithmic model that learns from coal chemical characteristic data and predicts the thickness of the coal colloidal layer. Common models include linear regression models, support vector machine models, and random forest models. Different models have different advantages and disadvantages. By flexibly selecting the most suitable model, the accuracy and applicability of the prediction can be improved. Multi-model selection enhances the robustness of the method. By combining the prediction results of multiple models, the robustness and reliability of the prediction can be improved, avoiding the errors and instability that may be caused by a single model.

[0040] S2: setting multiple neighborhood radii, clustering the coal chemical characteristic data of multiple coal samples based on these neighborhood radii, and generating clusters under different radii.

[0041] The multiple coal samples in this embodiment come from different locations of the coal mine; the coal chemical characteristic data of the multiple coal samples can be stored historical data or obtained through measurement; and the thickness of the colloid layer of each coal sample is known.

[0042] It should be noted that, in this embodiment, when measuring each coal chemical characteristic data of multiple coal samples, the corresponding set temperature interval may be the same as or different from the temperature of the coal chemical characteristic data in step S1.

[0043] In this embodiment, the DBSCAN algorithm is used to cluster multiple coal samples.

[0044] DBSCAN is a density clustering algorithm that can effectively identify clusters of any shape and is particularly suitable for noisy data sets. Clustering is performed by setting the neighborhood radius and the minimum number of sample points.

[0045] In this embodiment, the process of obtaining multiple neighborhood radii is as follows:

[0046] According to the Minkowski algorithm, different P values ​​are selected to calculate the distance values ​​between the coal chemical characteristic data of any two coal samples respectively, and the distance values ​​are processed in ascending order to draw a distance distribution diagram; the inflection point in the distance distribution diagram is obtained, and the distance value corresponding to the inflection point is used as the neighborhood radius, so as to obtain multiple neighborhood radii under different P values.

[0047] like Figure 2 As shown, the grayscale diagram of the distance values ​​at the inflection points in the distance distribution diagram of an embodiment of the present invention includes inflection point 1, inflection point 2, inflection point 3 and inflection point 4, wherein the horizontal axis represents the sorted Minkowski distance value, each point represents the distance value between the coal chemical characteristic data of two coal samples, and the vertical axis represents the cumulative distribution number of Minkowski distance values.

[0048] Among them, the Minkowski algorithm is a generalized method for calculating the distance between two points. The parameter P represents the parameter in the Minkowski distance, which is used to adjust the way to calculate the distance. Common parameters P include 1, 2, and infinity, where 1 represents Manhattan distance, 2 represents Euclidean distance, and infinity represents Chebyshev distance. The distance distribution graph represents a graph that sorts and plots the calculated distance values, which is used to visualize the distance distribution between data points. The inflection point represents a point in the distance distribution graph where the curve changes significantly, usually representing a critical point of the distance value, which is used to determine the neighborhood radius.

[0049] The parameter P is selected by the Minkowski algorithm, and the distance values ​​between the coal chemical characteristic data of any two coal samples are calculated respectively. After these distance values ​​are processed in ascending order, the distance distribution diagram is drawn. The inflection points in the distance distribution diagram are detected by the difference method, and multiple neighborhood radii are determined. This process realizes the automation and precision of the neighborhood radius selection, ensures the accuracy and scientificity of clustering analysis, and avoids the subjective errors caused by manual selection. In addition, by selecting the neighborhood radius under different parameters P, it can dynamically adapt to the characteristics of different data sets and ensure the stability and reliability of the clustering effect.

[0050] In one embodiment, different parameters P will lead to different distance calculation methods and neighborhood shapes, affecting the size of the density radius, and then affecting the proximity between points. A larger parameter P is more likely to identify outliers as objects far away from other points, and a smaller parameter P can treat outliers as closer points, that is, normal. However, when parameter P tends to ∞, it is approximated to the Chebyshev distance, which only focuses on the dimension of the difference in multiple dimensions between two points, and cannot identify the influence of other dimensions, and a single dimension cannot measure the abnormal state of the data. Therefore, calculating different parameters P is extremely important for outlier processing of high-dimensional coal sample data.

[0051] In the above embodiment, multiple neighborhood radii are set, and DBSCAN clustering is performed on all coal chemical characteristic data corresponding to multiple coal samples based on these radii to obtain multiple cluster clusters corresponding to different neighborhood radii.

[0052] This embodiment combines Minkowski distance calculation and inflection point detection to automatically select the neighborhood radius, thereby accurately determining the optimal clustering parameters. This method can dynamically adapt to the characteristics of different data sets and improve the accuracy and scientificity of clustering analysis. In addition, the use of different distance calculation methods further enhances the model's ability to handle outliers in high-dimensional data, ensuring the stability and reliability of the clustering results. The DBSCAN algorithm can adaptively adjust to the characteristics of different data sets to ensure the stability and robustness of the clustering effect. This efficient clustering method significantly improves the quality of clustering analysis of coal sample data and provides a reliable data basis for the training of machine learning models.

[0053] S3: Use the improved CH index to evaluate the clusters under each neighborhood radius and calculate the corresponding clustering quality index.

[0054] Among them, the CH index, namely the Calinski-Harabasz index, is an indicator used to evaluate clustering quality. The improved CH index is based on the original CH index and adds feature weights to enhance the accuracy of the evaluation.

[0055] Specifically, the clustering quality index in this embodiment is obtained by using an improved CH index, specifically, by weighting the intra-cluster dispersion matrix of the cluster using feature weights.

[0056] The improved CH index is:

[0057] ;

[0058] In the formula, represents the trace of the inter-cluster dispersion matrix, that is, the sum of the squares of the distances from each cluster center to the average of all cluster centers, represents the trace of the intra-cluster dispersion matrix, that is, the sum of the squares of the distances from all coal chemical characteristic data within each cluster to the center of the cluster. Indicates the first Coal chemical characteristics data, Indicates the cluster The center of the cluster, Represents the total number of coal chemical characteristic data, represents the number of clusters, Indicates The characteristic weight of each coal sample.

[0059] The feature weights are:

[0060] ;

[0061] In the formula, Indicates The characteristic weight of each coal sample, Indicates The variance of the coal sample, Indicates The standard deviation of the coal samples, represents the set of variances, represents the set of standard deviations, represents the adjustment parameters, It means to take the maximum value in the set. Indicates taking the minimum value in a set.

[0062] This method can significantly improve the accuracy of predicting the thickness of coal colloid layer, and is applicable to various types of coal samples, with broad application prospects.

[0063] S4: Select the cluster with the largest clustering quality index and its corresponding thickness as the training set, input the training set into the machine learning model for training, and obtain a trained model.

[0064] The thickness of the coal colloid layer is obtained from the thickness database corresponding to the historically stored coal chemical characteristic data.

[0065] In order to more accurately evaluate the clustering quality, the second evaluation index is further calculated, such as the Dunn index, which comprehensively measures the distance between clusters and the distance within clusters. The steps to obtain the final clustering quality index are as follows: calculate the second evaluation index, and obtain the final clustering quality index based on the clustering quality index and the second evaluation index. The formula is as follows:

[0066] ;

[0067] In the formula, score represents the final clustering quality index, represents the weight, represents the improved CH index, represents the second evaluation index, and log represents the logarithmic function.

[0068] By calculating the second evaluation index, such as the Dunn index, combined with the improved CH index, the clustering quality is comprehensively evaluated. The weighted calculation method is used to comprehensively consider the contribution of multiple evaluation indicators to ensure that the final clustering quality index is more scientific and reasonable. The comprehensive evaluation method combines the measurement of inter-cluster distance and intra-cluster distance, which can comprehensively evaluate the clustering effect, avoid the deviation that may be caused by a single indicator, and improve the accuracy and credibility of the evaluation results.

[0069] As other implementations for calculating the second evaluation index, the Rand index algorithm, the adjusted Rand index algorithm, or the adjusted mutual information score may also be used.

[0070] After determining the final clustering quality index, the cluster with the largest clustering quality index is selected as the optimal clustering cluster. The corresponding coal colloid layer thickness is extracted from the cluster to form a training set. This training set contains the coal chemical characteristics and corresponding coal colloid layer thickness of all samples in the cluster with the best clustering quality.

[0071] The embodiment integrates indicators of multiple different dimensions to carefully measure the internal compactness of clusters and the separation between clusters, which can more accurately reflect the quality of clustering and ensure the comprehensiveness and accuracy of the evaluation results.

[0072] Taking the support vector machine model as an example, the training process of the support vector machine is introduced as follows:

[0073] The optimal clusters obtained above and the thickness of each coal sample in the cluster are used as training sets; the coal chemical characteristic data of each coal sample at different temperatures are used as input, and the thickness is used as output to train the support vector machine to obtain a trained support vector machine model. During the training process, cross-validation and adjustment of hyperparameters are used to improve the generalization ability and prediction accuracy of the model, so that the model can accurately predict the thickness of the coal sample to be tested based on the input coal chemical characteristic data.

[0074] S5: Input the coal chemical characteristic data to be tested into the machine learning model to obtain the thickness of the coal sample to be tested.

[0075] Taking the trained support vector machine model as an example, the chemical characteristic data of the coal to be tested is input into the support vector machine model to obtain the thickness of the colloid layer of the coal sample to be tested.

[0076] The scheme of the present invention collects coal chemical characteristic data at different temperatures of coal samples, sets multiple neighborhood radii for clustering, and uses the improved CH index for evaluation, selects the cluster with the largest clustering quality index and its corresponding thickness as a training set, and trains a machine learning model to predict the thickness of the coal sample. The clustering process is optimized by the Minkowski algorithm and the DBSCAN clustering algorithm, and the improved CH index is combined with the second evaluation index to make the clustering quality evaluation more accurate, thereby being able to accurately predict the thickness of the coal sample, thereby improving the measurement accuracy and efficiency.

[0077] An embodiment of a coal colloid layer thickness measurement system based on big data:

[0078] like Figure 3 As shown, a structural block diagram of a coal colloid layer thickness measurement system based on big data in an embodiment of the present invention includes a processor and a memory.

[0079] The present invention also provides a coal colloid layer thickness measurement system based on big data. The system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a coal colloid layer thickness measurement method based on big data according to the present invention is implemented.

[0080] The coal colloid layer thickness measurement system based on big data also includes other components familiar to those skilled in the art, such as a communication interface, and its settings and functions are known in the art, so they will not be described in detail here.

[0081] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0082] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0083] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for measuring the thickness of a coal colloid layer based on big data, characterized in that: include: Obtain coal chemical characteristic data of the coal sample to be tested at different temperatures; The coal chemical characteristic data is input into the machine learning model to obtain the thickness of the coal sample to be tested; wherein the training process of the machine learning model includes: Collect coal chemical characteristic data of multiple coal samples at different temperatures; set multiple neighborhood radii, cluster the multiple coal samples based on the neighborhood radius, and obtain clusters under each neighborhood radius; The process of obtaining multiple neighborhood radii is as follows: According to the Minkowski algorithm, different parameters P are selected to calculate the distance values ​​between the coal chemical characteristic data of any two coal samples, and then the distance values ​​are processed in ascending order to draw a distance distribution diagram; Obtain the inflection point in the distance distribution graph, use the distance value corresponding to the inflection point as the neighborhood radius, and then obtain multiple neighborhood radii under different parameters P; The algorithm for clustering multiple coal samples based on neighborhood radius is the DBSCAN algorithm; The improved CH index is used to evaluate the clusters under each neighborhood radius to obtain the corresponding clustering quality index, and the cluster with the largest clustering quality index and its corresponding thickness are used as the training set; the training set is input into the machine learning model for training to obtain a trained machine learning model; The improved CH index includes weighting the trace of the dispersion matrix within the cluster using feature weights; the feature weights are: ； In the formula, Indicates The characteristic weight of each coal sample, Indicates The variance of the coal sample, Indicates The standard deviation of the coal samples, represents the set of variances, represents the set of standard deviations, represents the adjustment parameters, It means to take the maximum value in the set. Indicates taking the minimum value in a set; The method also includes the steps of adjusting the clustering quality index to obtain the final clustering quality index, specifically: The second evaluation index is calculated, and the final clustering quality index is obtained based on the clustering quality index and the second evaluation index. The formula is as follows: ； In the formula, score represents the final clustering quality index, represents the weight, represents the improved CH index, represents the second evaluation index, log represents the logarithmic function; The second evaluation index is calculated using the Rand index algorithm, the adjusted Rand index algorithm, the adjusted mutual information score or the Dunn index algorithm.

2. The method for measuring the thickness of a coal colloid layer based on big data according to claim 1, characterized in that: The coal chemical characteristic data includes at least one of volatile matter data, ash data and sulfur data.

3. The method for measuring the thickness of a coal colloid layer based on big data according to claim 1, characterized in that: The inflection point in the distance distribution graph is obtained by using a difference method, a curvature method or an interpolation method.

4. The method for measuring the thickness of a coal colloid layer based on big data according to claim 1, characterized in that: The machine learning models include: linear regression model, support vector machine model and random forest model.

5. A coal colloid layer thickness measurement system based on big data, characterized in that: It comprises a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a method for measuring the thickness of a coal colloid layer based on big data as described in any one of claims 1 to 4 is implemented.

Images