Raman spectrum-based method for rapidly discriminating and sorting hydrophilicity and hydrophobicity of coal rock components

Abstract

The invention discloses a Raman spectrum-based method for rapidly judging and sorting hydrophilicity and hydrophobicity of coal rock components, and relates to the technical field of coal processing and sorting. The method comprises the following steps: establishing a coal rock Raman spectrum characteristic-microscopic component-macroscopic contact angle associated database with a large sample size; a quantitative analysis model capable of directly and rapidly predicting the coal sample contact angle according to Raman spectrum data is trained through a machine learning algorithm; and finally, the model is integrated into separation equipment, lossless and rapid judgment on the hydrophilicity and hydrophobicity of the coal particles is achieved, and an execution mechanism is driven to complete accurate separation according to the lossless and rapid judgment. The method has the outstanding advantages of rapidness, no damage, high precision, online performance and the like.

Description

Technical Field

[0001] This invention relates to the field of coal processing and sorting technology, and in particular to a rapid method for determining and sorting the hydrophilicity and hydrophobicity of coal petrographic components based on Raman spectroscopy. Background Technology

[0002] Coal is China's primary energy source, and its clean and efficient utilization is crucial. In coal washing, flotation, and coal-water slurry preparation, the hydrophilicity / hydrophobicity of coal petrographic components is a key physicochemical property determining the effectiveness of these processes. Traditionally, the hydrophilicity / hydrophobicity of coal is measured primarily through contact angle measurements, including the seated drop method and the bubble trapping method. However, these methods have the following inherent drawbacks: Coal samples typically require cutting and polishing to create smooth sections, which damages the original morphology of the sample. The sample preparation and measurement processes are complex and cannot meet the demands for rapid detection in industrial production. Single-point measurements are insufficient to represent the average properties of a whole lump of coal or a large amount of coal powder, and the non-uniformity of coal petrographic components leads to significant fluctuations in point measurement results. Furthermore, existing technologies are difficult to integrate into continuous, high-speed industrial production lines.

[0003] Therefore, developing a method that can quickly, non-destructively, and online determine the hydrophilicity and hydrophobicity of coal and rock components and achieve automated sorting is of great theoretical and practical significance. Summary of the Invention

[0004] The purpose of this invention is to provide a rapid method for determining and sorting the hydrophilicity and hydrophobicity of coal and petrographic components based on Raman spectroscopy, aiming to solve or improve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, the present invention provides the following solution: A rapid method for determining and sorting the hydrophilicity and hydrophobicity of coal petrographic components based on Raman spectroscopy includes: A standardized sample library is constructed; the standardized sample library includes purified samples rich in specific microscopic components; the specific microscopic components include vitrinite, inertinite, and chitinite; Microscopic Raman measurements were performed on individual samples in the standardized sample library, and the corresponding macroscopic contact angle values ​​were calculated to generate data pairs consisting of Raman spectra and macroscopic contact angle values. The acquired Raman spectra were preprocessed and feature extracted, and the characteristics corresponding to the micro-components were determined. By integrating Raman spectroscopy, microscopic component spectral characteristics, and macroscopic contact angles, a correlated database is obtained; The associated database was supervised learning using a partial least squares regression model, and the number of latent variables in the model was optimized. The trained model was then used as the quantitative analysis model. The quantitative analysis model is deployed in the control equipment of the sorting production line, and the Raman spectrum acquired in real time is used to predict the predicted contact angle value. The predicted contact angle value is compared with the dynamic contact angle threshold to determine the hydrophilicity / hydrophobicity classification and generate the corresponding sorting decision instruction.

[0006] Optionally, the construction process of the standardized sample library is as follows: Coal samples were collected from the target area; the coal samples covered the complete coalification sequence from low-rank coal to high-rank coal. The low-rank coal included lignite, long-flame coal, non-caking coal, and weakly caking coal; the high-rank coal included anthracite, lean coal, coking coal, fat coal, 1 / 3 coking coal, lean coal, and gas-fat coal. The coal sample was screened using a heavy liquid separation method to obtain a purified sample rich in specific microscopic components. The purified sample was then cut and polished to obtain a smooth sheet sample of a set size.

[0007] Optionally, the step of performing micro Raman measurements on individual samples in the standardized sample library and calculating the corresponding macroscopic contact angle values ​​to generate data pairs consisting of Raman spectra and macroscopic contact angle values ​​specifically includes: Select the point to be measured under a microscope and collect the Raman spectrum of the point to be measured; Without moving the sample, the static contact angle of deionized water at the same test point was measured using a micro-droplet system and the droplet morphology was recorded by a high-speed camera. The corresponding contact angle value was calculated by fitting the Young-Laplace equation. Data pairs are generated based on the Raman spectra of each measurement point and the corresponding contact angle values.

[0008] Optionally, the preprocessing and feature extraction of the acquired Raman spectra, and the determination of the features corresponding to the micro-components, specifically include: The acquired Raman spectra are subjected to noise filtering, baseline correction, and vector normalization to obtain preprocessed spectral data; the noise filtering adopts Savitzky-Golay convolutional smoothing algorithm or wavelet transform for noise reduction; Quantitative characteristic parameters are extracted from the preprocessed spectral data and matched with microscopic components; the quantitative characteristic parameters include the intensity ratio of D peak to G peak, the half width at half maximum (WHM) of G peak to D peak, the peak position shift of D peak to G peak, and the area ratio of (S peak) / (G+D peak).

[0009] Optionally, the step of using a partial least squares regression model to perform supervised learning on the associated database and optimizing the number of latent variables in the model to determine the trained model as a quantitative analysis model specifically includes: The associated database is input into the partial least squares regression model for training. When the evaluation indicators are met, the number of latent variables in the model is optimized, and the trained model is determined to be a quantitative analysis model. The evaluation indicators include the coefficient of determination, root mean square error, and mean absolute error.

[0010] Optionally, the step of deploying the quantitative analysis model in the control equipment of the sorting line and predicting the predicted contact angle value based on the real-time acquired Raman spectrum specifically includes: On the sorting production line, an industrial-grade online Raman probe is installed and positioned directly in front of the detection area on the conveyor belt to collect scanning data for each coal and rock sample. The scanned data is preprocessed and features are extracted. The processed data is then input into the quantitative analysis model, which calculates and outputs the predicted contact angle value.

[0011] Optionally, comparing the predicted contact angle value with the dynamic contact angle threshold to determine the hydrophilicity / hydrophobicity classification and generate the corresponding sorting decision instruction specifically includes: A dynamic contact angle threshold is set, the predicted contact angle value is compared with the dynamic contact angle threshold, and a sorting decision instruction is generated based on the comparison result; the dynamic contact angle threshold is a threshold that is dynamically updated according to the industrial scenario; the industrial scenario includes the pre-flotation waste removal scenario and the clean coal sorting scenario; The sorting decision command is sent to the corresponding actuator for hydrophilic and hydrophobic coal sorting; the actuator is an array of high-pressure pneumatic nozzles.

[0012] Optionally, the dynamic contact angle threshold specifically includes: In the pre-flotation waste disposal scenario, a single threshold is set, and the coal is screened according to the single threshold. Coal with a value less than the single threshold is identified as gangue with strong hydrophilicity or extremely poor quality coal and is directly discarded. In the clean coal sorting scenario, a dual threshold is set, and three intervals are divided according to the dual threshold: "high-quality clean coal" with high hydrophobicity, "ordinary clean coal" with medium hydrophobicity, and "tailings" with hydrophilicity.

[0013] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects: This invention discloses a rapid method for identifying and sorting the hydrophilicity and hydrophobicity of coal petrographic components based on Raman spectroscopy. The method includes constructing a standardized sample library; the standardized sample library includes purified samples rich in specific microscopic components; the specific microscopic components include vitrinite, inertinite, and chitinite; performing microscopic Raman measurements on individual samples in the standardized sample library and calculating the corresponding macroscopic contact angle values ​​to generate data pairs composed of Raman spectra and macroscopic contact angle values; preprocessing and extracting features from the acquired Raman spectra and determining the features corresponding to the microscopic components; integrating the Raman spectra, microscopic component spectral features, and macroscopic contact angles to obtain a correlation database; using a partial least squares regression model to supervise learning the correlation database and optimizing the number of latent variables in the model, determining the trained model as a quantitative analysis model; deploying the quantitative analysis model in the control equipment of the sorting line and predicting the real-time acquired Raman spectra to obtain predicted contact angle values; comparing the predicted contact angle values ​​with dynamic contact angle thresholds to determine the hydrophilicity / hydrophobicity classification and generate corresponding sorting decision instructions.

[0014] This invention establishes a large-scale database linking "coal and petrographic Raman spectral characteristics - microscopic components - macroscopic contact angles"; it then trains a quantitative analysis model using machine learning algorithms to directly and rapidly predict the contact angle of coal samples based on Raman spectral data; finally, this model is integrated into a sorting device to achieve non-destructive and rapid discrimination of the hydrophilicity and hydrophobicity of coal particles, thereby driving the actuator to complete precise sorting. This method has outstanding advantages such as speed, non-destructive nature, high precision, and online operation capability. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of the steps of the rapid identification and sorting method for hydrophilicity and hydrophobicity of coal and petrographic components based on Raman spectroscopy in this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] The purpose of this invention is to provide a rapid method for determining and sorting the hydrophilicity and hydrophobicity of coal and petrographic components based on Raman spectroscopy, aiming to solve or improve at least one of the above-mentioned technical problems.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1 As shown, this invention provides a rapid method for determining and sorting the hydrophilicity and hydrophobicity of coal and petrographic components based on Raman spectroscopy, comprising: Step 100: Construct a standardized sample library; the standardized sample library includes purified samples rich in specific microscopic components; the specific microscopic components include vitrinite, inertinite, and chitinite. Specifically: The process of constructing the standardized sample library is as follows: Coal samples were collected from the target area; the coal samples covered the complete coalification sequence from low-rank coal to high-rank coal; the low-rank coal included lignite, long-flame coal, non-caking coal and weakly caking coal; the high-rank coal included anthracite, lean coal, coking coal, fat coal, 1 / 3 coking coal, lean coal and gas-fat coal; the target area included major coal-forming ages in China (such as Carboniferous-Permian, Jurassic) and different geographical-tectonic regions (such as Shanxi, Ordos, Xinjiang).

[0021] The coal sample was screened using a heavy liquid separation method to obtain a purified sample rich in specific microscopic components in order to obtain "pure" data of the component end-members. The purified sample was then cut and polished to obtain a smooth sheet sample of a set size.

[0022] Step 200: Perform micro-Raman measurements on each individual sample in the standardized sample library and calculate the corresponding macroscopic contact angle values ​​to generate data pairs consisting of Raman spectra and macroscopic contact angle values. This is crucial for ensuring the accuracy of data correlation. A confocal micro-Raman spectroscopy system equipped with an optical microscope and a precision platform is used. Specifically: First, select a representative point to be tested under a microscope (such as vitrinite bands or filamentous fragments).

[0023] Then, high-quality Raman spectra are acquired at this point. Spectral acquisition parameters need to be standardized: typically, a 532 nm or 785 nm laser wavelength is used to avoid fluorescence interference and ensure high spectral signal intensity and excellent signal-to-noise ratio.

[0024] Next, without moving the sample, the static contact angle of deionized water at the same measurement point was precisely measured using a micro-dropping system based on the seated drop method. The droplet morphology was recorded using a high-speed camera, and the precise contact angle value was calculated by fitting the Young-Laplace equation.

[0025] This process is repeated dozens to hundreds of times on each sample to obtain a sufficient number of statistically significant data pairs.

[0026] Step 300: Preprocess and extract features from the acquired Raman spectra, and determine the characteristics corresponding to the microscopic components. Specifically: The acquired Raman spectra were subjected to noise filtering, baseline correction, and vector normalization to obtain preprocessed spectral data. Noise filtering employed the Savitzky-Golay convolutional smoothing algorithm or wavelet transform for denoising. Baseline correction used adaptive iterative reweighted penalized least squares (airPLS) or polynomial fitting to eliminate fluorescence background. Vector normalization was used to eliminate spectral intensity fluctuations caused by sample surface roughness, focusing differences, etc.

[0027] Quantitative characteristic parameters are extracted from the preprocessed spectral data and matched with microscopic components. These quantitative characteristic parameters include the intensity ratio of the D peak to the G peak, the full width at half maximum (FWHM) of the G peak and the D peak, the peak position shift of the D peak and the G peak, and the area ratio of the (S peak) to the (G+D peak). The intensity ratio of the D peak to the G peak reflects the ratio of disorder / defect degree to ordered graphitization degree in the aromatic ring structure. The FWHM of the G peak and the D peak represents the degree of order in the carbon structure; a wider FWHM indicates a lower degree of order. The peak position shift of the D peak and the G peak reflects the size and stress state of the aromatic lamellars. The (S peak) to (G+D peak) area ratio is used because the S peak is often associated with alkyl-aryl carbon chains or sp... 3 It is related to hybrid carbon, and its relative strength is related to the aliphatic content, which significantly affects hydrophobicity.

[0028] Step 400: Integrate Raman spectra, microscopic component spectral characteristics, and macroscopic contact angles to obtain a correlated database. Specifically: Integrate all data pairs obtained in the above steps, including Raman spectra, extracted feature parameters, and corresponding precise contact angle values, into a structured database.

[0029] Step 500: Supervised learning of the associated database is performed using a partial least squares regression model, and the number of latent variables in the model is optimized. The trained model is then determined as the quantitative analysis model. Specifically: The database is randomly divided into a training set (e.g., 70%), a validation set (e.g., 15%), and a test set (e.g., 15%). Then, a partial least squares regression model is used to supervise the learning of the associated database. This model can effectively handle the high collinearity problem of spectral data and maximize the covariance with the target variable (contact angle) while reducing dimensionality, making it very suitable for this type of task.

[0030] The model is trained using the training set data, and the hyperparameters (number of latent variables) of the model are optimized using the validation set to prevent overfitting and pursue the best generalization ability of the model.

[0031] The final model was evaluated using an independent test set. Key evaluation metrics included the coefficient of determination (R²). 2 The root mean square error (RMSE) and mean absolute error (MAE) are also important metrics. A reliable model should have a high R-squared value. 2 Values ​​(e.g., >0.85) and low RMSE are associated with MAE.

[0032] Step 600: Deploy the quantitative analysis model in the control equipment of the sorting line, and predict the contact angle value by analyzing the real-time acquired Raman spectra. Specifically: An industrial-grade online Raman probe is installed on the sorting line. This probe must be dustproof, waterproof, and vibration-resistant, and integrate autofocus and laser safety features. The probe is positioned directly over the detection area on the conveyor belt to ensure that every passing coal particle is scanned quickly. It is used in conjunction with a high-speed linear array camera or photoelectric sensor to precisely trigger the spectral acquisition.

[0033] After the online system acquires the Raman spectrum of the coal particle, it first applies the same preprocessing procedure and feature extraction algorithm as in the first stage to process the spectrum. Then, the processed feature data is input into a pre-trained quantitative analysis model deployed in an industrial computer. The model performs calculations within milliseconds and outputs the predicted contact angle value for the coal particle.

[0034] Step 700: Compare the predicted contact angle value with the dynamic contact angle threshold to determine the hydrophilicity / hydrophobicity classification and generate the corresponding sorting decision instruction.

[0035] The control system presets one or more dynamic contact angle thresholds according to specific process requirements.

[0036] In the pre-disposal scenario before flotation, a single threshold θ = 50° is set. Coal particles with a predicted contact angle < 50° are identified as highly hydrophilic gangue or extremely low-quality coal and are directly discarded as tailings.

[0037] In the clean coal sorting scenario, dual thresholds can be set to classify coal particles into highly hydrophobic "high-quality clean coal", moderately hydrophobic "ordinary clean coal", and hydrophilic "tailings".

[0038] The system compares the predicted value with the threshold and immediately generates sorting decision instructions (such as "keep" or "discard", or "category A" or "category B").

[0039] The sorting decision command is sent to a high-speed actuator, typically an array of high-pressure pneumatic nozzles (air guns). The system precisely calculates the impact delay time based on the speed and position of the coal particles. When the target coal particle moves directly below the air gun, the corresponding air gun instantly activates, using compressed air to blow it off its original trajectory and into the target collection chamber, thus achieving efficient and automatic separation of hydrophilic and hydrophobic coal particles.

[0040] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0041] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A rapid method for determining and sorting the hydrophilicity and hydrophobicity of coal petrographic components based on Raman spectroscopy, characterized in that, include: Construct a standardized sample library; The standardized sample library includes purified samples rich in specific microscopic components; these specific microscopic components include vitrinite, inertinite, and chitinite. Microscopic Raman measurements were performed on individual samples in the standardized sample library, and the corresponding macroscopic contact angle values ​​were calculated to generate data pairs consisting of Raman spectra and macroscopic contact angle values. The acquired Raman spectra were preprocessed and feature extracted, and the characteristics corresponding to the micro-components were determined. By integrating Raman spectroscopy, microscopic component spectral characteristics, and macroscopic contact angles, a correlated database is obtained; The associated database was supervised learning using a partial least squares regression model, and the number of latent variables in the model was optimized. The trained model was then used as the quantitative analysis model. The quantitative analysis model is deployed in the control equipment of the sorting production line, and the Raman spectrum acquired in real time is used to predict the predicted contact angle value. The predicted contact angle value is compared with the dynamic contact angle threshold to determine the hydrophilicity / hydrophobicity classification and generate the corresponding sorting decision instruction.

2. The rapid identification and sorting method for hydrophilicity and hydrophobicity of coal and petrographic components based on Raman spectroscopy according to claim 1, characterized in that, The process of constructing the standardized sample library is as follows: Coal samples were collected from the target area; the coal samples covered the complete coalification sequence from low-rank coal to high-rank coal; the low-rank coal included lignite, long-flame coal, non-caking coal and weakly caking coal; the high-rank coal included anthracite, lean coal, coking coal, fat coal, 1 / 3 coking coal, lean coal and gas-fat coal. The coal sample was screened using a heavy liquid separation method to obtain a purified sample rich in specific microscopic components. The purified sample was then cut and polished to obtain a smooth sheet sample of a set size.

3. The rapid identification and sorting method for hydrophilicity and hydrophobicity of coal and petrographic components based on Raman spectroscopy according to claim 1, characterized in that, The step of performing microscopic Raman measurements on individual samples in the standardized sample library and calculating the corresponding macroscopic contact angle values ​​to generate data pairs consisting of Raman spectra and macroscopic contact angle values ​​specifically includes: Select the point to be measured under a microscope and collect the Raman spectrum of the point to be measured; Without moving the sample, the static contact angle of deionized water at the same test point was measured using a micro-droplet system and the droplet morphology was recorded by a high-speed camera. The corresponding contact angle value was calculated by fitting the Young-Laplace equation. Data pairs are generated based on the Raman spectra of each measurement point and the corresponding contact angle values.

4. The rapid identification and sorting method for hydrophilicity and hydrophobicity of coal and petrographic components based on Raman spectroscopy according to claim 1, characterized in that, The preprocessing and feature extraction of the acquired Raman spectra, and the determination of the characteristics corresponding to the micro-components, specifically include: The acquired Raman spectra are subjected to noise filtering, baseline correction, and vector normalization to obtain preprocessed spectral data; the noise filtering adopts Savitzky-Golay convolutional smoothing algorithm or wavelet transform for noise reduction; Quantitative characteristic parameters are extracted from the preprocessed spectral data and matched with microscopic components; the quantitative characteristic parameters include the intensity ratio of D peak to G peak, the half width at half maximum (WHM) of G peak to D peak, the peak position shift of D peak to G peak, and the area ratio of (S peak) / (G+D peak).

5. The rapid identification and sorting method for hydrophilicity and hydrophobicity of coal petrographic components based on Raman spectroscopy according to claim 1, characterized in that, The process involves using a partial least squares regression model to supervise learning the associated database, optimizing the number of latent variables in the model, and determining the trained model as a quantitative analysis model. Specifically, this includes: The associated database is input into the partial least squares regression model for training. When the evaluation indicators are met, the number of latent variables in the model is optimized, and the trained model is determined to be a quantitative analysis model. The evaluation indicators include the coefficient of determination, root mean square error, and mean absolute error.

6. The rapid identification and sorting method for hydrophilicity and hydrophobicity of coal and petrographic components based on Raman spectroscopy according to claim 1, characterized in that, The step of deploying the quantitative analysis model in the control equipment of the sorting production line and predicting the predicted contact angle value based on the real-time acquired Raman spectra specifically includes: On the sorting production line, an industrial-grade online Raman probe is installed and positioned directly in front of the detection area on the conveyor belt to collect scanning data for each coal and rock sample. The scanned data is preprocessed and features are extracted. The processed data is then input into the quantitative analysis model, which calculates and outputs the predicted contact angle value.

7. The rapid identification and sorting method for hydrophilicity and hydrophobicity of coal and petrographic components based on Raman spectroscopy according to claim 1, characterized in that, The step of comparing the predicted contact angle value with the dynamic contact angle threshold to determine the hydrophilicity / hydrophobicity classification and generate corresponding sorting decision instructions specifically includes: A dynamic contact angle threshold is set, the predicted contact angle value is compared with the dynamic contact angle threshold, and a sorting decision instruction is generated based on the comparison result; the dynamic contact angle threshold is a threshold that is dynamically updated according to the industrial scenario; the industrial scenario includes the pre-flotation waste removal scenario and the clean coal sorting scenario; The sorting decision command is sent to the corresponding actuator for hydrophilic and hydrophobic coal sorting; the actuator is an array of high-pressure pneumatic nozzles.

8. The rapid identification and sorting method for hydrophilicity and hydrophobicity of coal petrographic components based on Raman spectroscopy according to claim 7, characterized in that, The dynamic contact angle threshold specifically includes: In the pre-flotation waste disposal scenario, a single threshold is set, and the coal is screened according to the single threshold. Coal with a value less than the single threshold is identified as gangue with strong hydrophilicity or extremely poor quality coal and is directly discarded. In the clean coal sorting scenario, a dual threshold is set, and three intervals are divided according to the dual threshold: "high-quality clean coal" with high hydrophobicity, "ordinary clean coal" with medium hydrophobicity, and "tailings" with hydrophilicity.

Images