A coal quality online detection method and system based on fusion of raman and infrared spectra

Abstract

The application discloses a coal quality online detection method and system based on Raman and infrared spectrum fusion, relates to the field of analytical chemistry and process detection, and comprises the following steps: synchronously or asynchronously collecting Raman spectrum and infrared absorption spectrum of a coal sample to be detected; extracting coal quality sensitive features and full spectrum features of the coal sample to be detected according to the Raman spectrum and the infrared absorption spectrum; splicing the coal quality sensitive features and the full spectrum features to obtain a fusion feature vector; and determining industrial analysis parameters of the coal sample to be detected by using a pre-trained machine learning model according to the fusion feature vector, so as to determine the quality of the coal sample to be detected. The application realizes rapid, accurate and online nondestructive detection of the industrial analysis parameters of coal by combining machine learning with Raman and infrared spectrum fusion and feature splicing, and significantly improves the detection efficiency, accuracy and intelligent level of coal utilization.

Description

Technical Field

[0001] This application relates to the fields of analytical chemistry and process detection, and in particular to an online coal quality detection method and system based on the fusion of Raman and infrared spectroscopy. Background Technology

[0002] Currently, coal quality testing mainly relies on traditional chemical analysis methods (such as industrial analysis and elemental analysis) and single rapid spectroscopic techniques. Traditional chemical analysis methods provide accurate results, but are cumbersome, time-consuming (ranging from hours to days), have inherent time lags, and damage samples, failing to meet the needs of real-time optimization and control in production processes. Single near-infrared spectroscopy, while rapid, relies on numerous standards and lacks sensitivity to key coal structures (such as aromatic rings and mineral crystal phases), and its robustness is easily affected by changes in coal type. Single Raman spectroscopy is sensitive to the carbon skeleton structure, but the signal is weak, susceptible to fluorescence interference, and insensitive to hydrogen-containing functional groups (such as -OH). Offline infrared spectroscopy provides rich information on functional groups, but typically requires sample preparation (KBr pelleting), making online measurement difficult.

[0003] In summary, existing single-spectral techniques are unable to comprehensively and accurately acquire the chemical structure information and multiple industrial analytical parameters of coal without sample preparation and under rapid online conditions. Summary of the Invention

[0004] The purpose of this application is to provide a coal quality online detection method and system based on the fusion of Raman and infrared spectroscopy, which can realize rapid, accurate, online non-destructive detection of coal industry analytical parameters, and significantly improve detection efficiency, accuracy and the level of intelligent coal utilization.

[0005] To achieve the above objectives, this application provides the following solution: Firstly, this application provides an online coal quality detection method based on the fusion of Raman and infrared spectroscopy, comprising: The Raman and infrared absorption spectra of the coal sample to be tested are acquired synchronously or asynchronously. Based on the Raman spectrum and the infrared absorption spectrum, extract the coal quality sensitive characteristics and full-spectral characteristics of the coal sample to be tested; The coal quality sensitive features are concatenated with the full-spectrum features to obtain a fused feature vector; Based on the fused feature vector, a pre-trained machine learning model is used to determine the industrial analysis parameters of the coal sample to be tested, thereby determining the quality of the coal sample.

[0006] In one embodiment, before synchronously or asynchronously acquiring the Raman spectrum and infrared absorption spectrum of the coal sample to be tested, the method further includes: shaping the measurement surface of the coal sample to be tested so that the spectral illumination surface is focused and flat.

[0007] In one embodiment, before extracting the coal quality sensitive features and full-spectrum features of the coal sample to be tested based on the Raman spectrum and the infrared absorption spectrum, the method further includes: performing smoothing and noise reduction, baseline correction and standardization processing on the Raman spectrum and the infrared absorption spectrum, respectively.

[0008] In one embodiment, based on the Raman spectrum and the infrared absorption spectrum, the coal quality sensitive features and full-spectrum features of the coal sample to be tested are extracted, including: extracting the spectral peak parameters, spectral statistics and absorbance of the Raman spectrum and the infrared absorption spectrum respectively to obtain the coal quality sensitive features; performing principal component analysis on the Raman spectrum and the infrared absorption spectrum respectively, retaining the first n principal components to obtain the full-spectrum features.

[0009] In one embodiment, the spectral peak parameters include peak height, peak position, peak intensity, peak area, and peak ratio for a specific spectral band; the spectral statistics include mean, variance, skewness, and kurtosis.

[0010] In one embodiment, the machine learning model includes a first-level analysis model and a second-level analysis model; the first-level analysis model is pre-trained using a first training sample set, and the second-level analysis model is pre-trained using a second training sample set; the training samples in the first training sample set include fused feature samples and corresponding preliminary industrial analysis parameters; the training samples in the second training sample set include fused feature set samples and corresponding industrial analysis parameters; based on the fused feature vector, the pre-trained machine learning model is used to determine the industrial analysis parameters of the coal sample to be tested, including: based on the fused feature vector, using the first-level analysis model to determine the preliminary industrial analysis parameters of the coal sample to be tested; fusing the preliminary industrial analysis parameters with the fused feature vector to obtain a fused feature set; and based on the fused feature set, using the second-level analysis model to determine the industrial analysis parameters of the coal sample to be tested.

[0011] In one embodiment, the first-level analysis model is a support vector machine or a random forest model; the second-level analysis model is a tree ensemble model.

[0012] In one embodiment, the machine learning model includes a Raman analysis model and an infrared analysis model; the Raman analysis model is pre-trained using a third training sample set, and the infrared analysis model is pre-trained using a fourth training sample set; the training samples in the third training sample set include Raman spectral feature samples and corresponding Raman industrial analysis parameters; the training samples in the fourth training sample set include infrared spectral feature samples and corresponding infrared industrial analysis parameters; based on the fused feature vector, the pre-trained machine learning model is used to determine the industrial analysis parameters of the coal sample to be tested, including: extracting Raman spectral features and infrared spectral features from the fused feature vector respectively; based on the Raman spectral features, using the Raman analysis model to determine the Raman industrial analysis parameters; based on the infrared spectral features, using the infrared analysis model to determine the infrared industrial analysis parameters; and fusing the Raman industrial analysis parameters and the infrared industrial analysis parameters to obtain the industrial analysis parameters of the coal sample to be tested.

[0013] Secondly, this application provides an online coal quality detection system based on the fusion of Raman and infrared spectroscopy, comprising: An integrated sampling probe is used to synchronously or asynchronously acquire Raman and infrared absorption spectra of coal samples to be tested; The spectral analysis module is used to extract the coal quality sensitive features and full-spectral features of the coal sample to be tested based on the Raman spectrum and the infrared absorption spectrum; to concatenate the coal quality sensitive features and the full-spectral features to obtain a fused feature vector; and to determine the industrial analysis parameters of the coal sample to be tested based on the fused feature vector using a pre-trained machine learning model, so as to determine the quality of the coal sample to be tested.

[0014] In one embodiment, the integrated sampling probe includes a laser, a halogen lamp, a Raman spectrometer, and an infrared spectrometer; the laser and the halogen lamp are used to mix and emit excitation light onto the measurement surface of the coal sample to be tested through an arc-shaped reflective surface; the Raman spectrometer is used to collect the Raman spectrum of the coal sample to be tested; and the infrared spectrometer is used to collect the infrared absorption spectrum of the coal sample to be tested.

[0015] According to the specific embodiments provided in this application, this application has the following technical effects: By synchronously or asynchronously acquiring Raman and infrared absorption spectra, and utilizing their complementary responses to the aromatic structure of the carbon skeleton and oxygen- and hydrogen-containing polar functional groups, more comprehensive and multi-dimensional coal quality information is obtained, effectively overcoming the limitations of overlapping features or missing information in single-spectral techniques. At the feature extraction level, full-spectral features are spliced ​​and fused with coal-sensitive features, preserving complete spectral information while highlighting key parameters. The constructed fused feature vector has higher discriminative power and robustness. Combined with machine learning models, rapid and accurate prediction of industrial analytical parameters (moisture, ash, volatile matter, fixed carbon) is achieved. This method requires no complex sample pretreatment, and the detection cycle is shortened from several hours in traditional chemical methods to seconds, fully meeting the online real-time quality control requirements of production processes such as belt conveying and furnace combustion. Simultaneously, non-contact measurement avoids human error and chemical reagent consumption, combining environmental friendliness and economy. Practical applications show that the prediction accuracy and stability of the fusion model are significantly better than those of the single spectral scheme. It can provide reliable data support for intelligent coal sorting, coal blending optimization and combustion condition adjustment, effectively improving coal utilization efficiency and cleanliness. Significant technological breakthroughs have been achieved in terms of detection speed, accuracy, information comprehensiveness and online adaptability. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic flowchart of an online coal quality detection method based on the fusion of Raman and infrared spectroscopy, provided as an embodiment of this application.

[0018] Figure 2 This is a framework diagram of an online coal quality detection method based on the fusion of Raman and infrared spectroscopy, provided as an embodiment of this application.

[0019] Figure 3 A processing framework diagram of Raman spectroscopy and infrared excitation spectroscopy is provided for one embodiment of this application.

[0020] Figure 4 This is a schematic diagram of a collinear optical path in one embodiment of this application. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] In one exemplary embodiment, an online coal quality detection method based on Raman and infrared spectroscopy fusion is provided. This method is suitable for rapid, online detection and structural characterization of solid fuels such as coal and biomass. The method is executed by a computer device, specifically a terminal or server, or both. In this embodiment, as shown... Figure 1 and Figure 2 As shown, the method includes steps 101 to 106.

[0024] Step 101: Shape the measuring surface of the coal sample to be tested to ensure the spectral illumination surface is focused and flat. In this application, the coal sample does not require pressing or grinding; only simple flattening is needed. The measuring surface shaping is achieved by designing a pressure roller structure or a pressure-bearing surface to shape the coal powder surface. When shaping the coal powder surface using a pressure roller structure, the coal powder is deposited onto the concave surface of the pressure roller and then shaped on the roller. When shaping the coal powder surface using a pressure-bearing surface, the coal powder is fed into a fixed-shape mold and pressed with a flat surface. The shaped coal sample is then placed into the sampling probe box.

[0025] Step 102: Synchronously or asynchronously acquire the Raman spectrum and infrared absorption spectrum of the coal sample to be tested.

[0026] Step 103: Perform smoothing, noise reduction, baseline correction, and normalization on the Raman spectrum and the infrared absorption spectrum, respectively.

[0027] Specifically, the Raman and infrared absorption spectra are first preprocessed to eliminate measurement noise and baseline drift. Then, Savitzky-Golay filtering and moving average methods are used to smooth and denoise the Raman and infrared absorption spectra, reducing random noise. Next, polynomial fitting and iterative fitting methods are employed for baseline correction to subtract baseline drift caused by instrument background or scattering. Finally, the spectra are standardized or normalized to eliminate the influence of sample size or optical path differences; common methods include vector normalization and standard normalization.

[0028] like Figure 3As shown, the coal sample to be tested is first transmitted to the sampling probe through the coal powder pipeline. Raman spectra and infrared excitation spectra are collected by Raman spectrometer and infrared spectrometer, respectively, and stored in Raman spectral database and infrared spectral database. Raman spectral analysis and infrared spectral analysis are performed respectively, and then coal quality data correction is performed to output the signal.

[0029] Step 104: Extract the coal quality sensitive features and full-spectrum features of the coal sample to be tested based on the Raman spectrum and the infrared absorption spectrum.

[0030] In a specific application example, the peak parameters, spectral statistics, and absorbance of the Raman spectrum and the infrared absorption spectrum are extracted respectively to obtain the coal quality sensitive characteristic F. t =[f1,f2,...,f m ]. Among them, F t Let f1~f be the coal quality sensitive characteristics at time t. m denoted as spectral peak parameter, and m as the number of spectral peak parameters.

[0031] Specifically, for spectra with distinct peaks, peak parameters are extracted using peak-finding algorithms. Peak parameters include peak height, peak position, peak intensity, peak area, peak ratio (e.g., the intensity ratio of the D peak to the G peak in Raman spectroscopy, or the absorption ratio of aromatic CH to aliphatic CH in infrared spectroscopy), and full width at half maximum (FWHM). Spectral statistics include mean, variance, skewness, and kurtosis. When features are not obvious or correlated with multiple components, principal component analysis or partial least squares methods can be used to reduce the dimensionality of high-dimensional spectral data and extract a few comprehensive variables that best represent the differences or attributes of the samples.

[0032] Principal component analysis (PCA) was performed on the Raman spectrum and the infrared absorption spectrum respectively, and the first n principal components were retained to obtain the full spectrum characteristics.

[0033] Specifically, the PCA model is fitted only on the training set to determine the principal component directions. Using the trained PCA model, the full-spectral data from both the training and test sets are transformed to obtain the dimensionality-reduced principal component scores. Assuming the first n principal components are retained, each spectrum yields a score vector F. p =[p1,p2,...,p n As a full-spectrum feature. Among them, F p Let p1~p be the score vector. n The score is the score of the first n principal components.

[0034] Step 105: The coal quality sensitive features are spliced ​​with the full-spectrum features to obtain a fused feature vector.

[0035] Specifically, before splicing, the coal quality-sensitive features and full-spectrum features are standardized (e.g., Z-score standardization). The standardized coal quality-sensitive features and full-spectrum features are then spliced ​​along the feature dimension (i.e., column-wise concatenation), resulting in the final fused feature vector: F = [F...]. t ,F p The dimension of the merged feature vector after concatenation is m+n.

[0036] Step 106: Based on the fused feature vector, a pre-trained machine learning model is used to determine the industrial analysis parameters of the coal sample to be tested, so as to determine the quality of the coal sample to be tested.

[0037] In a specific application example, a cascaded modeling strategy is used to determine the industrial analysis parameters of the coal sample to be tested. The machine learning model includes a first-level analysis model and a second-level analysis model. The first-level analysis model is pre-trained using a first training sample set, and the second-level analysis model is pre-trained using a second training sample set. The training samples in the first training sample set include fused feature samples and corresponding preliminary industrial analysis parameters. The training samples in the second training sample set include fused feature set samples and corresponding industrial analysis parameters. The first-level analysis model is a Support Vector Machine (SVM) or Random Forest model. The second-level analysis model is a tree ensemble model.

[0038] In this application example, step 106 includes: determining preliminary industrial analysis parameters of the coal sample to be tested using the first-level analysis model based on the fused feature vector; fusing the preliminary industrial analysis parameters with the fused feature vector to obtain a fused feature set; and determining the industrial analysis parameters of the coal sample to be tested using the second-level analysis model based on the fused feature set.

[0039] Specifically, firstly, an SVM or random forest model is established using fused feature vectors to predict the preliminary industrial analysis parameters of the coal sample to be tested. The industrial analysis and elemental analysis indicators of coal are used as input features, and the key coal quality characteristics to be predicted (such as calorific value, ash fusion point, etc.) are used as output targets. The first-level analysis model is constructed through data preprocessing, model training and optimization. The relevant parameters of the first-level analysis model are set as follows: (1) Kernel function selection: Coal quality data often have nonlinear relationships, so the radial basis kernel function is selected. (2) Key parameter optimization: Optimization is achieved through cross-validation grid search. (3) Penalty parameter C: Controls the tolerance of the model to misclassification. A large C value makes the model more complex and may lead to overfitting; a small C value makes the model simpler and may lead to underfitting. (4) RBF kernel parameter γ: Controls the influence range of a single sample. A large γ value makes the model more complex; a small γ value makes the model smoother.

[0040] Secondly, preliminary industrial analysis parameters (such as volatile matter and fixed carbon) are introduced as auxiliary variables and combined with fused feature vectors to construct a more advanced model for predicting more complex parameters such as elemental analysis or coking properties, forming a cascaded prediction network of "spectral analysis → basic indicators → advanced indicators". The core idea of ​​constructing the more advanced model is to use the output of the first-level analysis model as enhanced input information for the second-level analysis model. The trained first-level analysis model is used to predict industrial analysis indicators for samples in the entire dataset (training and testing sets). Using the fused new feature set as input and the final advanced coal quality indicators to be predicted (such as ash flow temperature FT and calorific value Q) as output, a more powerful "meta-model" is trained. The second-level analysis model can automatically handle complex interactions and nonlinear relationships between features and evaluate feature importance, determining the contribution of the predicted industrial analysis indicators.

[0041] In another specific application example, a sub-principle modeling strategy is used to determine the industrial analysis parameters of the coal sample to be tested. The machine learning model includes a Raman analysis model and an infrared analysis model. The Raman analysis model is pre-trained using a third training sample set, and the infrared analysis model is pre-trained using a fourth training sample set. The training samples in the third training sample set include Raman spectral feature samples and corresponding Raman industrial analysis parameters. The training samples in the fourth training sample set include infrared spectral feature samples and corresponding infrared industrial analysis parameters. The Raman analysis model and the infrared analysis model employ Partial Least Squares (PLS), SVM, or Random Forest models.

[0042] In this application example, step 106 includes: extracting Raman spectral features and infrared spectral features from the fused feature vector, respectively; determining Raman industrial analysis parameters based on the Raman spectral features using the Raman analysis model; determining infrared industrial analysis parameters based on the infrared spectral features using the infrared analysis model; and fusing the Raman industrial analysis parameters and the infrared industrial analysis parameters to obtain the industrial analysis parameters of the coal sample to be tested.

[0043] Raman and infrared spectroscopy models predict the same industrial analytical parameters (such as volatile matter and calorific value), but they rely on completely different spectral information sources. Raman models focus more on the "carbon skeleton," while infrared spectroscopy models focus more on "functional groups (especially hydrogen-containing groups)." Integrating the measurement results from both principles improves the accuracy of coal quality parameters by correcting for correlations.

[0044] Specifically, firstly, two models predicting the same coal quality parameter (e.g., volatile matter) are independently trained using Raman spectroscopy and infrared spectroscopy features, respectively. Then, the correlation between the prediction errors (predicted values ​​- actual values) of the two models on the validation set is calculated. Because their principles are complementary, their error patterns may differ. Finally, the performance of the two models in cross-validation (e.g., R-squared values) is used to determine the correlation. 2 Weights are assigned using either the Raman prediction model or the infrared prediction model, with higher weights for models that perform better. The final value = w1 × Raman prediction value + w2 × Infrared prediction value. Where w1 is the weight of the Raman analysis model and w2 is the weight of the infrared analysis model.

[0045] This application further stores the industrial analysis parameters of the coal sample to be tested in a memory and transmits them to a remote management terminal.

[0046] Based on the same inventive concept, this application also provides a system for implementing the methods described above. The solution provided by this system is similar to the solution described in the methods above; therefore, specific limitations in one or more system embodiments provided below can be found in the limitations of the methods described above, and will not be repeated here.

[0047] In one exemplary embodiment, an online coal quality detection system based on Raman and infrared spectroscopy fusion is provided, comprising: a coal sample shaping module, an integrated sampling probe, and a spectral analysis module.

[0048] The coal sample shaping module is used to shape the measurement surface of the coal sample to ensure that the spectral illumination surface is focused and flat. Measurement surface shaping is achieved through a designed pressure roller structure or a pressure-bearing surface. When shaping the coal powder surface using a pressure roller structure, the coal powder is deposited onto the concave surface of the pressure roller and then shaped on the roller. When shaping the coal powder surface using a pressure-bearing surface, the coal powder is fed into a fixed-shape mold and pressed by a flat surface. The measured coal sample is then returned to the pipeline via backflushing.

[0049] The integrated sampling probe is used to synchronously or asynchronously acquire Raman and infrared absorption spectra of coal samples to be tested.

[0050] Specifically, the integrated sampling probe employs a collinear optical path design, integrating the Raman excitation laser (e.g., 532nm) and the infrared measurement light source (halogen tungsten lamp) measurement optical paths onto the same probe head. This ensures that both technologies probe the same tiny sample area, resolving representativeness errors caused by asynchronous or out-of-sync measurement points. Figure 4As shown, the integrated sampling probe includes a laser 401, a halogen lamp 402, a Raman spectrometer 403, and an infrared spectrometer 404. The laser 401 and the halogen lamp 402 are used to mix and emit excitation light onto the measurement surface of the coal sample 405 through an arc-shaped reflective surface. The Raman spectrometer 403 is used to collect the Raman spectrum of the coal sample. The infrared spectrometer 404 is used to collect the infrared absorption spectrum of the coal sample 405.

[0051] Furthermore, the front end of the measurement optical path integrates a retractable sapphire or diamond optical window, which, together with a negative pressure air curtain, prevents coal dust pollution. The integrated sampling probe also has a built-in miniature temperature control unit to reduce the impact of ambient temperature fluctuations on the infrared signal.

[0052] The spectral analysis module is used to extract the coal quality sensitive features and full-spectrum features of the coal sample to be tested based on the Raman spectrum and the infrared absorption spectrum; the coal quality sensitive features and the full-spectrum features are spliced ​​together to obtain a fused feature vector; based on the fused feature vector, a pre-trained machine learning model is used to determine the industrial analysis parameters of the coal sample to be tested, so as to determine the quality of the coal sample to be tested.

[0053] In summary, this application provides an integrated, interference-resistant, and online-operable method for online coal quality detection based on the fusion of Raman and infrared spectroscopy. This method can perform real-time online testing of coal fed into thermal power plants and coal conveyed in belts, reducing the modeling workload of the coal quality system and enabling synchronous or asynchronous, rapid, and non-destructive detection of coal moisture, ash, volatile matter, fixed carbon, calorific value, sulfur content, and key functional group indicators.

[0054] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0055] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0056] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0057] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0058] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0059] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, etc., and are not limited to these.

[0060] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0061] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, 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 this application.

Claims

1. A method for online coal quality detection based on the fusion of Raman and infrared spectroscopy, characterized in that, include: The Raman and infrared absorption spectra of the coal sample to be tested are acquired synchronously or asynchronously. Based on the Raman spectrum and the infrared absorption spectrum, extract the coal quality sensitive characteristics and full-spectral characteristics of the coal sample to be tested; The coal quality sensitive features are concatenated with the full-spectrum features to obtain a fused feature vector; Based on the fused feature vector, a pre-trained machine learning model is used to determine the industrial analysis parameters of the coal sample to be tested, thereby determining the quality of the coal sample.

2. The online coal quality detection method based on Raman and infrared spectroscopy fusion according to claim 1, characterized in that, Before synchronously or asynchronously acquiring the Raman and infrared absorption spectra of the coal sample to be tested, the method further includes: The measurement surface of the coal sample to be tested is shaped to make the spectral irradiation surface focused and flat.

3. The online coal quality detection method based on Raman and infrared spectroscopy fusion according to claim 1, characterized in that, Before extracting the coal quality-sensitive characteristics and full-spectrum characteristics of the coal sample to be tested based on the Raman spectrum and the infrared absorption spectrum, the method further includes: The Raman spectrum and the infrared absorption spectrum are respectively subjected to smoothing, noise reduction, baseline correction and normalization.

4. The online coal quality detection method based on Raman and infrared spectroscopy fusion according to claim 1, characterized in that, Based on the Raman spectrum and the infrared absorption spectrum, the coal quality-sensitive characteristics and full-spectral characteristics of the coal sample to be tested are extracted, including: The peak parameters, spectral statistics, and absorbance of the Raman spectrum and the infrared absorption spectrum are extracted respectively to obtain the sensitive characteristics of coal quality. Principal component analysis was performed on the Raman spectrum and the infrared absorption spectrum respectively, and the first n principal components were retained to obtain the full spectrum characteristics.

5. The online coal quality detection method based on Raman and infrared spectroscopy fusion according to claim 4, characterized in that, The spectral peak parameters include peak height, peak position, peak intensity, peak area, and peak ratio for a specific spectral band; the spectral statistics include mean, variance, skewness, and kurtosis.

6. The online coal quality detection method based on Raman and infrared spectroscopy fusion according to claim 1, characterized in that, The machine learning model includes a first-level analysis model and a second-level analysis model; the first-level analysis model is pre-trained using a first training sample set, and the second-level analysis model is pre-trained using a second training sample set. The training samples in the first training sample set include fused feature samples and corresponding preliminary industrial analysis parameters; The training samples in the second training sample set include fused feature set samples and corresponding industrial analysis parameters; Based on the fused feature vector, a pre-trained machine learning model is used to determine the industrial analysis parameters of the coal sample to be tested, including: Based on the fused feature vector, the first-level analysis model is used to determine the preliminary industrial analysis parameters of the coal sample to be tested; The preliminary industrial analysis parameters are fused with the fused feature vector to obtain a fused feature set; Based on the fusion feature set, the industrial analysis parameters of the coal sample to be tested are determined using the second-level analysis model.

7. The online coal quality detection method based on Raman and infrared spectroscopy fusion according to claim 6, characterized in that, The first-level analysis model is a support vector machine or random forest model; the second-level analysis model is a tree ensemble model.

8. The online coal quality detection method based on Raman and infrared spectroscopy fusion according to claim 1, characterized in that, The machine learning model includes a Raman analysis model and an infrared analysis model; the Raman analysis model is pre-trained using a third training sample set, and the infrared analysis model is pre-trained using a fourth training sample set. The training samples in the third training sample set include Raman spectral feature samples and corresponding Raman industrial analysis parameters; the training samples in the fourth training sample set include infrared spectral feature samples and corresponding infrared industrial analysis parameters. Based on the fused feature vector, a pre-trained machine learning model is used to determine the industrial analysis parameters of the coal sample to be tested, including: Raman spectral features and infrared spectral features are extracted from the fused feature vector, respectively. Based on the Raman spectral characteristics, the Raman analysis model is used to determine the Raman industrial analysis parameters; Based on the infrared spectral characteristics, the infrared analysis model is used to determine the infrared industrial analysis parameters; The Raman industrial analysis parameters and the infrared industrial analysis parameters are fused to obtain the industrial analysis parameters of the coal sample to be tested.

9. A coal quality online detection system based on Raman and infrared spectroscopy fusion, applied to the coal quality online detection method based on Raman and infrared spectroscopy fusion as described in any one of claims 1-8, characterized in that, The system includes: An integrated sampling probe is used to synchronously or asynchronously acquire Raman and infrared absorption spectra of coal samples to be tested; The spectral analysis module is used to extract the coal quality sensitive features and full-spectral features of the coal sample to be tested based on the Raman spectrum and the infrared absorption spectrum; to concatenate the coal quality sensitive features and the full-spectral features to obtain a fused feature vector; and to determine the industrial analysis parameters of the coal sample to be tested based on the fused feature vector using a pre-trained machine learning model, so as to determine the quality of the coal sample to be tested.

10. The online coal quality detection system based on Raman and infrared spectroscopy fusion according to claim 9, characterized in that, The integrated sampling probe includes a laser, a halogen lamp, a Raman spectrometer, and an infrared spectrometer; The laser and the halogen lamp are used to mix and emit excitation light onto the measuring surface of the coal sample to be tested through the arc-shaped reflective surface; the Raman spectrometer is used to collect the Raman spectrum of the coal sample to be tested; and the infrared spectrometer is used to collect the infrared absorption spectrum of the coal sample to be tested.

Images