Coal gangue identification method and system based on multi-sensor fusion of acoustic characteristics

Abstract

The present invention provides a coal gangue identification method and system based on multi-sensor fusion of acoustic characteristics, which relates to the field of mineral mining. The method includes: obtaining sound signals, vibration signals and ultrasonic signals generated by the impact of the tail beam by the coal gangue that has fallen across the coal gangue; performing noise reduction processing; performing feature extraction to obtain a spectrum contrast feature matrix and a chromaticity feature matrix; performing feature fusion to obtain a more dimensional fusion feature matrix; inputting a preset model for training to obtain a trained coal gangue identification model; inputting the real-time collected signal into the model to output the coal gangue ratio identification result. The embodiment of the present invention extracts features from each signal and then fuses the features, then performs model training with the fused feature matrix, identifies the coal gangue ratio based on the trained model, improves the recognition accuracy of the gangue content in the top coal that has fallen across the coal, and can improve the recovery rate of the top coal and the quality of the coal.

Description

Technical Field

[0001] The present invention relates to the technical field of mineral mining, and in particular to a coal gangue identification method and system based on multi-sensor fusion of acoustic characteristics. Background Art

[0002] The research on comprehensive top coal caving technology has important driving significance for achieving high production, high efficiency and low consumption in the field of coal mining. The traditional top coal caving process is generally to manually control the top coal caving process, that is, the so-called "closing the window when seeing gangue", that is, when the operator finds that the top coal that has been caving contains gangue, the coal caving will be stopped immediately, which will inevitably lead to insufficient caving and cause a large amount of coal resources to be wasted; after seeing the gangue, the coal will continue to be caved for a period of time. Since it is difficult for humans to determine the gangue content by naked eyes, it may eventually cause the gangue content to exceed a certain amount, thereby causing over-cavitation and reducing the quality of coal. In addition, coal gangue is an energy source that cannot be directly used. If it is not properly handled, it will cause environmental pollution.

[0003] Manual coal caving is not only labor-intensive, but also has a harsh underground environment, threatening the safety of operators. Automated coal caving has become the first problem in the top coal caving process. The most important thing to achieve automated coal caving is to accurately identify the gangue content in the top coal that has fallen. The existing coal gangue identification method has insufficient actual identification effect on the working face. Summary of the invention

[0004] To solve the above problems, an embodiment of the present invention provides a coal gangue identification method based on multi-sensor fusion of acoustic characteristics, the method comprising: obtaining sound signals, vibration signals and ultrasonic signals generated by the fallen coal gangue impacting the tail beam; performing noise reduction processing on the sound signals, the vibration signals and the ultrasonic signals; performing feature extraction on the noise-reduced sound signals, the vibration signals and the ultrasonic signals to obtain a spectrum contrast feature matrix and a chromaticity feature matrix; performing feature fusion on the spectrum contrast feature matrix and the chromaticity feature matrix to obtain a more dimensional fusion feature matrix; inputting the fusion feature matrix and the corresponding coal gangue ratio data into a preset model for training to obtain a trained coal gangue identification model; inputting the real-time collected sound signals, vibration signals and ultrasonic signals into the trained coal gangue identification model to output a coal gangue ratio identification result.

[0005] Optionally, the feature extraction of the sound signal, the vibration signal and the ultrasonic signal after noise reduction to obtain a spectral contrast feature matrix and a chromaticity feature matrix includes: respectively performing frame processing on the sound signal, the vibration signal and the ultrasonic signal after noise reduction, and then performing short-time Fourier transform on each frame to obtain spectral contrast features and chromaticity features of each signal; and using the obtained spectral contrast features and chromaticity features to respectively extract features from each signal to obtain a spectral contrast feature matrix and a chromaticity feature matrix corresponding to each signal.

[0006] Optionally, performing a short-time Fourier transform on each sub-frame to obtain a spectrum contrast feature and a chromaticity feature of each signal includes: performing a short-time Fourier transform on each sub-frame, dividing the transformed result into 8 sub-bands; determining the energy contrast by comparing the average energy of the top digit and the bottom digit of each sub-band, and taking the logarithm log mean as the peak value Peak k , and then calculate the valley value Valley k , and then calculate the difference between the two SC k =Peak k -Valley k , {SC k ,Valley k} as the spectrum contrast feature of the original spectrum; extract the signal amplitude spectrum after short-time Fourier transform, then perform peak detection on the signal amplitude spectrum to find the largest first n peaks within a fixed frequency range, calculate the difference between any two peaks, and then combine the standard frequency specified in the twelve-tone equal temperament to obtain the fundamental frequency of the original signal ;Finally, the chromaticity feature vector is obtained by mapping each half-order tone class through the spectrum peak energy. The calculation formula of the element value of each dimension in the vector is:

[0007] in the formula and are the amplitude and frequency of the i-th peak, The frequency is The weight of the signal for the half-order tone class is calculated as: First determine the center reference frequency of each semitone class f n :f n =f b 2 k / 2 ,k=1,2……,12; Defines the center reference frequency of each signal peak and half-order tone class The interval distance between them is: , where m is an integer adjustment factor such that The value of is the smallest, and the weight value can be obtained by the following formula: ,in l is the preset weighted window length.

[0008] Optionally, the spectral contrast feature matrix and the chromaticity feature matrix are subjected to feature fusion to obtain a more dimensional fusion feature matrix, including: splicing the spectral contrast feature matrix and the chromaticity feature matrix corresponding to each signal in the vertical direction respectively; and splicing the spliced ​​matrices of each signal in the horizontal direction to obtain a more dimensional fusion feature matrix.

[0009] Optionally, the denoising process for the sound signal includes: generating multiple sound samples through a generator of a generative adversarial network, and judging the authenticity of the sound samples through a discriminator; the discriminator passes the sound signal and the false sound sample to the generator, and the generator outputs a waveform to remove the noise signal judged by the discriminator to be false.

[0010] Optionally, the denoising process for the vibration signal includes: decomposing the vibration signal into a plurality of band-limited intrinsic mode functions (BIMFs) by using a variational mode decomposition algorithm; and reconstructing a noise-free original signal by adding the plurality of BMIFs.

[0011] Optionally, the method also includes: inputting the core parameters of coal placement corresponding to the coal-gangue ratio data into the preset model for training to obtain a trained coal-gangue recognition model; the core parameters of coal placement include remaining coal placement time, target tail beam swing angle, and target plug plate extension distance; inputting the real-time collected sound signals, vibration signals and ultrasonic signals into the trained coal-gangue recognition model to output the corresponding core parameters of coal placement.

[0012] Optionally, the method further includes: sending the corresponding coal caving core parameters to an operator terminal.

[0013] Optionally, the method further includes: sending the coal gangue ratio identification result to an underground terminal, so that the underground terminal controls the color of underground lights according to the coal gangue ratio identification result.

[0014] An embodiment of the present invention provides a coal gangue identification system based on multi-sensor fusion of acoustic characteristics, the system comprising: a sound sensor, a vibration acceleration sensor, an acoustic emission sensor and a single-chip microcomputer; the single-chip microcomputer is used to execute the above-mentioned coal gangue identification method based on multi-sensor fusion of acoustic characteristics.

[0015] The coal gangue identification method and system based on multi-sensor fusion of acoustic characteristics provided in the embodiment of the present invention realize coal gangue identification by fusing the sound signal, vibration signal and ultrasonic signal generated by the fallen coal gangue impacting the tail beam, extracts features from each signal and then fuses the features, then trains the model with the fused feature matrix, and identifies the coal gangue ratio based on the trained model, thereby improving the recognition accuracy of the gangue content in the fallen top coal, and can improve the recovery rate of the top coal and enhance the coal quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0017] Figure 1 A structural block diagram of a coal gangue identification system based on multi-sensor fusion of acoustic characteristics provided by an embodiment of the present invention; Figure 2 A schematic flow chart of a coal gangue identification method based on multi-sensor fusion of acoustic characteristics provided by an embodiment of the present invention; Figure 3 A flow chart of a coal gangue identification method using multi-sensor data fusion provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned objects, features and advantages of the present invention more clearly understood, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0019] The embodiment of the present invention provides a coal gangue identification method and system based on multi-sensor fusion of acoustic characteristics, which can realize the accurate prediction of the gangue content of top coal, not only can improve the recovery rate of top coal and enhance the coal quality, but also provide a theoretical basis for the automation of coal caving, and is the technical support for realizing automatic mining of fully-mechanized caving working face.

[0020] The multi-sensor fusion coal gangue identification system based on acoustic characteristics provided by the embodiment of the present invention has a hardware part including a sound sensor, a vibration acceleration sensor, an acoustic emission sensor and a single-chip microcomputer. The sound sensor includes a housing for installing a sound pickup module, a signal noise reduction module, etc.; the vibration acceleration sensor includes a housing, a signal acquisition card, a noise reduction module, etc.; the acoustic emission sensor includes a signal acquisition card, an ultrasonic signal noise reduction module, etc.

[0021] During implementation, sound sensors, vibration acceleration sensors, and acoustic emission sensors can be pre-installed. Specifically, the sensors are installed on the tail beam of the underground top coal caving hydraulic support. The number of sensors installed should be the same as the number of parallel hydraulic supports, so as to ensure the integrity of the data collection of underground top coal caving. At the same time, explosion-proof devices are installed for each sensor to ensure that each sensor can normally collect data in the harsh environment underground.

[0022] For example, Figure 1 A structural block diagram of a coal gangue identification system based on multi-sensor fusion of acoustic characteristics provided in an embodiment of the present invention. Figure 1 It shows that the output data of the sound sensor is denoised through a generative adversarial network; the vibration acceleration sensor is connected to a data acquisition card, and the output data is decomposed and reconstructed by VMD for denoising; the acoustic emission sensor is connected to an ultrasonic signal acquisition card, and the output data is subjected to ultrasonic denoising; the denoised data of the three are transmitted to the microcontroller module and the data storage module via Ethernet.

[0023] Furthermore, the single-chip computer module and the data storage module are connected to the handheld terminal via an Ethernet chip and a wireless communication module, and the handheld terminal is connected to a recognition result display module and a waveform display module. Figure 1 Also shown is a light prompt module, which specifically includes a strobe light and a color changing device.

[0024] Figure 2 A flow chart of a coal gangue identification method based on multi-sensor fusion of acoustic characteristics provided by an embodiment of the present invention, the method comprising the following steps: S202, obtaining sound signals, vibration signals and ultrasonic signals generated by the impact of the falling coal gangue on the tail beam.

[0025] For example, the sound pickup module is used to collect the sound signal generated by the coal flow with different gangue contents hitting the tail beam of the hydraulic support during the complete top coal caving process; the vibration acceleration signal sensor is used to collect the vibration signal generated by the impact of the tail beam of the hydraulic support during the complete span of coal gangue; the acoustic emission sensor is used to collect the ultrasonic signal that cannot be collected by the sound and vibration sensors underground to prevent signal loss. Among them, the acoustic emission sensor can collect high-frequency signals, such as sound signals with a frequency higher than 20KHz, and the sound signals below 20KHz can be collected by the above-mentioned sound sensor.

[0026] S204, performing noise reduction processing on the sound signal, the vibration signal and the ultrasonic signal.

[0027] For example, the noise reduction process of the sound signal includes: generating a large number of sound samples through the generator of the generative adversarial network, judging the authenticity of the sound samples through the discriminator; the discriminator passes the sound signal and the false sound sample to the generator, and the generator outputs a waveform to remove the noise signal judged by the discriminator to be false. Among them, the discriminator judges the authenticity of the sound sample, the discriminator passes the sound and the false sound sample to the generator, and the generator outputs a waveform to remove the noise signal judged by the discriminator to be false. The sound signal after noise reduction is passed to the single-chip microcomputer module through Ethernet as a training data set for the sound signal For example, the vibration signal is subjected to noise reduction processing, including: using the variational mode decomposition (VMD) algorithm to decompose the vibration signal into multiple band-limited intrinsic mode functions (BIMF); and reconstructing the noise-free original signal by adding multiple BMIFs.

[0028] Each BIMF represents the oscillation of the signal within a certain frequency range, and the amplitude and frequency change over time. By adding these BMIFs, the noise-free original signal can be reconstructed and then transmitted to the MCU module via Ethernet as a training data set for the vibration signal. S206, extracting features from the denoised sound signal, vibration signal and ultrasonic signal to obtain a spectrum contrast feature matrix and a chromaticity feature matrix.

[0029] Optionally, the denoised sound signal, vibration signal and ultrasonic signal are framed, and then each frame is subjected to short-time Fourier transform to obtain the spectrum contrast feature and chromaticity feature of each signal; then, the obtained spectrum contrast feature and chromaticity feature are used to extract features of each signal, and the spectrum contrast feature matrix and chromaticity feature matrix corresponding to each signal are obtained. Further, the feature extraction result can be stored in the single-chip microcomputer module.

[0030] Exemplarily, a short-time Fourier transform is performed on each sub-frame to obtain the spectrum contrast characteristics and chromaticity characteristics of each signal, including: Perform short-time Fourier transform on each sub-frame and divide the transformed result into 8 sub-bands; The energy contrast is determined by comparing the average energy of the top quantile and the bottom quantile of each subband, and the logarithmic mean is taken as the peak value. k , and then calculate the valley value Valley k , and then calculate the difference between the two SC k =Peak k -Valley k , {SC k,Valley k} as the spectral contrast feature of the original spectrum; Extract the signal amplitude spectrum after short-time Fourier transform, then perform peak detection on the signal amplitude spectrum to find the largest first n peaks within a fixed frequency range, calculate the difference between any two peaks, and then combine the standard frequency specified in the twelve-tone equal temperament to obtain the fundamental frequency of the original signal ; Finally, the chromaticity feature vector is obtained by mapping each half-order tone class through the spectrum peak energy. The calculation formula of the element value of each dimension in the vector is:

[0031] in the formula and are the amplitude and frequency of the i-th peak, The frequency is The weight of the signal for the half-order tone class is calculated as: First determine the center reference frequency of each semitone class f n :f n =f b 2 k / 2 ,k=1,2……,12; Defines the center reference frequency of each signal peak and half-order tone class The interval distance between them is: , where m is an integer adjustment factor such that The value of is the smallest, and the weight value can be obtained by the following formula: ,in l is the preset weighted window length.

[0032] S208, performing feature fusion on the above spectrum contrast feature matrix and chromaticity feature matrix to obtain a multi-dimensional fusion feature matrix.

[0033] Optionally, the spectral contrast feature matrix and the chromaticity feature matrix corresponding to each signal are spliced ​​in the vertical direction respectively; then, the spliced ​​matrices of each signal are spliced ​​in the horizontal direction to obtain a more dimensional fusion feature matrix.

[0034] For example, the spectrum contrast feature matrix A and the chromaticity feature matrix C of the sound signal, the spectrum contrast feature matrix B and the chromaticity feature matrix D of the vibration signal; Then the spectral contrast feature matrix A and the chromaticity feature matrix C of the sound signal are concatenated in the vertical direction to obtain the matrix ; The spectral contrast feature matrix B and the chromaticity feature matrix D of the vibration signal are spliced ​​in the vertical direction to obtain the matrix ; The matrix of the sound signal and the vibration signal is spliced ​​in the horizontal direction to obtain a more dimensional fusion feature matrix .

[0035] S210, inputting the fused feature matrix and the corresponding gangue ratio data into a preset model for training to obtain a trained gangue recognition model.

[0036] Optionally, the new fusion feature matrix is ​​transmitted to the host computer through the single-chip microcomputer module for model training.

[0037] S212, inputting the real-time collected sound signal, vibration signal and ultrasonic signal into the trained coal gangue recognition model, and outputting the coal gangue ratio recognition result.

[0038] The coal gangue identification method based on multi-sensor fusion of acoustic characteristics provided in the embodiment of the present invention realizes coal gangue identification by fusing the sound signal, vibration signal and ultrasonic signal generated by the fallen coal gangue impacting the tail beam, extracts features from each signal and then fuses the features, then trains the model with the fused feature matrix, and identifies the coal gangue proportion based on the trained model, thereby improving the recognition accuracy of the gangue content in the fallen top coal, and can improve the recovery rate of the top coal and enhance the coal quality.

[0039] On the basis of the model's coal gangue ratio recognition training, it can also be trained on the coal caving core parameters, which should at least include: coal caving time, target tail beam swing angle, target plug plate telescopic distance, etc. Based on this, the above method can also include: inputting the coal caving core parameters corresponding to the coal gangue ratio data into the preset model for training to obtain the trained coal gangue recognition model; then, inputting the real-time collected sound signals, vibration signals and ultrasonic signals into the trained coal gangue recognition model to output the corresponding coal caving core parameters.

[0040] According to the above-mentioned core parameters of coal caving, the tail beam and the plug plate of the hydraulic support perform corresponding actions according to the generated core parameters during the top coal caving process, thereby avoiding the occurrence of "under-caving" or "over-caving" during the top coal caving process, thereby reducing the waste of coal resources and improving the quality of coal.

[0041] For example, the above-mentioned coal placing core parameter can be a three-dimensional array, which is composed of the remaining coal placing time, the target tail beam swing angle, the target plugboard extension distance, and the target tail beam swing angle. ,in t The remaining coal-laying time, θ is the tail beam swing angle, l The extension distance of the plug board.

[0042] In this embodiment, the entire top coal caving process is divided into multiple coal caving stages (pure coal caving stage, coal-gangue mixed stage, pure gangue caving stage), and the gangue content in the coal-gangue flow of each coal caving stage changes continuously. The current coal caving state is predicted based on the results after model training and the core parameters of coal caving are generated.

[0043] Furthermore, the corresponding coal placing core parameters can be sent to the operator terminal, and the operator can perform manual or automatic control based on the coal placing core parameters.

[0044] Furthermore, the gangue ratio recognition result may be sent to an underground terminal, so that the underground terminal controls the color of the underground light according to the gangue ratio recognition result.

[0045] The vibration signal generated by the impact of the tail beam caused by the falling of coal gangue after noise reduction is analyzed after training with a convolutional neural network model to obtain the recognition result. The current coal discharge status is transmitted to the handheld terminal via the Ethernet chip and wireless communication module, and the handheld terminal then controls the color of the underground light according to the recognition result.

[0046] The operator can know the current coal discharge status according to the color of the light prompt module. When the identification result shows that the current coal discharge status is the pure coal discharge stage, no response is made. If it is the coal-gangue mixed stage and the gangue content is greater than the calibrated range, the operator operates the control console to extend the tail beam and stop coal discharge.

[0047] Figure 3 The flowchart of the coal gangue identification method using multi-sensor data fusion provided by an embodiment of the present invention is shown, and includes the following steps: S301, turn on the sound sensor, vibration acceleration sensor, and acoustic emission sensor.

[0048] S302, audio, vibration, and ultrasonic signal acquisition.

[0049] S303, reducing noise on the collected signal.

[0050] S304, feature extraction + feature fusion.

[0051] S305, inputting the fused features into the single chip computer module for training.

[0052] S306, determine whether the test accuracy is less than a set threshold. If yes, execute S307; otherwise, return to execute S305.

[0053] S307, the model training stops and is saved.

[0054] S308, the coal placing window is opened and the coal placing process begins.

[0055] S309, the sound, vibration, and acoustic emission sensors respectively input signals into the model for recognition.

[0056] S310, determine whether the recognition result reaches the set coal gangue ratio value. If yes, execute S312; otherwise, return to execute S309.

[0057] S311, the lighting prompt device displays a light of a specific color according to the recognition result to inform the operator of the current coal placing status.

[0058] S312, the single chip microcomputer module issues a control command to extend the tail boom until the coal discharge stops.

[0059] The embodiment of the present invention can accurately identify the coal gangue ratio, thereby avoiding the occurrence of "under-releasing" or "over-releasing" during the top coal caving process, thereby reducing the waste of coal resources and improving the quality of coal.

[0060] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the control device through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it may include the processes of the above-mentioned method embodiments, wherein the storage medium may be a memory, a disk, an optical disk, etc.

[0061] In this article, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements. The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A coal gangue identification method based on multi-sensor fusion of acoustic characteristics, characterized in that: The method comprises: Acquire the sound signal, vibration signal and ultrasonic signal generated by the impact of the falling coal gangue on the tail beam; Performing noise reduction processing on the sound signal, the vibration signal and the ultrasonic signal; Performing feature extraction on the denoised sound signal, the vibration signal and the ultrasonic signal to obtain a spectrum contrast feature matrix and a chromaticity feature matrix; Performing feature fusion on the spectrum contrast feature matrix and the chromaticity feature matrix to obtain a multi-dimensional fusion feature matrix; Inputting the fused feature matrix and the corresponding gangue ratio data into a preset model for training to obtain a trained gangue recognition model; The sound signal, vibration signal and ultrasonic signal collected in real time are input into the trained coal gangue recognition model, and the coal gangue ratio recognition result is output.

2. The method according to claim 1, characterized in that The feature extraction of the de-noised sound signal, the vibration signal and the ultrasonic signal to obtain a spectrum contrast feature matrix and a chromaticity feature matrix includes: The denoised sound signal, the vibration signal and the ultrasonic signal are respectively subjected to frame processing, and then each frame is subjected to short-time Fourier transform to obtain the spectrum contrast characteristics and chromaticity characteristics of each signal; The obtained spectrum contrast feature and chromaticity feature are used to extract features of each signal respectively, so as to obtain a spectrum contrast feature matrix and a chromaticity feature matrix corresponding to each signal.

3. The method according to claim 2, characterized in that The short-time Fourier transform is performed on each sub-frame to obtain the spectrum contrast characteristics and chromaticity characteristics of each signal, including: Perform short-time Fourier transform on each sub-frame and divide the transformed result into 8 sub-bands; The energy contrast is determined by comparing the average energy of the top quantile and the bottom quantile of each subband, and the logarithmic mean is taken as the peak value. k , and then calculate the valley value Valley k , and then calculate the difference between the two SC k =Peak k -Valley k , {SC k ,Valley k } as the spectral contrast feature of the original spectrum; Extract the signal amplitude spectrum after short-time Fourier transform, then perform peak detection on the signal amplitude spectrum to find the largest first n peaks within a fixed frequency range, calculate the difference between any two peaks, and then combine the standard frequency specified in the twelve-tone equal temperament to obtain the fundamental frequency of the original signal ;Finally, the chromaticity feature vector is obtained by mapping each half-order tone class through the spectrum peak energy. The calculation formula of the element value of each dimension in the vector is: in the formula and are the amplitude and frequency of the i-th peak, The frequency is The weight of the signal for the half-order tone class is calculated as: First determine the center reference frequency of each semitone class f n :f n =f b 2 k / 2 ,k=1,2……,12; Defines the center reference frequency of each signal peak and half-order tone class The interval distance between them is: , where m is an integer adjustment factor such that The value of is the smallest, and the weight value can be obtained by the following formula: ,in l is the preset weighted window length.

4. The method according to claim 2, characterized in that: The spectral contrast feature matrix and the chromaticity feature matrix are subjected to feature fusion to obtain a multi-dimensional fusion feature matrix, including: The spectrum contrast feature matrix and the chromaticity feature matrix corresponding to each signal are respectively spliced ​​in the vertical direction; The matrices of the spliced ​​signals are spliced ​​horizontally to obtain a more dimensional fusion feature matrix.

5. The method according to claim 1, characterized in that The performing noise reduction processing on the sound signal comprises: Generate multiple sound samples through the generator of the generative adversarial network, and use the discriminator to determine the authenticity of the sound samples; The discriminator transmits the sound signal and the false sound sample to the generator, and the generator outputs a waveform with the noise signal determined to be false by the discriminator removed.

6. The method according to claim 1, characterized in that The performing noise reduction processing on the vibration signal comprises: Decomposing the vibration signal into a plurality of band-limited intrinsic mode functions (BIMFs) using a variational mode decomposition algorithm; By adding the multiple BMIFs, a noise-free original signal is reconstructed.

7. The method according to claim 1, characterized in that The method further comprises: Inputting the coal placing core parameters corresponding to the coal gangue ratio data into the preset model for training to obtain a trained coal gangue recognition model; the coal placing core parameters include the remaining coal placing time, the target tail beam swing angle, and the target plugboard telescopic distance; The sound signal, vibration signal and ultrasonic signal collected in real time are input into the trained coal gangue recognition model, and the corresponding coal discharge core parameters are output.

8. The method according to claim 7, characterized in that The method further comprises: The corresponding coal placing core parameters are sent to the operator terminal.

9. The method according to claim 1, characterized in that: The method further comprises: The coal gangue ratio recognition result is sent to an underground terminal, so that the underground terminal controls the color of the underground light according to the coal gangue ratio recognition result.

10. A coal gangue identification system based on multi-sensor fusion of acoustic characteristics, characterized in that: The system comprises: a sound sensor, a vibration acceleration sensor, an acoustic emission sensor and a single chip microcomputer; The single chip microcomputer is used to execute the coal gangue identification method based on multi-sensor fusion of acoustic characteristics as described in any one of claims 1-9.

Images