Coal gangue identification method based on acoustic emission signals

Abstract

The invention provides a coal gangue identification method based on an acoustic emission signal, and the method comprises the steps: obtaining an original acoustic emission signal generated when a target object falls to impact a steel plate through an acoustic emission sensor disposed in the range of a tail beam of a hydraulic support at a coal discharge port, and carrying out the preprocessing of the original acoustic emission signal, so as to obtain a first acoustic emission signal; a first feature of the first acoustic emission signal is extracted, the first feature comprises one or more of ringing count, event count, signal energy, signal amplitude, duration and rise time, the ringing count is the number of times exceeding a preset threshold voltage, the event count is the number of independent waveform events, the signal energy is an energy value within a fixed time, and the signal amplitude is the number of times exceeding the preset threshold voltage; the signal amplitude is the maximum amplitude, the duration is the time length from beginning to ending, and the rising time is the time from beginning to reaching the maximum amplitude; and training an identification model based on the first feature, the identification model being used for identifying the category of the target object, the category being one of coal, gangue and a coal gangue mixture.

Description

Technical Field

[0001] The present invention relates to the field of acoustic technology, and in particular to a coal gangue identification method based on acoustic emission signals. Background Art

[0002] Coal is an important global energy resource. During the mining process, gangue is mined along with the coal as a by-product mineral. The inclusion of gangue not only reduces the quality and calorific value of the coal, but also increases transportation costs and the difficulty of subsequent processing. With the continuous expansion of coal mining scale and the increase in mining depth, the output of gangue is also increasing. Accurately identifying and separating gangue is crucial to improving coal production efficiency, reducing costs, and reducing environmental pollution. Traditional gangue identification methods mainly rely on manual observation on the conveyor belt to identify gangue. This method is inefficient and labor-intensive, and the recognition accuracy is greatly affected by human factors, making it difficult to meet the needs of large-scale production. Summary of the Invention

[0003] The purpose of the present invention is to provide a coal gangue identification method based on acoustic emission signals, and to apply acoustic emission technology to the field of coal gangue detection under fully-mechanized coal caving process, aiming to provide an efficient and accurate coal gangue detection method.

[0004] To achieve the above objectives, in a first aspect, the present invention provides a coal gangue identification method based on acoustic emission signals, comprising:

[0005] An acoustic emission sensor installed within the tail beam of the hydraulic support at the coal caving port is used to obtain an original acoustic emission signal generated by the target object falling and hitting the steel plate, and the original acoustic emission signal is preprocessed to obtain a first acoustic emission signal;

[0006] Extracting a first feature of the first acoustic emission signal, the first feature including one or more of a ring count, an event count, a signal energy, a signal amplitude, a duration, and a rise time, wherein the ring count is the number of times the first acoustic emission signal exceeds a preset threshold voltage, the event count is the number of independent waveform events in the first acoustic emission signal, the signal energy is the energy value of the first acoustic emission signal within a fixed time, the signal amplitude is the maximum amplitude of the first acoustic emission signal, the duration is the time length from the start to the end of the first acoustic emission signal, and the rise time is the time from the start of the first acoustic emission signal to the maximum amplitude;

[0007] A recognition model is trained based on the first feature, and the recognition model is used to identify the category of the target object, which is one of coal, gangue, and coal-gangue mixture.

[0008] Specifically, it also includes: extracting a second feature and a third feature of the first acoustic emission signal; the second feature includes a Mel frequency cepstral coefficient, and the third feature includes a gammatone frequency cepstral coefficient;

[0009] Training a recognition model based on the first feature includes:

[0010] The recognition model is trained based on the first feature, the second feature, and the third feature.

[0011] Specifically, a second feature and a third feature of the first acoustic emission signal are extracted, wherein the second feature includes a Mel-frequency cepstral coefficient, including:

[0012] Performing a short-time Fourier transform on the first acoustic emission signal to obtain a spectrum, calculating a power spectrum based on the spectrum, filtering the power spectrum through a Mel filter bank to obtain a second acoustic emission signal, calculating an energy value of the second acoustic emission signal, taking the logarithm of the energy value to obtain a logarithmic Mel spectrum, and performing a discrete cosine transform on the logarithmic Mel spectrum to obtain a Mel-frequency cepstrum coefficient.

[0013] Specifically, the second feature and the third feature of the first acoustic emission signal are extracted, wherein the third feature includes a gammatone frequency cepstral coefficient, including:

[0014] Performing a short-time Fourier transform on the first acoustic emission signal to obtain a spectrum, calculating a power spectrum based on the spectrum, filtering the power spectrum through a gammatone filter bank to obtain a third acoustic emission signal, calculating an energy value of the third acoustic emission signal, taking a logarithm of the energy value to obtain a logarithmic gammatone spectrum, and performing a discrete cosine transform on the logarithmic gammatone spectrum to obtain a gammatone frequency cepstral coefficient.

[0015] Preferably, the recognition model is a neural network model or a support vector machine model.

[0016] Specifically, preprocessing the original acoustic emission signal to obtain the first acoustic emission signal includes: filtering the original acoustic emission signal through a bandpass filter to obtain the first acoustic emission signal.

[0017] Preferably, the installation position of the acoustic emission sensor is optimized according to the coal gangue flow law and / or the hydraulic support action law.

[0018] In a second aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute the method described in the first aspect.

[0019] In a third aspect, the present invention provides a computing device comprising a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the method described in the first aspect is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A flow chart of a coal gangue identification method based on acoustic emission signals provided by an embodiment of the present invention;

[0021] Figure 2 An installation position diagram of an acoustic emission sensor provided in an embodiment of the present invention;

[0022] Figure 3 This is a flowchart for calculating Mel-frequency cepstral coefficients extracted by an embodiment of the present invention;

[0023] Figure 4 A schematic diagram of a Mel filter bank provided by an embodiment of the present invention;

[0024] Figure 5 Mel-frequency cepstral coefficient calculation flow chart provided by the embodiment of the present invention

[0025] Figure 6 A schematic diagram of a gammatone filter bank provided by an embodiment of the present invention;

[0026] Figure 7 A schematic diagram of a recognition model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments.

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be described below with reference to the accompanying drawings. It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0029] In the description of the embodiments of the present invention, words such as "exemplary," "for example," or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for example" in the embodiments of the present invention should not be construed as preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "for example," or "for example" is intended to present the relevant concepts in a concrete manner.

[0030] Distinguishing coal from gangue during fully mechanized coal caving is a crucial step in coal mining. Traditional gangue identification methods rely primarily on manual observation on conveyor belts. This method is inefficient, labor-intensive, and its accuracy is significantly affected by human factors, making it difficult to meet the needs of large-scale production.

[0031] In order to overcome the shortcomings of the existing technology and based on the above content, a coal gangue identification method based on acoustic emission signals is proposed. Acoustic emission technology is a non-destructive testing technology based on the release of elastic wave signals by materials or structures during stress or deformation. It captures transient acoustic wave signals generated by defect expansion, crack initiation, phase change or plastic deformation inside the material, and analyzes its characteristics to evaluate the integrity, degree of damage and failure risk of the material or structure. The present invention analyzes the acoustic emission signals of coal gangue to obtain common acoustic emission features, Mel frequency cepstral coefficients and gammatone frequency cepstral coefficients, and trains a recognition model based on the three types of features of the acoustic emission signals. The recognition model is used to distinguish between coal, gangue and coal gangue mixtures.

[0032] Figure 1 The flow chart of a coal gangue identification method based on acoustic emission signals provided by an embodiment of the present invention. Figure 1 As shown, the method mainly includes the following steps:

[0033] Step S101: An acoustic emission sensor installed within the tail beam of the hydraulic support at the coal caving port captures a raw acoustic emission signal generated by a falling object impacting a steel plate. The raw acoustic emission signal is preprocessed to obtain a first acoustic emission signal. The specific method for preprocessing the raw acoustic emission signal may vary in different embodiments. In one embodiment, the first acoustic emission signal may be obtained by filtering the raw acoustic emission signal using a bandpass filter.

[0034] In another embodiment, the installation position of the acoustic emission sensor is optimized according to the coal gangue flow law and / or the hydraulic support action law.

[0035] Figure 2 This diagram illustrates the installation location of an acoustic emission sensor provided in an embodiment of the present invention. As shown, the sensor is installed within the tail beam of the hydraulic support at the coal vent to capture the acoustic emission signal generated when the target (coal gangue) falls and impacts the support's steel plate, causing it to shatter. The sensor's installation location and method should be optimized based on the flow patterns of the coal gangue and / or the impact of the hydraulic support's operation. Signal preprocessing is performed by filtering the acoustic emission signal using a bandpass filter, for example, within a frequency band between 1kHz and 300kHz.

[0036] Step S102: Extracting a first feature of the first acoustic emission signal, where the first feature includes one or more of a ring count, an event count, a signal energy, a signal amplitude, a duration, and a rise time, wherein the ring count is the number of times the first acoustic emission signal exceeds a preset threshold voltage, the event count is the number of independent waveform events in the first acoustic emission signal, the signal energy is the energy value of the first acoustic emission signal within a fixed time, the signal amplitude is the maximum amplitude of the first acoustic emission signal, the duration is the time length from the start to the end of the first acoustic emission signal, and the rise time is the time from the start of the first acoustic emission signal to the maximum amplitude.

[0037] Exemplarily, by calculating some common acoustic emission characteristic values ​​of the acoustic emission signal, a first feature of the first acoustic emission signal is obtained, where the first feature includes one or more of ring count, event count, signal energy, signal amplitude, duration, and rise time. For example,

[0038] Ring count: The number of times the acoustic emission signal exceeds the preset threshold voltage. The calculation formula is Where V i is the i-th signal amplitude, V th is the threshold voltage, N r Count the rings;

[0039] Event count: The number of independent waveform events in the acoustic emission signal, calculated as N e = count(unique(V)), where V is the amplitude sequence of the acoustic emission signal, N e Count events;

[0040] Signal energy: The energy value of the acoustic emission signal within a fixed time. The calculation formula is: Where V(t) is the function of the acoustic emission signal changing with time, and E is the signal energy;

[0041] Signal amplitude: The maximum amplitude of the acoustic emission signal, calculated as A = max(V), where V is the acoustic emission signal amplitude sequence and A is the signal amplitude;

[0042] Duration: The length of time from the start to the end of the acoustic emission signal, calculated as t d =t e -t s , where t e is the signal end time, t s is the signal start time, t d is the duration;

[0043] Rise time: The time from the start of the acoustic emission signal to the maximum amplitude, calculated as T r =tmax -t s , where t max is the time when the signal reaches the maximum amplitude, t s is the signal start time, T r is the rise time.

[0044] Step S103: training a recognition model based on the first feature, wherein the recognition model is used to identify the category of the target object, which is one of coal, gangue, and coal-gangue mixture.

[0045] In one embodiment, a second feature and a third feature of the first acoustic emission signal may also be extracted; the second feature includes Mel-frequency cepstral coefficients, and the third feature includes gamma-tone frequency cepstral coefficients. Furthermore, the recognition model may be trained based on the first, second, and third features.

[0046] Exemplarily, a Mel frequency cepstral coefficient (MFCC) is obtained as a second feature, a short-time Fourier transform is performed on the first acoustic emission signal to obtain a spectrum, a power spectrum is calculated based on the spectrum, the power spectrum is filtered by a Mel filter group to obtain a second acoustic emission signal, the energy value of the second acoustic emission signal is calculated, the logarithm of the energy value is taken to obtain a logarithmic Mel spectrum, and a discrete cosine transform is performed on the logarithmic Mel spectrum to obtain a Mel frequency cepstral coefficient.

[0047] Figure 3 The flowchart for calculating Mel-frequency cepstral coefficients provided by the embodiment of the present invention is shown in the figure.

[0048] S301 Short-time Fourier transform: Pre-emphasize, frame, and window the first acoustic emission signal, and perform fast Fourier transform (FFT) on each frame of the windowed signal to obtain a spectrum. The calculation formula is:

[0049]

[0050] Among them, s i (n) is the data of the i-th frame, h(n) is the time domain signal, n ranges from 1 to 400, e -j2πkn / N The complex exponential basis function of Discrete Fourier Transform (DFT) represents the sine wave component with frequency k, S i (k) represents the kth complex coefficient of the i-th frame.

[0051] S302 Calculate the power spectrum: The power spectrum represents the power distribution of the signal at different frequencies. For a time domain signal, its power spectrum is a representation of the power of the signal at each frequency component. The power spectrum is calculated from the STFT result, that is, the amplitude of each frequency component is squared to obtain the energy of the frequency component. The calculation formula is:

[0052]

[0053] Among them, S i (k) represents the kth complex coefficient of the i-th frame, P i (k) is the power spectrum of the i-th frame.

[0054] S303 Mel filter bank: Map the spectrum to the Mel frequency scale and perform weighted averaging of the spectrum using a set of triangular filters. The spectrum is then mapped to the Mel frequency scale. The relationship between the Mel frequency and the normal frequency f is: M(f) = 1125ln(1 + f / 700). Select approximately 20-40 (usually 26) triangular filter banks. The filter bank formula is as follows:

[0055]

[0056] in, f(m) is the center frequency.

[0057] Figure 4 A schematic diagram of a Mel filter bank provided for an embodiment of the present invention is shown in the figure. The Mel filter bank of an embodiment of the present invention is composed of 26 (filter) vectors with a length of 257. Most of the 257 values ​​of each filter are 0, and only the frequency range to be collected is non-zero. The power spectrum is filtered by the Mel filter bank.

[0058] S304 Filter Bank Energy: The input 257-point signal will pass through 26 filters, and the energy of the signal passing through each filter will be calculated.

[0059] S305 takes the logarithm: takes the logarithm of each value output by the filter bank to obtain a logarithmic Mel spectrum to simulate the human ear's perception of sound intensity.

[0060] S306 Discrete Cosine Transform: Perform discrete cosine transform on the logarithmic Mel spectrum to obtain Mel frequency cepstral coefficients, where the calculation formula is:

[0061]

[0062] Where s(m) is the input discrete signal sequence, C(n) is the transformed Mel-frequency cepstral coefficient, M is the number of triangular filters, n is the output frequency index, L refers to the MFCC coefficient order, and m is the input time index.

[0063] Exemplarily, a gammatone frequency cepstral coefficient (GFCC) is obtained as a third feature, a short-time Fourier transform is performed on the first acoustic emission signal to obtain a spectrum, a power spectrum is calculated based on the spectrum, the power spectrum is filtered by a gammatone filter group to obtain a third acoustic emission signal, the energy value of the third acoustic emission signal is calculated, the logarithm of the energy value is taken to obtain a logarithmic gammatone spectrum, and a discrete cosine transform is performed on the logarithmic gammatone spectrum to obtain a gammatone frequency cepstral coefficient.

[0064] Figure 5 The flowchart for calculating Mel-frequency cepstral coefficients provided by the embodiment of the present invention is shown in the figure.

[0065] S501 Short-time Fourier transform: Pre-emphasize, frame, and window the first acoustic emission signal, and perform fast Fourier transform (FFT) on each frame of the windowed signal to obtain a spectrum. The calculation formula is:

[0066]

[0067] Among them, s i (n) is the data of the i-th frame, h(n) is the time domain signal, n ranges from 1 to 400, e -j2πkn / N The complex exponential basis function of Discrete Fourier Transform (DFT) represents the sine wave component with frequency k, S i (k) represents the kth complex coefficient of the i-th frame.

[0068] S502 calculates the power spectrum: The power spectrum represents the power distribution of the signal at different frequencies. For a time domain signal, its power spectrum is a representation of the power of the signal at each frequency component. The power spectrum is calculated from the STFT result, that is, the amplitude of each frequency component is squared to obtain the energy of the frequency component. The calculation formula is:

[0069]

[0070] Among them, S i (k) represents the kth complex coefficient of the i-th frame, P i (k) is the power spectrum of the i-th frame.

[0071] S503 Gammatone Filter Bank: The gammatone filter is a commonly used cochlear filter model used to describe the impulse response and amplitude-frequency characteristics of the basilar membrane. Its spectral peak is flatter and slower than that of the Mel filter, which can effectively compensate for the insufficient energy of the triangular filter. The time domain expression of the gammatone filter is:

[0072] g(t)=at n-1 e -2πbtcos(2πf c t+φ0)

[0073] Among them, f c represents the center frequency of the filter, φ0 represents the initial phase, a represents the filter amplitude, n represents the filter order and is related to the filter shape, b represents the bandwidth of the filter, t represents time, and φ0 represents the filter phase.

[0074] Figure 6 This is a schematic diagram of a gammatone filter bank provided by an embodiment of the present invention. As shown in the figure, the gammatone filter bank of the embodiment of the present invention is composed of 32 gammatone filters, and the power spectrum is filtered by the gammatone filter bank.

[0075] S504 Filter Bank Energy: Calculates the energy of the signal passing through each gammatone filter.

[0076] S505 takes the logarithm: takes the logarithm of each value output by the gammatone filter bank to obtain a logarithmic gammatone spectrum.

[0077] S506 Discrete Cosine Transform: Performing discrete cosine transform on the logarithmic gammatone spectrum to obtain gammatone frequency cepstral coefficients.

[0078] In different embodiments, the specific type of the recognition model may be different. For example, in one embodiment, the recognition model may be a neural network model or a support vector machine model.

[0079] Figure 7 The schematic diagram of the recognition model provided in the embodiment of the present invention inputs the three types of features of the extracted acoustic emission signal (the first feature (common eigenvalue), the second feature (Mel-frequency cepstral coefficient), and the third feature (gammatone frequency cepstral coefficient)) into the recognition network for recognition. Common classifiers in the recognition network include support vector machines (SVM), deep neural networks (DNN), random forests, etc. By training and optimizing the classifier, high-precision recognition of coal gangue audio is achieved. The main steps include: first, data preparation and feature input; second, creating a linear kernel vector machine (SVM) model and training it using the training set; then calculating the accuracy of the model on the test set; finally, using the trained model to predict the feature signals in the test set and visualize the prediction results. By using the trained recognition model to identify the real coal gangue acoustic emission signals, coal, gangue, and coal gangue mixtures are separated.

[0080] It is understood that the method steps in the embodiments of the present invention can be implemented by hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, mobile hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be an integral part of the processor. The processor and the storage medium can be located in an ASIC.

[0081] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted via the computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, hard disk, tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0082] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A coal gangue identification method based on acoustic emission signals, comprising: An acoustic emission sensor installed within the tail beam of the hydraulic support at the coal caving port is used to obtain an original acoustic emission signal generated by the target object falling and hitting the steel plate, and the original acoustic emission signal is preprocessed to obtain a first acoustic emission signal; Extracting a first feature of the first acoustic emission signal, the first feature including one or more of a ring count, an event count, a signal energy, a signal amplitude, a duration, and a rise time, wherein the ring count is the number of times the first acoustic emission signal exceeds a preset threshold voltage, the event count is the number of independent waveform events in the first acoustic emission signal, the signal energy is the energy value of the first acoustic emission signal within a fixed time, the signal amplitude is the maximum amplitude of the first acoustic emission signal, the duration is the time length from the start to the end of the first acoustic emission signal, and the rise time is the time from the start of the first acoustic emission signal to the maximum amplitude; A recognition model is trained based on the first feature, and the recognition model is used to identify the category of the target object, which is one of coal, gangue, and coal-gangue mixture.

2. The method according to claim 1, wherein The invention also includes: extracting a second feature and a third feature of the first acoustic emission signal; wherein the second feature includes a Mel frequency cepstral coefficient, and the third feature includes a gammatone frequency cepstral coefficient; Training a recognition model based on the first feature includes: The recognition model is trained based on the first feature, the second feature, and the third feature.

3. The method according to claim 2, wherein: Extracting a second feature and a third feature of the first acoustic emission signal, wherein the second feature includes a Mel-frequency cepstral coefficient, includes: Performing a short-time Fourier transform on the first acoustic emission signal to obtain a spectrum, calculating a power spectrum based on the spectrum, filtering the power spectrum through a Mel filter bank to obtain a second acoustic emission signal, calculating an energy value of the second acoustic emission signal, taking the logarithm of the energy value to obtain a logarithmic Mel spectrum, and performing a discrete cosine transform on the logarithmic Mel spectrum to obtain a Mel-frequency cepstrum coefficient.

4. The method according to claim 2, wherein: Extracting a second feature and a third feature of the first acoustic emission signal, wherein the third feature includes a gammatone frequency cepstral coefficient, includes: Performing a short-time Fourier transform on the first acoustic emission signal to obtain a spectrum, calculating a power spectrum based on the spectrum, filtering the power spectrum through a gammatone filter bank to obtain a third acoustic emission signal, calculating an energy value of the third acoustic emission signal, taking a logarithm of the energy value to obtain a logarithmic gammatone spectrum, and performing a discrete cosine transform on the logarithmic gammatone spectrum to obtain a gammatone frequency cepstral coefficient.

5. The method according to claim 1, wherein The recognition model is a neural network model or a support vector machine model.

6. The method according to claim 1, wherein Preprocessing the original acoustic emission signal to obtain a first acoustic emission signal includes: filtering the original acoustic emission signal through a bandpass filter to obtain the first acoustic emission signal.

7. The method according to claim 1, wherein The method further includes optimizing the installation position of the acoustic emission sensor according to the coal gangue flow law and / or the hydraulic support action law.

8. An electronic device comprising: A processor, a memory, and computer program instructions stored in the memory and executable on the processor, wherein the processor is configured to implement the method according to any one of claims 1 to 7 when executing the computer program instructions.

9. A computer-readable storage medium, wherein: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Images