A method and system for monitoring and early warning of dynamic disasters in coal mines based on acoustic and electrical signals.

Abstract

This invention provides a method and system for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals, belonging to the field of coal mine safety monitoring and early warning. This invention collects signals through acoustic emission sensors and electromagnetic radiation sensors, and adopts modal reconstruction technology to effectively eliminate the interference of noise on the monitoring and early warning results, thereby improving the accuracy of monitoring and early warning. At the same time, it adopts a deep subdomain adaptive network model to accurately capture the precursor signals of coal mine dynamic disasters and issue early warnings, realizing the intelligentization of the monitoring and early warning process and the ability to predict coal mine dynamic disasters.

Description

Technical Field

[0001] This invention relates to the field of coal mine safety monitoring and early warning, and in particular to a method and system for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals. Background Technology

[0002] Coal mines, as an important source of energy resources, occupy a vital position in the national economy. Deep, high ground stress, coupled with disturbances during coal mining, can lead to dynamic disasters such as rock bursts and coal and gas outbursts, threatening the safe production of coal mines.

[0003] Traditional methods for monitoring dynamic disasters in coal mines largely rely on geological exploration and manual observation. These methods, limited by the sensitivity of monitoring equipment and observation conditions, often struggle to capture early warning signs of disasters in a timely and accurate manner. Furthermore, traditional methods typically require significant manpower and, due to their slow response time, fail to provide effective early warnings before accidents occur.

[0004] In recent years, with the development of sensing and information technologies, disaster monitoring using acoustic and electrical signals has become an emerging research direction. During coal and gas outbursts, energy in the coal and rock mass is released in the form of elastic energy, acoustic emission energy, thermal radiation energy, gas expansion energy, and electromagnetic radiation energy. By monitoring various geophysical signals in the mine tunnels during coal mining, the instability of the surrounding rock mass can be determined, enabling accurate prediction of outburst disasters. Acoustic emission and electromagnetic radiation monitoring, due to their non-contact, high-penetration, and high-frequency real-time characteristics, have been widely used for monitoring and early warning of coal and rock dynamic disasters. However, the working environment in coal mine tunnels is complex, and some normal operations can generate interference noise that affects the early warning effect. Furthermore, each tunnel has a large amount of monitoring data to be processed, necessitating an intelligent monitoring method that is unaffected by noise and requires minimal human intervention. Summary of the Invention

[0005] To address the problems in the prior art, this invention provides a method and system for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals. This invention collects signals using acoustic emission sensors and electromagnetic radiation sensors, and employs modal reconstruction technology to effectively eliminate noise interference with the monitoring and early warning results, thereby improving the accuracy of monitoring and early warning. Simultaneously, it utilizes a deep subdomain adaptive network model to accurately capture precursor signals of coal mine dynamic disasters and issue early warnings, realizing intelligent monitoring and early warning processes and the ability to predict coal mine dynamic disasters. To achieve the above objectives, the technical solution is as follows:

[0006] On the one hand, the present invention provides a method for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals, the method comprising:

[0007] S1. Install the acoustic emission sensor and the electromagnetic radiation sensor directly in front of the coal mine working face, and collect the acoustic and electrical signal data of the coal mine through the acoustic emission sensor and the electromagnetic radiation sensor.

[0008] S2. Based on the acoustic and electrical signal dataset of the coal mine, the signal component dataset of the acoustic and electrical signals is obtained through improved fully adaptive noise empirical mode decomposition processing.

[0009] S3. Based on the signal component dataset of the acoustic signal, the modal reconstruction method is used to process it to obtain the reconstructed acoustic signal dataset.

[0010] S4. Convert the reconstructed acoustic-electric signal dataset into a time-frequency image, and use continuous wavelet transform technology to obtain the reconstructed acoustic-electric signal time-frequency image dataset.

[0011] S5. Based on the reconstructed acoustic and electrical signal time-frequency image dataset, the risk of coal mine dynamic disasters is determined by identifying and judging through a deep subdomain adaptive network model.

[0012] Optionally, the acoustic emission sensor and the electromagnetic radiation sensor move with the coal mine working face, and the distance between the acoustic emission sensor and the electromagnetic radiation sensor and the coal mine working face is 10 meters to 20 meters.

[0013] The sampling frequency of the acoustic emission sensor is 0.05 Hz, and the sampling frequency of the electromagnetic radiation sensor is 0.05 Hz.

[0014] Optionally, in S2, based on the acoustic and electrical signal dataset of the coal mine, an improved fully adaptive noise empirical mode decomposition process is used to obtain the signal component dataset of the acoustic and electrical signals, including:

[0015] S21. Based on the acoustic and electrical signal dataset of the coal mine, the first residual of the acoustic and electrical signal is obtained through formula (1).

[0016] (1)

[0017] In the formula, This is the first residual of the acoustic signal. This refers to the signal data in the acoustic and electrical signal dataset of a coal mine. The initial noise figure, For the EMD decomposition operator of the first residual, It is Gaussian white noise with zero mean and unit variance. This is the local mean obtained through EMD calculation;

[0018] S22. Based on the acoustic signal dataset of the coal mine and the first residual of the acoustic signal, the first signal component of the acoustic signal is obtained through formula (2).

[0019] (2)

[0020] In the formula: This is the first signal component of the acoustic-electric signal;

[0021] S23. Based on the first residual of the acoustic signal, the residual dataset of the acoustic signal is obtained by calculating using formula (3).

[0022] (3)

[0023] In the formula, Let k be the k-th residual of the acoustic signal. The noise figure for the (k-1)th residual. For the EMD decomposition operator of the k-th residual, For local average operator, The number of decompositions;

[0024] S24. Based on the residual dataset of the acoustic signal, the signal component dataset of the acoustic signal is obtained through formula (4).

[0025] (4)

[0026] In the formula: Let be the kth signal component of the acoustic-electric signal.

[0027] Optionally, in step S3, the signal component dataset of the acoustic-electric signal is processed using a modal reconstruction method to obtain a reconstructed acoustic-electric signal dataset, including:

[0028] S31. Based on the signal component dataset of the acoustic-electric signal, the phase space vector dataset of the signal components is obtained through formula (5). , , (5)

[0029] In the formula: Let be the phase space vector of the m-dimensional signal component at the i-th sampling point. Let m be the k-th signal component at the i-th sampling point, where m is the vector dimension and n is the sampling time.

[0030] S32. Based on the phase space vector dataset of the signal component, calculate the distance between the phase space vectors of the signal component and compare it with the set tolerance to obtain the vector logarithm dataset where the distance between the phase space vectors of the signal component is less than the set tolerance.

[0031] S33. Based on the logarithmic dataset of vectors whose distance between phase space vectors of the signal component is less than the set tolerance, the tolerance probability dataset of the phase space vectors is obtained through formula (6).

[0032] (6)

[0033] In the formula: Let be the tolerance probability of an m-dimensional phase space vector. The number of vector pairs whose distance between the phase space vectors of the signal components at the i-th sampling point is less than the set tolerance;

[0034] S34. Based on the tolerance probability dataset of the phase space vector, the sample entropy dataset of the signal components is obtained through formula (7).

[0035] (7)

[0036] In the formula: Let r be the sample entropy of the signal component, and r be the set tolerance.

[0037] S35. Based on the sample entropy dataset of the signal component and the signal component dataset of the acoustic signal, select the signal component data of the acoustic signal corresponding to the sample entropy data in the sample entropy dataset of the signal component that are greater than the sample entropy threshold, and obtain the noisy component dataset of the signal component.

[0038] S36. Based on the noisy component dataset of the signal component, the wavelet soft thresholding denoising method is used to perform denoising processing to obtain the denoised component dataset of the signal component.

[0039] S37. The signal component data of the acoustic-electric signal corresponding to the denoised component dataset of the signal component and the sample entropy data of the sample entropy dataset of the signal component that are not greater than the sample entropy threshold are accumulated to obtain the reconstructed acoustic-electric signal dataset.

[0040] Optionally, in step S4, the reconstructed acoustic-electric signal dataset is converted into a time-frequency image, and continuous wavelet transform is used to obtain the reconstructed acoustic-electric signal time-frequency image dataset, including:

[0041] S41. Based on the reconstructed acoustic and electrical signal dataset, a two-dimensional time-frequency image of the acoustic and electrical signal dataset is obtained by using continuous wavelet transform technology.

[0042] S42. Based on the two-dimensional time-frequency image of the acoustic-electric signal dataset, the reconstructed acoustic-electric signal time-frequency image dataset is obtained.

[0043] Optionally, in step S5, based on the reconstructed acoustic-electric signal time-frequency image dataset, a deep subdomain adaptive network model is used for identification and judgment to obtain the risk of coal mine dynamic hazards, including:

[0044] S51. Based on the reconstructed acoustic-electric signal time-frequency image dataset, input it into the deep subdomain adaptive network model to obtain the probability distribution of the reconstructed acoustic-electric signal recognition results;

[0045] S52. Based on the probability distribution of the reconstructed acoustic and electrical signal recognition results, the risk of coal mine dynamic disasters is obtained by summing and normalizing the results.

[0046] Optionally, the deep subdomain adaptive network model includes:

[0047] The feature extraction module is used to extract features from the reconstructed acoustic-electric signal time-frequency image dataset;

[0048] The classification module is used to generate corresponding predicted labels using the features extracted by the feature extraction module from the reconstructed acoustic-electric signal time-frequency image dataset.

[0049] Optionally, the training process of this deep subdomain adaptive network model includes:

[0050] S61. Based on the acoustic emission sensor and the electromagnetic radiation sensor, through steps S1 to S4, multiple data acquisitions and processing are performed to obtain a training dataset.

[0051] S62. Compare the number of acoustic emission signals and electromagnetic radiation signals in the training dataset, select the signals with fewer signals to obtain the source domain signals, and select the signals with more signals to obtain the target domain signals.

[0052] S63. Based on the source domain signal, determine the normal signal or the abnormal precursor signal of the source domain signal by using the signal identification criteria.

[0053] S64. Based on the normal signal or the abnormal precursor signal of the source domain signal, the risk of coal mine dynamic disasters in the training dataset is obtained by judgment.

[0054] S65. Input the source domain signal and the target domain signal into the initialized deep subdomain adaptive network model, and train it by constructing a loss function to obtain the deep subdomain adaptive network model during the training process and the risk of coal mine dynamic disasters output by the model.

[0055] S66. If the risk of coal mine dynamic disasters in the training dataset is consistent with the risk of coal mine dynamic disasters output by the model, save the model parameters and obtain the parameters of the deep subdomain adaptive network model.

[0056] S67. Update the parameters of the deep subdomain adaptive network model to the deep subdomain adaptive network model during the training process to obtain the deep subdomain adaptive network model.

[0057] Optionally, the signal identification criteria include:

[0058] Normal acoustic and electrical signals: The main frequency of the signal is 0~0.01Hz, the high amplitude point is located in the 0~0.01Hz region, and there is no high amplitude point in the 0.01~0.25Hz region;

[0059] Abnormal precursor signals of acoustic and electrical signals: High amplitude points exist in the main frequency range of 0~0.25Hz, 0~0.01Hz and 0.01~0.25Hz.

[0060] On the other hand, the present invention provides a coal mine dynamic disaster monitoring and early warning system based on acoustic and electrical signals. This system is applied to a coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals. The system includes:

[0061] The signal acquisition module is used to install an acoustic emission sensor and an electromagnetic radiation sensor directly in front of the coal mine working face, and to acquire acoustic and electrical signal datasets of the coal mine through the acoustic emission sensor and the electromagnetic radiation sensor.

[0062] The mode decomposition module is used to obtain the signal component dataset of the acoustic and electrical signals from the acoustic and electrical signal dataset of the coal mine through improved fully adaptive noise empirical mode decomposition processing.

[0063] The modal reconstruction module is used to process the signal component dataset of the acoustic-electric signal using the modal reconstruction method to obtain the reconstructed acoustic-electric signal dataset.

[0064] The signal conversion module is used to convert the reconstructed acoustic-electric signal dataset into a time-frequency image. It uses continuous wavelet transform technology to obtain the reconstructed acoustic-electric signal time-frequency image dataset.

[0065] The hazard assessment module is used to identify and assess the risk of coal mine dynamic disasters based on the reconstructed acoustic and electrical signal time-frequency image dataset through a deep subdomain adaptive network model.

[0066] Compared with the prior art, the technical solution of the present invention has at least the following beneficial effects:

[0067] The above scheme, on the one hand, collects signals through acoustic emission sensors and electromagnetic radiation sensors, and uses modal reconstruction technology to effectively eliminate the interference of noise on the monitoring and early warning results, thereby improving the accuracy of monitoring and early warning. On the other hand, it adopts a deep subdomain adaptive network model to accurately capture the precursor signals of coal mine dynamic disasters and issue early warnings, realizing the intelligentization of the monitoring and early warning process and the ability to predict coal mine dynamic disasters. Attached Figure Description

[0068] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0069] Figure 1 This is a flowchart of an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention;

[0070] Figure 2 This is a flowchart illustrating the acquisition of the signal component dataset of acoustic and electrical signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention.

[0071] Figure 3 This is a flowchart illustrating the process of obtaining the reconstructed acoustic and electrical signal dataset in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention.

[0072] Figure 4 This is a flowchart illustrating the process of obtaining the reconstructed time-frequency image dataset of acoustic and electrical signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention.

[0073] Figure 5 This is a flowchart illustrating the process of obtaining the hazard of coal mine dynamic disasters in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention.

[0074] Figure 6 This is a flowchart illustrating the training process of the deep subdomain adaptive network model in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention.

[0075] Figure 7 This is a schematic diagram of the acoustic emission signal decomposition and processing results in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention;

[0076] Figure 8 This is a schematic diagram of the electromagnetic radiation signal decomposition and processing results in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention;

[0077] Figure 9 This is a comparative schematic diagram showing the results of modal reconstruction processing of acoustic emission signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention;

[0078] Figure 10 This is a comparative schematic diagram showing the results of modal reconstruction processing of electromagnetic radiation signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention;

[0079] Figure 11 This is a schematic diagram of a normal acoustic emission signal in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention;

[0080] Figure 12 This is a schematic diagram of a normal electromagnetic radiation signal in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention;

[0081] Figure 13 This is a schematic diagram of abnormal precursor signals of acoustic emission signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention;

[0082] Figure 14 This is a schematic diagram of abnormal precursor signals of electromagnetic radiation signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention;

[0083] Figure 15 This is a system block diagram of an embodiment of the coal mine dynamic disaster monitoring and early warning system based on acoustic and electrical signals of the present invention. Detailed Implementation

[0084] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0085] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0086] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0087] like Figure 1The flowchart shown is an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention. The present invention provides a coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals, which is implemented by a coal mine dynamic disaster monitoring and early warning system based on acoustic and electrical signals. The method includes:

[0088] S1. Install the acoustic emission sensor and the electromagnetic radiation sensor directly in front of the coal mine working face, and collect the acoustic and electrical signal data of the coal mine through the acoustic emission sensor and the electromagnetic radiation sensor.

[0089] Specifically, the acoustic emission sensor and the electromagnetic radiation sensor move with the coal mine working face, and the distance between the acoustic emission sensor and the electromagnetic radiation sensor and the coal mine working face is 10 meters to 20 meters.

[0090] The sampling frequency of the acoustic emission sensor is 0.05 Hz, and the sampling frequency of the electromagnetic radiation sensor is 0.05 Hz.

[0091] The antenna of the electromagnetic radiation sensor should be oriented towards the coal mine working face and parallel to the coal seam strike. The acoustic emission sensor is installed on the support anchor bolts of the coal mine working face. The sampling frequencies of the acoustic emission sensor and the electromagnetic radiation sensor can be changed according to the on-site data storage requirements.

[0092] S2. Based on the acoustic and electrical signal dataset of the coal mine, the signal component dataset of the acoustic and electrical signals is obtained through improved fully adaptive noise empirical mode decomposition processing.

[0093] Specifically, such as Figure 2 The flowchart shown in this embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals illustrates the process of obtaining the signal component dataset of acoustic and electrical signals. In step S2, based on the acoustic and electrical signal dataset of the coal mine, an improved fully adaptive noise empirical mode decomposition process is used to obtain the signal component dataset of the acoustic and electrical signals, including:

[0094] S21. Based on the acoustic and electrical signal dataset of the coal mine, the first residual of the acoustic and electrical signal is obtained through formula (1).

[0095] (1)

[0096] In the formula, This is the first residual of the acoustic signal. This refers to the signal data in the acoustic and electrical signal dataset of a coal mine. The initial noise figure, For the EMD decomposition operator of the first residual, It is Gaussian white noise with zero mean and unit variance. This is the local mean obtained through EMD calculation;

[0097] S22. Based on the acoustic signal dataset of the coal mine and the first residual of the acoustic signal, the first signal component of the acoustic signal is obtained through formula (2).

[0098] (2)

[0099] In the formula: This is the first signal component of the acoustic-electric signal;

[0100] S23. Based on the first residual of the acoustic signal, the residual dataset of the acoustic signal is obtained by calculating using formula (3).

[0101] (3)

[0102] In the formula, Let k be the k-th residual of the acoustic signal. The noise figure for the (k-1)th residual. For the EMD decomposition operator of the k-th residual, For local average operator, The number of decompositions;

[0103] Furthermore, the number of decompositions It can be obtained based on the following conditions; satisfying any one of these conditions is sufficient.

[0104] Condition 1: In the formula: for The second norm;

[0105] Condition 2: There are no local extrema in the middle.

[0106] like Figure 7 The diagram shown is a schematic representation of the acoustic emission signal decomposition and processing results in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention. The acoustic emission signal decomposition and processing results contain 10 signal components. The order of magnitude of the 2-norm is ;

[0107] like Figure 8 The diagram shown is a schematic representation of the electromagnetic radiation signal decomposition and processing results in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention. The electromagnetic radiation signal decomposition and processing results contain nine signal components. The order of magnitude of the 2-norm is .

[0108] S24. Based on the residual dataset of the acoustic signal, the signal component dataset of the acoustic signal is obtained through formula (4).

[0109] (4)

[0110] In the formula: Let be the kth signal component of the acoustic-electric signal.

[0111] Furthermore, For a set of time series data: .

[0112] S3. Based on the signal component dataset of the acoustic signal, the modal reconstruction method is used to process it to obtain the reconstructed acoustic signal dataset.

[0113] Specifically, such as Figure 3 The flowchart shown in this embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention illustrates the process of obtaining the reconstructed acoustic and electrical signal dataset. In step S3, the signal component dataset of the acoustic and electrical signals is processed using a modal reconstruction method to obtain the reconstructed acoustic and electrical signal dataset, which includes:

[0114] S31. Based on the dataset of the acoustic-electric signal, the phase space vector dataset of the signal components is obtained through formula (5). , , (5)

[0115] In the formula: Let be the phase space vector of the m-dimensional signal component at the i-th sampling point. Let m be the k-th signal component at the i-th sampling point, where m is the vector dimension and n is the sampling time.

[0116] S32. Based on the phase space vector dataset of the signal component, calculate the distance between the phase space vectors of the signal component and compare it with the set tolerance to obtain the vector logarithm dataset where the distance between the phase space vectors of the signal component is less than the set tolerance.

[0117] S33. Based on the logarithmic dataset of vectors whose distance between phase space vectors of the signal component is less than the set tolerance, the tolerance probability dataset of the phase space vectors is obtained through formula (6).

[0118] (6)

[0119] In the formula: Let be the tolerance probability of an m-dimensional phase space vector. The number of vector pairs whose distance between the phase space vectors of the signal components at the i-th sampling point is less than the set tolerance;

[0120] S34. Based on the tolerance probability dataset of the phase space vector, the sample entropy dataset of the signal components is obtained through formula (7).

[0121] (7)

[0122] In the formula: Let r be the sample entropy of the signal component, and r be the set tolerance.

[0123] S35. Based on the sample entropy dataset of the signal component and the signal component dataset of the acoustic signal, select the signal component data of the acoustic signal corresponding to the sample entropy data in the sample entropy dataset of the signal component that are greater than the sample entropy threshold, and obtain the noisy component dataset of the signal component.

[0124] Furthermore, such as Figure 9 The diagram shows a comparison of the modal reconstruction processing results of acoustic emission signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention, and as shown in the figure. Figure 10 The diagram shows a comparison of the modal reconstruction results of electromagnetic radiation signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention. The selection of the sample entropy threshold will affect the denoising result of the reconstructed signal. If the reconstructed signal has a lot of noise components, the sample entropy threshold can be reduced relatively. In this embodiment, the sample entropy threshold is 0.1.

[0125] S36. Based on the noisy component dataset of the signal component, the wavelet soft thresholding denoising method is used to perform denoising processing to obtain the denoised component dataset of the signal component.

[0126] S37. The signal component data of the acoustic-electric signal corresponding to the denoised component dataset of the signal component and the sample entropy data of the sample entropy dataset of the signal component that are not greater than the sample entropy threshold are accumulated to obtain the reconstructed acoustic-electric signal dataset.

[0127] S4. Convert the reconstructed acoustic-electric signal dataset into a time-frequency image, and use continuous wavelet transform technology to obtain the reconstructed acoustic-electric signal time-frequency image dataset.

[0128] Specifically, such as Figure 4 The flowchart shown in the embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention illustrates the process of obtaining the reconstructed acoustic and electrical signal time-frequency image dataset. In step S4, the reconstructed acoustic and electrical signal dataset is converted into a time-frequency image using continuous wavelet transform technology, resulting in the reconstructed acoustic and electrical signal time-frequency image dataset, including:

[0129] S41. Based on the reconstructed acoustic and electrical signal dataset, a two-dimensional time-frequency image of the acoustic and electrical signal dataset is obtained by using continuous wavelet transform technology.

[0130] S42. Based on the two-dimensional time-frequency image of the acoustic-electric signal dataset, the reconstructed acoustic-electric signal time-frequency image dataset is obtained.

[0131] S5. Based on the reconstructed acoustic and electrical signal time-frequency image dataset, the risk of coal mine dynamic disasters is determined by identifying and judging through a deep subdomain adaptive network model.

[0132] Specifically, such as Figure 5 The flowchart shown in the embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention obtains the hazard of coal mine dynamic disasters. In step S5, the hazard of coal mine dynamic disasters is obtained by identification and judgment through a deep subdomain adaptive network model based on the reconstructed acoustic and electrical signal time-frequency image dataset, including:

[0133] S51. Based on the reconstructed acoustic-electric signal time-frequency image dataset, input it into the deep subdomain adaptive network model to obtain the probability distribution of the reconstructed acoustic-electric signal recognition results;

[0134] This deep subdomain adaptive network model includes:

[0135] The feature extraction module is used to extract features from the reconstructed acoustic-electric signal time-frequency image dataset;

[0136] The classification module is used to generate corresponding predicted labels using the features extracted by the feature extraction module from the reconstructed acoustic-electric signal time-frequency image dataset.

[0137] like Figure 6 The flowchart shown is a presentation of the training process of a deep subdomain adaptive network model in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention. The training process of the deep subdomain adaptive network model includes:

[0138] S61. Based on the acoustic emission sensor and the electromagnetic radiation sensor, through steps S1 to S4, multiple data acquisitions and processing are performed to obtain a training dataset.

[0139] S62. Compare the number of acoustic emission signals and electromagnetic radiation signals in the training dataset, select the signals with fewer signals to obtain the source domain signals, and select the signals with more signals to obtain the target domain signals.

[0140] S63. Based on the source domain signal, determine the normal signal or the abnormal precursor signal of the source domain signal by using the signal identification criteria.

[0141] Furthermore, such as Figure 11 The diagram shown is a normal signal diagram of the acoustic emission signal in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention. Figure 12The diagram shown is a normal signal diagram of electromagnetic radiation signal in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention. Figure 13 The diagram shown illustrates abnormal precursor signals of acoustic emission signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention, and as shown in the figure. Figure 14 The diagram shown illustrates an abnormal precursor signal of electromagnetic radiation signals in an embodiment of the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals of the present invention. The identification criteria for this signal include:

[0142] Normal acoustic and electrical signals: The main frequency of the signal is 0~0.01Hz, the high amplitude point is located in the 0~0.01Hz region, and there is no high amplitude point in the 0.01~0.25Hz region;

[0143] Abnormal precursor signals of acoustic and electrical signals: High amplitude points exist in the main frequency range of 0~0.25Hz, 0~0.01Hz and 0.01~0.25Hz.

[0144] S64. Based on the normal signal or the abnormal precursor signal of the source domain signal, the risk of coal mine dynamic disasters in the training dataset is obtained by judgment.

[0145] S65. Input the source domain signal and the target domain signal into the initialized deep subdomain adaptive network model, and train it by constructing a loss function to obtain the deep subdomain adaptive network model during the training process and the risk of coal mine dynamic disasters output by the model.

[0146] Furthermore, the formula for calculating this loss function is as follows:

[0147] (8)

[0148] (9)

[0149] (10)

[0150] In the formula, The prediction loss for the classification module, Let cross-entropy be the loss function. The predicted labels generated for the classification module To be the true label of the source domain signal, For subdomain alignment differences, and The source domain signal features extracted by the model, and The target domain signal features extracted by the model. For Gaussian kernel function, for Weight, For adaptation factors;

[0151] The calculation formula is as follows:

[0152] (11)

[0153] By adjusting Weight and adaptation factors The value of this parameter can control the convergence direction of the deep subdomain adaptive network model. If during the convergence process... Difficult to decrease, can increase Or change The calculation formula.

[0154] S66. If the risk of coal mine dynamic disasters in the training dataset is consistent with the risk of coal mine dynamic disasters output by the model, save the model parameters and obtain the parameters of the deep subdomain adaptive network model.

[0155] S67. Update the parameters of the deep subdomain adaptive network model to the deep subdomain adaptive network model during the training process to obtain the deep subdomain adaptive network model.

[0156] S52. Based on the probability distribution of the reconstructed acoustic and electrical signal recognition results, the risk of coal mine dynamic disasters is obtained by summing and normalizing the results.

[0157] Furthermore, in this embodiment, the identification result of the acoustic emission signal is [abnormal precursor signal, normal signal] = The identification result of the electromagnetic radiation signal is [abnormal precursor signal, normal signal] = Therefore, the probability of a coal mine dynamic disaster is 0.997460275, which is consistent with the coal mine site records.

[0158] like Figure 15 The diagram shown is a system block diagram of an embodiment of the coal mine dynamic disaster monitoring and early warning system based on acoustic and electrical signals of the present invention. The present invention provides a coal mine dynamic disaster monitoring and early warning system based on acoustic and electrical signals. This system is applied to a coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals. The system includes a signal acquisition module, a mode decomposition module, a mode reconstruction module, a signal conversion module, and a hazard judgment module. Specifically,

[0159] The signal acquisition module is used to install an acoustic emission sensor and an electromagnetic radiation sensor directly in front of the coal mine working face, and to acquire acoustic and electrical signal datasets of the coal mine through the acoustic emission sensor and the electromagnetic radiation sensor.

[0160] The mode decomposition module is used to obtain the signal component dataset of the acoustic and electrical signals from the acoustic and electrical signal dataset of the coal mine through improved fully adaptive noise empirical mode decomposition processing.

[0161] The modal reconstruction module is used to process the signal component dataset of the acoustic-electric signal using the modal reconstruction method to obtain the reconstructed acoustic-electric signal dataset.

[0162] The signal conversion module is used to convert the reconstructed acoustic-electric signal dataset into a time-frequency image. It uses continuous wavelet transform technology to obtain the reconstructed acoustic-electric signal time-frequency image dataset.

[0163] The hazard assessment module is used to identify and assess the risk of coal mine dynamic disasters based on the reconstructed acoustic and electrical signal time-frequency image dataset through a deep subdomain adaptive network model.

[0164] This invention provides a method and system for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals. The invention collects signals through acoustic emission sensors and electromagnetic radiation sensors, and adopts modal reconstruction technology to effectively eliminate the interference of noise on the monitoring and early warning results, thereby improving the accuracy of monitoring and early warning. At the same time, it adopts a deep subdomain adaptive network model to accurately capture the precursor signals of coal mine dynamic disasters and issue early warnings, realizing the intelligentization of the monitoring and early warning process and the ability to predict coal mine dynamic disasters.

[0165] It is understood that the present invention has been described through the above embodiments and should not be construed as limiting the implementation and scope of the present invention. Those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the present invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A method for monitoring and early warning of dynamic disasters in coal mines based on acoustic and electrical signals, characterized in that, The method includes: S1. Install the acoustic emission sensor and the electromagnetic radiation sensor directly in front of the coal mine working face, and collect the acoustic and electrical signal data of the coal mine through the acoustic emission sensor and the electromagnetic radiation sensor. S2. Based on the acoustic and electrical signal dataset of the coal mine, decompose it using an improved fully adaptive noise empirical mode decomposition to obtain the signal component dataset of the acoustic and electrical signals. S3. Based on the signal component dataset of the acoustic-electric signal, the modal reconstruction method is used to process it to obtain the reconstructed acoustic-electric signal dataset. S4. The reconstructed acoustic-electric signal dataset is converted into a time-frequency image, and the reconstructed acoustic-electric signal time-frequency image dataset is obtained by using continuous wavelet transform technology. S5. Based on the reconstructed acoustic and electrical signal time-frequency image dataset, the risk of coal mine dynamic disasters is determined by identification and judgment through a deep subdomain adaptive network model. Wherein, S2 includes: S21. Based on the acoustic and electrical signal dataset of the coal mine, the first residual of the acoustic and electrical signal is obtained using formula (1). （1） In the formula, This is the first residual of the acoustic signal. This refers to the signal data in the acoustic and electrical signal dataset of a coal mine. The initial noise figure, For the EMD decomposition operator of the first residual, It is Gaussian white noise with zero mean and unit variance. This is the local mean obtained through EMD calculation; S22. Based on the acoustic and electrical signal dataset of the coal mine and the first residual of the acoustic and electrical signal, the first signal component of the acoustic and electrical signal is obtained through formula (2). （2） In the formula: This is the first signal component of the acoustic-electric signal; S23. Based on the first residual of the acoustic signal, the residual dataset of the acoustic signal is obtained by formula (3). （3） In the formula, The first sound signal k One residual, For the first k-1 The noise figure of each residual. For the first k EMD decomposition operator for each residual, For local average operator, The number of decompositions; S24. Based on the residual dataset of the acoustic-electric signal, the signal component dataset of the acoustic-electric signal is obtained using formula (4). （4） In the formula: The first sound signal k One signal component; Wherein, S3 includes: S31. Based on the signal component dataset of the acoustic-electric signal, obtain the phase space vector dataset of the signal components using formula (5). ， ， （5） In the formula: For the first i Each sampling point at time m The phase space vector of the 3D signal component. For the first i The sampling point at time n k Each signal component m For vector dimensions, n Sampling time; S32. Based on the phase space vector dataset of the signal components, calculate the distance between the phase space vectors of the signal components and compare it with a set tolerance to obtain a vector logarithm dataset where the distance between the phase space vectors of the signal components is less than the set tolerance. S33. Based on the vector logarithm dataset where the distance between the phase space vectors of the signal components is less than the set tolerance, obtain the tolerance probability dataset of the phase space vectors using formula (6). （6） In the formula: for m Tolerance probability of a 3D phase space vector For the first i The distance between the phase space vectors of the signal components at each sampling point is less than the number of vector logs with the set tolerance. S34. Based on the tolerance probability dataset of the phase space vector, obtain the sample entropy dataset of the signal components using formula (7). （7） In the formula: The sample entropy of the signal component. r To set tolerances; S35. Based on the sample entropy dataset of the signal component and the signal component dataset of the acoustic signal, select the signal component data of the acoustic signal corresponding to the sample entropy data in the sample entropy dataset of the signal component that is greater than the sample entropy threshold, and obtain the noisy component dataset of the signal component. S36. Based on the noisy component dataset of the signal components, the wavelet soft thresholding denoising method is used to perform denoising processing to obtain the denoised component dataset of the signal components. S37. The signal component data of the acoustic-electric signal corresponding to the denoised component dataset of the signal component and the sample entropy data of the signal component that are not greater than the sample entropy threshold are accumulated to obtain the reconstructed acoustic-electric signal dataset.

2. The method for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals according to claim 1, characterized in that, The acoustic emission sensor and the electromagnetic radiation sensor move with the coal mine working face, and the distance between the acoustic emission sensor and the electromagnetic radiation sensor and the coal mine working face is 10 meters to 20 meters. The sampling frequency of the acoustic emission sensor is 0.05 Hz, and the sampling frequency of the electromagnetic radiation sensor is 0.05 Hz.

3. The method for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals according to claim 1, characterized in that, In step S4, the reconstructed acoustic-electric signal dataset is converted into a time-frequency image. Continuous wavelet transform is used to obtain the reconstructed acoustic-electric signal time-frequency image dataset, including: S41. Based on the reconstructed acoustic and electrical signal dataset, a two-dimensional time-frequency image of the acoustic and electrical signal dataset is obtained by using continuous wavelet transform technology. S42. Based on the two-dimensional time-frequency image of the acoustic-electric signal dataset, obtain the reconstructed acoustic-electric signal time-frequency image dataset.

4. The method for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals according to claim 1, characterized in that, In step S5, based on the reconstructed acoustic-electric signal time-frequency image dataset, a deep subdomain adaptive network model is used for identification and judgment to obtain the risk of coal mine dynamic disasters, including: S51. Based on the reconstructed acoustic-electric signal time-frequency image dataset, input it into the deep subdomain adaptive network model to obtain the probability distribution of the reconstructed acoustic-electric signal recognition results; S52. Based on the probability distribution of the reconstructed acoustic and electrical signal recognition results, the risk of coal mine dynamic disasters is obtained by summing and normalizing the results.

5. The method for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals according to claim 1, characterized in that, The deep subdomain adaptive network model includes: The feature extraction module is used to extract features from the reconstructed acoustic-electric signal time-frequency image dataset; The classification module is used to generate corresponding predicted labels using the features of the reconstructed acoustic-electric signal time-frequency image dataset extracted by the feature extraction module.

6. The method for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals according to claim 1, characterized in that, The training process of the deep subdomain adaptive network model includes: S61. Based on the acoustic emission sensor and the electromagnetic radiation sensor, through steps S1 to S4, multiple data acquisitions and processing are performed to obtain a training dataset. S62. Compare the number of acoustic emission signals and electromagnetic radiation signals in the training dataset, select the signal with fewer signals to obtain the source domain signal, and select the signal with more signals to obtain the target domain signal. S63. Based on the source domain signal, determine the normal signal or the abnormal precursor signal of the source domain signal by using the signal identification criteria. S64. Based on the normal signal of the source domain signal or the abnormal precursor signal of the source domain signal, the risk of coal mine dynamic disasters in the training dataset is obtained by judgment. S65. Input the source domain signal and the target domain signal into the initialized deep subdomain adaptive network model, and train it by constructing a loss function to obtain the deep subdomain adaptive network model during the training process and the risk of coal mine dynamic disasters output by the model. S66. If the risk of coal mine dynamic disasters in the training dataset is consistent with the risk of coal mine dynamic disasters output by the model, save the model parameters to obtain the parameters of the deep subdomain adaptive network model. S67. Update the parameters of the deep subdomain adaptive network model to the deep subdomain adaptive network model during the training process to obtain the deep subdomain adaptive network model.

7. The method for monitoring and early warning of coal mine dynamic disasters based on acoustic and electrical signals according to claim 6, characterized in that, The signal identification criteria include: Normal acoustic and electrical signals: The main frequency of the signal is 0~0.01Hz, the high amplitude point is located in the 0~0.01Hz region, and there is no high amplitude point in the 0.01~0.25Hz region; Abnormal precursor signals of acoustic and electrical signals: High amplitude points exist in the main frequency range of 0~0.25Hz, 0~0.01Hz and 0.01~0.25Hz.

8. A coal mine dynamic disaster monitoring and early warning system based on acoustic and electrical signals, used to implement the coal mine dynamic disaster monitoring and early warning method based on acoustic and electrical signals as described in any one of claims 1-7, characterized in that, The system includes: The signal acquisition module is used to install acoustic emission sensors and electromagnetic radiation sensors directly in front of the coal mine working face, and to acquire acoustic and electrical signal datasets of the coal mine through the acoustic emission sensors and the electromagnetic radiation sensors. The mode decomposition module is used to obtain the signal component dataset of the acoustic and electrical signals based on the acoustic and electrical signal dataset of the coal mine through improved fully adaptive noise empirical mode decomposition processing. The modal reconstruction module is used to process the signal component dataset of the acoustic-electric signal using a modal reconstruction method to obtain the reconstructed acoustic-electric signal dataset. The signal conversion module is used to convert the reconstructed acoustic-electric signal dataset into a time-frequency image by using continuous wavelet transform technology to obtain the reconstructed acoustic-electric signal time-frequency image dataset. The hazard assessment module is used to identify and assess the risk of coal mine dynamic disasters based on the reconstructed acoustic and electrical signal time-frequency image dataset through a deep subdomain adaptive network model.

Images