Intelligent prediction methods, equipment and media for mine roadway deformation

By deploying microseismic and deformation sensors around mine roadways, and combining microseismic coding blocks and long short-term memory neural networks, the problem of insufficient information density in mine roadway deformation prediction was solved, and faster and more accurate deformation prediction was achieved.

CN120541398BActive Publication Date: 2026-11-17INSPUR SOFTWARE CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202510593203.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-05-09
Publication Date
2026-11-17
Estimated Expiration
2045-05-09

AI Technical Summary

Technical Problem

Existing technologies for predicting deformation in mine roadways suffer from insufficient information density, resulting in inadequate prediction accuracy and speed.

Method used

Microseismic sensors and deformation sensors are deployed around the mine roadways. By using microseismic coding blocks and multi-source heterogeneous data decoders, combined with long short-term memory neural networks, microseismic and deformation data are processed to characterize the damage to the surrounding rock of the roadways and predict future deformation.

Benefits of technology

It improves information density, enables faster and more accurate prediction of mine roadway deformation, provides more useful information, and enhances the accuracy and speed of prediction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120541398B_ABST
    Figure CN120541398B_ABST
Patent Text Reader

Abstract

This invention discloses an intelligent prediction method, equipment, and medium for mine roadway deformation, belonging to the fields of geotechnical engineering testing, artificial intelligence, and the Internet of Things. The technical problem this invention aims to solve is how to improve the information density of mine roadways, thereby enhancing the accuracy and speed of roadway deformation prediction. The technical solution adopted is as follows: deploying microseismic sensors and deformation sensors around the mine roadway, with both sensors having real-time data transmission capabilities; transmitting the collected microseismic data and roadway deformation data to a database based on the type and location of the microseismic and deformation sensors; processing and extracting features from the microseismic data to obtain damage characterization of the surrounding rock of the mine roadway; wherein, damage characterization of the surrounding rock of the mine roadway refers to resolving the microseismic data into fracture size, magnitude, energy, and fracture type.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the fields of geotechnical engineering testing, artificial intelligence, and the Internet of Things, specifically to a method, equipment, and medium for intelligent prediction of deformation in mine roadways. Background Technology

[0002] In recent years, deep learning methods have achieved widespread success in many fields, inspiring more scholars to use them to predict rock deformation processes. The essence of deep learning is a data-driven approach; its learning process can be viewed as starting with historical data and allowing the computer to find a potentially better representation from the data samples that is more conducive to computation.

[0003] Currently, common deep learning-based algorithms for predicting rock deformation and failure directly input deformation or load information during rock failure as time-series signals, using recurrent neural networks as deep learning models to predict the trend of rock deformation over a future period. Essentially, this algorithm infers the constitutive relationship of the rock from historical data, implicitly storing it as tensors in the model parameters, and using these parameters to fit the time-deformation relationship of the rock material. However, when using this strategy, due to the enormous number of parameters in the deep learning model and the limited information provided by a single strain data point, overfitting or underfitting often occurs. Furthermore, these methods, starting only from single deformation data points, cannot reflect the internal damage conditions of the rock.

[0004] Therefore, how to improve the information density of mine roadways, thereby enhancing the accuracy and speed of mine roadway deformation prediction, is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] The technical objective of this invention is to provide a method, device, and medium for intelligent prediction of mine roadway deformation, in order to address the problem of how to improve the information density of mine roadways, thereby enhancing the accuracy and speed of mine roadway deformation prediction.

[0006] The technical objective of this invention is achieved as follows: a method for intelligent prediction of deformation in mine roadways, the specific method of which is as follows: Microseismic sensors and deformation sensors are deployed around the mine roadways, and both microseismic sensors and deformation sensors have real-time data transmission capabilities; the number of microseismic sensors shall not be less than eight. Based on the type and location of the microseismic sensors and deformation sensors, the collected microseismic data and tunnel deformation data are transmitted to the database; Data processing and feature extraction are performed on microseismic data to obtain damage characterization of the surrounding rock of mine roadways; the damage characterization of the surrounding rock of mine roadways refers to resolving microseismic data into fracture size, magnitude, energy and fracture type. The damage characterization of the surrounding rock of the mine roadway, which represents the microseismic signal, is transformed into a microseismic characterization by a microseismic coding block. The microseismic characterization and roadway deformation data are then processed by a multi-source heterogeneous data decoder to predict the changes in mine roadway deformation over a future period. The microseismic characterization reflects the damage level of the rock within the current time window.

[0007] As a preferred approach, the microseismic data is processed and features are extracted to obtain the damage characterization of the surrounding rock in the mine roadway, as follows: Microseismic Information Characterization: Calculate the radius of the microseismic source separately Magnitude and energy The failure characteristic parameters are obtained, and the rupture type of the microseismic event is determined based on the moment tensor decomposition method. Then, the characteristic vector of the microseismic event is defined. ; Microseismic feature normalization: The set of feature vectors λ is Λ. Λ is normalized to obtain the normalized sequence of microseismic feature vectors. Microseismic transformation: The normalized microseismic feature vector sequence is transformed into a microseismic representation by using a microseismic coding block based on a long short-term memory neural network.

[0008] More specifically, the microseismic information characterization is as follows: Calculate the source radius of the microseismic events respectively. Magnitude and energy The destructive characteristic parameters are given by the following formula: ; ; ; in, Brune constant, ; Indicates the shear wave velocity; This represents the corner frequency of the acoustic emission signal. For indoor acoustic emission tests, Replaced by the peak frequency of the signal. For the first The maximum amplitude of the acoustic emission signal waveform received by the micro-vibration sensor; Indicates acoustic emission events and the first The distance between the micro-vibration sensors is in mm; This indicates the number of microseismic sensors used for amplitude calculation. For the first The absolute energy of the waveform received by each micro-vibration sensor is obtained by integrating the waveform envelope; Based on the moment tensor decomposition method, the rupture type of microseismic events is determined using the following formula: ; in, , For tensor The three eigenvalues ​​represent the maximum couple; when When the earthquake source is in a shear-dominant state, it is determined to be a shear failure; when At that time, the earthquake source was in a mixed state between shear and tension; At that time, the epicenter was determined to be a tensile failure; Suppose At what moment does the microseismic probe receive the microseismic event? Through microseismic events Inverse magnitude calculation ,radius ,energy Information on the type of rupture (e.g., shear-type or tension-type acoustic emission events). microseismic events The corresponding eigenvector Characterized as .

[0009] More preferably, the microseismic characteristics are normalized as follows: Will The data is distributed on a standard distribution with a mean of 0 and a variance of 1, as shown in the following formula: ; in, Indicates the translation coefficient; Indicates the scaling factor; Introducing the retranslation coefficient and rescaling factor Two new learnable variables, further study of the variables The transformation is performed, and the formula is as follows: ; After normalization, the microseismic coding process is represented as follows: ; in, This represents the input variable, and then the coefficients are shifted. and rescaling factor Learning is performed using gradient descent.

[0010] More preferably, the microseismic coding is as follows: The normalized microseismic feature vector sequence is used as input, and the microseismic feature vector sequence is converted into a local microseismic hidden state sequence using a long short-term memory neural network. By fusing global microseismic latent states, local microseismic latent states are converted into global microseismic latent states through summation and averaging. The latent state is transformed into a microseismic characterization through microseismic characterization transformation.

[0011] More preferably, the microseismic coding block implementation method based on long short-term memory neural network is as follows: It has For the microseismic feature tensor that has already been normalized by the feature layer, when processing the feature vector... At the same time, the characteristic vector of the previous microseismic event is processed again. The resulting hidden variables Then, within the current time step, the hidden variables The calculation formula is as follows: ; To enable the network to process microseismic feature vector sequences of arbitrary length, two artificially defined vectors representing the start and end of the sequence are introduced, denoted as [vectors to be inserted here]. and ;in, As the initial hidden state of a long short-term memory neural network layer, and Common input model Long Short-Term Memory Neural Network Layer, Vector It is a zero vector with the same shape as the other hidden states; This indicates that there are no other microseismic feature vectors within the current processing time window, and the microseismic coding block receives... When used as output, Together with the last hidden state before Input the data into a Long Short-Term Memory (LSTM) neural network layer and calculate the last hidden state. , The vector is a microseismic feature vector with a shape similar to the input. Given zero vectors with the same shape, the hidden states in the microseismic coding block are calculated using the following formula: ; Through continuous iteration, the Long Short-Term Memory (LSTM) neural network layer acquires a series of hidden states. ,variable Each microseismic event is considered as a local time window representation of the event. To obtain a microseismic event representation that can characterize the intensity of the event at the global level, global microseismic feature fusion is performed. The microseismic feature fusion first averages the hidden states of each independent local microseismic feature to obtain the global microseismic feature hidden state. And then Input to a multilayer perceptron to obtain the global microseismic encoding vector. The formula is as follows: ; .

[0012] As a preferred method, the damage characterization of the surrounding rock of the mine roadway, representing the microseismic signal, is transformed into a microseismic characterization using a microseismic coding block. The microseismic characterization and roadway deformation data are then processed by a multi-source heterogeneous data decoder to predict the changes in mine roadway deformation over a future period, as detailed below: Set in time window Inside, microseismic sensors detected microseismic events. Meanwhile, deformation sensors monitor continuous rock deformation data, and deformation sequences are obtained by sampling at equal intervals. ,in ; Deformation sequence Written in vector form, denoted as ; where vector The subscript indicates the sequence number of the time window. This indicates that the corresponding time window is the time window currently being monitored; Once a micro-seismic event occurs Microseismic characterization is converted from microseismic coding blocks. Characterization using microseismic events and deformation vector Predict the deformation vector for several subsequent time windows. .

[0013] Even better, once a microseismic event occurs... Microseismic characterization is converted from microseismic coding blocks. Characterization using microseismic events and deformation vector Predict the deformation vector for several subsequent time windows. Specifically as follows: The detected deformation vector Along with microseismic characterization As input, the modified sequence for the next time window is obtained from the Long Short-Term Memory neural network layer. and hidden state ; Long Short-Term Memory (LSTM) neural networks iteratively generate prediction time windows. Deformation vector in and hidden state and along with microseismic characterization Input the long short-term memory neural network to predict the first... Deformation vector of each time window and hidden state The process continues until the Long Short-Term Memory (LSTM) neural network determines that it cannot predict the information in the next time window based on the current information and outputs an end marker. .

[0014] An electronic device includes: a memory and at least one processor; The memory contains computer programs; The at least one processor executes the computer program stored in the memory, causing the at least one processor to perform the intelligent prediction method for mine roadway deformation as described above.

[0015] A computer-readable storage medium storing a computer program that can be executed by a processor to implement the intelligent prediction method for mine roadway deformation as described above.

[0016] The intelligent prediction method, equipment, and medium for mine roadway deformation of the present invention have the following advantages: (i) This invention achieves the fusion of diverse and heterogeneous data such as microseismic data and surrounding rock deformation by re-encoding microseismic information, thereby improving information density and providing more useful information for surrounding rock deformation prediction, thus achieving the goal of predicting tunnel deformation faster and more accurately. (ii) This invention establishes a microseismic signal encoder by deploying and collecting data from multiple monitoring devices such as strain and microseismic data in mine roadways, and then uses a multi-dimensional heterogeneous data decoder to predict mine roadway deformation, etc. (III) This invention characterizes the microseismic information during the damage process of the surrounding rock of the roadway into a damage state representation vector through microseismic coding blocks. Combined with strain monitoring data, it adopts an autoencoder network with a long short-term memory network as the backbone to realize the prediction of mine roadway deformation considering multi-source heterogeneous information such as external strain and internal damage (microseismic information). This invention is mainly applied to smart mines, unified network management and other fields. (iv) By deploying multiple sensors, this invention provides more useful information and increases information density, thereby enabling more accurate prediction of roadway deformation through microseismic coding blocks based on long short-term memory networks. (v) The present invention constructs a microseismic coding block based on a long short-term memory network. By encoding additional information, the data representation obtained can be regarded as an equivalent representation of the damage variable of rock material at the current moment, thereby providing a more advantageous prediction of mine roadway deformation in terms of computational accuracy, speed and operability. Attached Figure Description

[0017] The invention will be further described below with reference to the accompanying drawings.

[0018] Appendix Figure 1 A flowchart of a method for intelligent prediction of mine roadway deformation; Appendix Figure 2 A schematic diagram of an intelligent prediction method for mine roadway deformation; Appendix Figure 3 This is a schematic diagram of the micro-vibration encoder architecture; Appendix Figure 4 This is a schematic diagram of the architecture of a multi-source heterogeneous data decoder. Detailed Implementation

[0019] The intelligent prediction method, equipment and medium for mine roadway deformation of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] Example 1:

[0021] As attached Figure 1 As shown in the figure, this embodiment provides an intelligent prediction method for mine roadway deformation, which is as follows: S1. Deploy microseismic sensors and deformation sensors around the mine roadways, and ensure that the microseismic sensors and deformation sensors have real-time data transmission capabilities; wherein, the number of microseismic sensors shall not be less than eight. S2. Based on the type and location of the microseismic sensor and deformation sensor, transmit the collected microseismic data and tunnel deformation data to the database; S3. Perform data processing and feature extraction on the microseismic data to obtain damage characterization of the surrounding rock of the mine roadway; wherein, damage characterization of the surrounding rock of the mine roadway refers to resolving the microseismic data into fracture size, magnitude, energy and fracture type. S4. The damage characterization of the surrounding rock of the mine roadway, which represents the microseismic signal, is transformed into a microseismic characterization through a microseismic coding block. The microseismic characterization and roadway deformation data are processed by a multi-source heterogeneous data decoder to predict the changes in the deformation of the mine roadway in the future. Among them, the microseismic characterization reflects the damage level of the rock in the current time window.

[0022] As attached Figure 2 As shown, in step S3 of this embodiment, the data processing and feature extraction of microseismic data to obtain the damage characterization of the surrounding rock of the mine roadway are as follows: S301. Microseismic Information Characterization: Calculate the radius of the microseismic source. Magnitude and energy The failure characteristic parameters are obtained, and the rupture type of the microseismic event is determined based on the moment tensor decomposition method. Then, the characteristic vector of the microseismic event is defined. ; S302, Microseismic Feature Normalization: Eigenvector The set is ,Will Normalization is performed to obtain the normalized microseismic feature vector sequence; S303, Microseismic Transformation: The normalized microseismic feature vector sequence is converted into a microseismic representation by a microseismic coding block based on a long short-term memory neural network.

[0023] The microseismic information characterization in step S301 of this embodiment is specifically as follows: S30101. Calculate the radius of the microseismic source respectively. Magnitude and energy The destructive characteristic parameters are given by the following formula: ; ; ; in, Brune constant, ; Indicates the shear wave velocity; This represents the corner frequency of the acoustic emission signal. For indoor acoustic emission tests, Replaced by the peak frequency of the signal. For the first The maximum amplitude of the acoustic emission signal waveform received by the micro-vibration sensor; Indicates acoustic emission events and the first The distance between the micro-vibration sensors is in mm; This indicates the number of microseismic sensors used for amplitude calculation. For the first The absolute energy of the waveform received by each micro-vibration sensor is obtained by integrating the waveform envelope; S30102. Based on the moment tensor decomposition method, the rupture type of microseismic events is determined using the following formula: ; in, , For tensor The three eigenvalues ​​represent the maximum couple; when When the earthquake source is in a shear-dominant state, it is determined to be a shear failure; when At that time, the earthquake source was in a mixed state between shear and tension; At that time, the epicenter was determined to be a tensile failure; S30103, Assuming At what moment does the microseismic probe receive the microseismic event? Through microseismic events Inverse magnitude calculation ,radius ,energy Information on the type of rupture (e.g., shear-type or tension-type acoustic emission events). microseismic events The corresponding eigenvector Characterized as .

[0024] The microseismic feature normalization in step S302 of this embodiment is as follows: S30201, will The data is distributed on a standard distribution with a mean of 0 and a variance of 1, as shown in the following formula: ; in, Indicates the translation coefficient; Indicates the scaling factor; S30202, Introducing a re-translation coefficient and rescaling factor Two new learnable variables, further study of the variables The transformation is performed, and the formula is as follows: ; S30203. After normalization, the microseismic coding process is represented as follows: ; in, This represents the input variable, and then the coefficients are shifted. and rescaling factor Learning is performed using gradient descent.

[0025] The microseismic coding in step S303 of this embodiment is as follows: S30301. The normalized microseismic feature vector sequence is used as input, and the microseismic feature vector sequence is converted into a local microseismic hidden state sequence using a long short-term memory neural network. S30302. By fusing global microseismic latent states, local microseismic latent states are converted into global microseismic latent states by adding and averaging. S30303: Through microseismic characterization transformation, the latent state is transformed into a microseismic characterization.

[0026] As attached Figure 3 As shown, the specific implementation method of the microseismic coding block based on the long short-term memory neural network in this embodiment is as follows: (1) There is The microseismic feature tensor, which has been normalized by the feature layer, is processed to the feature vector. At the same time, the characteristic vector of the previous microseismic event is processed again. The resulting hidden variables Then, within the current time step, the hidden variables The calculation formula is as follows: ; (2) To enable the network to process microseismic feature vector sequences of arbitrary length, two artificially defined vectors representing the start and end of the sequence are introduced, denoted as [vectors to be inserted here]. and ;in, As the initial hidden state of a long short-term memory neural network layer, and Common input model Long Short-Term Memory Neural Network Layer, Vector It is a zero vector with the same shape as the other hidden states; This indicates that there are no other microseismic feature vectors within the current processing time window, and the microseismic coding block receives... When used as output, Together with the last hidden state before Input the data into a Long Short-Term Memory (LSTM) neural network layer and calculate the last hidden state. , The vector is a microseismic feature vector with a shape similar to the input. Given zero vectors with the same shape, the hidden states in the microseismic coding block are calculated using the following formula: ; (3) Through continuous iteration, the long short-term memory neural network layer acquires a series of hidden states. ,variable Each microseismic event is considered as a local time window representation of the event. To obtain a microseismic event representation that can characterize the intensity of the event at the global level, global microseismic feature fusion is performed. The microseismic feature fusion first averages the hidden states of each independent local microseismic feature to obtain the global microseismic feature hidden state. And then Input to a multilayer perceptron to obtain the global microseismic encoding vector. The formula is as follows: ; .

[0027] As attached Figure 4 As shown, in step S4 of this embodiment, the damage characterization of the surrounding rock of the mine roadway, representing the microseismic signal, is transformed into a microseismic characterization through a microseismic coding block. The microseismic characterization and roadway deformation data are then processed by a multi-source heterogeneous data decoder to predict the changes in mine roadway deformation over a future period, as detailed below: S401, Set in time window Inside, microseismic sensors detected microseismic events. Meanwhile, deformation sensors monitor continuous rock deformation data, and deformation sequences are obtained by sampling at equal intervals. ,in ; S402, Deformation sequence Written in vector form, denoted as ; where vector The subscript indicates the sequence number of the time window. This indicates that the corresponding time window is the time window currently being monitored; S403, in the event of a microseismic event Microseismic characterization is converted from microseismic coding blocks. Characterization using microseismic events and deformation vector Predict the deformation vector for several subsequent time windows. .

[0028] In step S403 of this embodiment, once a microseismic event occurs... Microseismic characterization is converted from microseismic coding blocks. Characterization using microseismic events and deformation vector Predict the deformation vector for several subsequent time windows. Specifically as follows: S40301, The detected deformation vector Along with microseismic characterization As input, the modified sequence for the next time window is obtained from the Long Short-Term Memory neural network layer. and hidden state ; S40302, Long Short-Term Memory Neural Network iteratively generates the predicted time window. Deformation vector in and hidden state and along with microseismic characterization Input the long short-term memory neural network to predict the first... Deformation vector of each time window and hidden state The process continues until the Long Short-Term Memory (LSTM) neural network determines that it cannot predict the information in the next time window based on the current information and outputs an end marker. .

[0029] Example 2: This embodiment also provides an electronic device, including: a memory and at least one processor; The memory stores computer-executed instructions; The at least one processor executes computer execution instructions stored in the memory, causing the at least one processor to execute the intelligent prediction method for mine roadway deformation in any embodiment of the present invention.

[0030] The processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor can be a microprocessor or any conventional processor.

[0031] Memory is used to store computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory, and by accessing data stored in the memory. Memory can mainly include a program storage area and a data storage area. The program storage area can store the operating system, at least one application program required for a function, etc.; the data storage area can store data created based on the use of the terminal, etc. In addition, memory can also include high-speed random access memory, and can also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart memory cards (SMC), secure digital cards (SD cards), flash memory cards, at least one disk storage device, flash memory devices, or other volatile solid-state storage devices.

[0032] Example 3:

[0033] This embodiment also provides a computer-readable storage medium storing multiple instructions, which are loaded by a processor to cause the processor to execute the intelligent prediction method for mine roadway deformation in any embodiment of the present invention. Specifically, a system or apparatus equipped with a storage medium may be provided, on which software program code implementing the functions of any of the above embodiments is stored, and the computer (or CPU or MPU) of the system or apparatus can read and execute the program code stored in the storage medium.

[0034] In this case, the program code read from the storage medium can itself implement the function of any of the above embodiments, and therefore the program code and the storage medium storing the program code constitute part of the present invention.

[0035] Storage media embodiments for providing program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RYM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer via a communication network.

[0036] Furthermore, it should be clear that not only can the program code read by the computer be executed, but also the operating system or other components operating on the computer can be instructed based on the program code to perform some or all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0037] Furthermore, it is understood that the program code read from the storage medium is written to the memory set in the expansion board inserted into the computer or to the memory set in the expansion unit connected to the computer. Then, based on the instructions of the program code, the CPU or other components installed on the expansion board or expansion unit execute some and all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0038] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent prediction of deformation in mine roadways, characterized in that, The method is as follows: Microseismic sensors and deformation sensors are deployed around the mine roadways, and both microseismic sensors and deformation sensors have real-time data transmission capabilities. Based on the type and location of the microseismic sensors and deformation sensors, the collected microseismic data and tunnel deformation data are transmitted to the database; Data processing and feature extraction are performed on microseismic data to obtain damage characterization of the surrounding rock of mine roadways; the damage characterization of the surrounding rock of mine roadways refers to resolving microseismic data into fracture size, magnitude, energy and fracture type. The damage characterization of the surrounding rock of the mine roadway, which represents the microseismic signal, is transformed into a microseismic characterization by a microseismic coding block. The microseismic characterization and roadway deformation data are then processed by a multi-source heterogeneous data decoder to predict the changes in mine roadway deformation over a future period. The microseismic characterization reflects the damage level of the rock within the current time window. Specifically, the microseismic data was processed and features were extracted to obtain the damage characterization of the surrounding rock in the mine roadway, as follows: Microseismic Information Characterization: Calculate the radius of the microseismic source separately Magnitude and energy The failure characteristic parameters are obtained, and the rupture type of the microseismic event is determined based on the moment tensor decomposition method. Then, the characteristic vector of the microseismic event is defined. ; Microseismic feature normalization: eigenvectors The set is ,Will Normalization is performed to obtain the normalized microseismic feature vector sequence; Microseismic transformation: The normalized microseismic feature vector sequence is converted into a microseismic representation by a microseismic coding block based on a long short-term memory neural network; The microseismic characteristics are normalized as follows: Will The data is distributed on a standard distribution with a mean of 0 and a variance of 1, as shown in the following formula: ; in, Indicates the translation coefficient; Indicates the scaling factor; Introducing the retranslation coefficient and rescaling factor Two new learnable variables, further study of the variables The transformation is performed, and the formula is as follows: ; After normalization, the microseismic coding process is represented as follows: ; in, This represents the input variable, and then the coefficients are shifted. and rescaling factor Learning is performed using gradient descent.

2. The intelligent prediction method for mine roadway deformation according to claim 1, characterized in that, The microseismic information characterization is specifically as follows: Calculate the source radius of the microseismic events respectively. Magnitude and energy The destructive characteristic parameters are given by the following formula: ; ; ; in, Brune constant, ; Indicates the shear wave velocity; This represents the corner frequency of the acoustic emission signal. For indoor acoustic emission tests, Replaced by the peak frequency of the signal. For the first The maximum amplitude of the acoustic emission signal waveform received by the micro-vibration sensor; Indicates acoustic emission events and the first The distance between the micro-vibration sensors is in mm; This indicates the number of microseismic sensors used for amplitude calculation. For the first The absolute energy of the waveform received by each micro-vibration sensor is obtained by integrating the waveform envelope; Based on the moment tensor decomposition method, the rupture type of microseismic events is determined using the following formula: ; in, , For tensor The three eigenvalues ​​represent the maximum couple; when When the earthquake source is in a shear-dominant state, it is determined to be a shear failure; when At that time, the earthquake source was in a mixed state between shear and tension; At that time, the epicenter was determined to be a tensile failure; Suppose At what moment does the microseismic probe receive the microseismic event? Through microseismic events Inverse magnitude calculation ,radius ,energy Information on the type of rupture microseismic events The corresponding eigenvector Characterized as .

3. The intelligent prediction method for mine roadway deformation according to claim 1, characterized in that, The microseismic transformation is as follows: The normalized microseismic feature vector sequence is used as input, and the microseismic feature vector sequence is converted into a local microseismic hidden state sequence using a long short-term memory neural network. By fusing global microseismic latent states, local microseismic latent states are converted into global microseismic latent states through summation and averaging. The latent state is transformed into a microseismic characterization through microseismic characterization transformation.

4. The intelligent prediction method for mine roadway deformation according to claim 1, characterized in that, The specific implementation method of the microseismic coding block based on the long short-term memory neural network is as follows: It has For the microseismic feature tensor that has already been normalized by the feature layer, when processing the feature vector... At the same time, the characteristic vector of the previous microseismic event is processed again. The resulting hidden variables Then, within the current time step, the hidden variables The calculation formula is as follows: ; Introduce two artificially defined vectors representing the start and end of the sequence, denoted as [vectors to be inserted here]. and ;in, As the initial hidden state of a long short-term memory neural network layer, and Common input model Long Short-Term Memory neural network layer, vector It is a zero vector with the same shape as the other hidden states; This indicates that there are no other microseismic feature vectors within the current processing time window, and the microseismic coding block receives... When used as output, Including the last hidden state before Input the data into a Long Short-Term Memory (LSTM) neural network layer and calculate the last hidden state. , The vector is a microseismic feature vector with a shape similar to the input. Given zero vectors with the same shape, the hidden states in the microseismic coding block are calculated using the following formula: ; Through continuous iteration, the Long Short-Term Memory (LSTM) neural network layer acquires a series of hidden states. ,variable Each microseismic event is considered as a local time window representation of the event. To obtain a microseismic event representation that can characterize the intensity of the event at the global level, global microseismic feature fusion is performed. The microseismic feature fusion first averages the hidden states of each independent local microseismic feature to obtain the global microseismic feature hidden state. Then Input to a multilayer perceptron to obtain the global microseismic encoding vector. The formula is as follows: ; 。 5. The intelligent prediction method for mine roadway deformation according to claim 1, characterized in that, The damage characterization of the surrounding rock of the mine roadway, representing microseismic signals, is transformed into a microseismic characterization using a microseismic coding block. A multi-source heterogeneous data decoder then processes the microseismic characterization and roadway deformation data to predict future changes in mine roadway deformation, as detailed below: Set in time window Inside, microseismic sensors detected microseismic events. Meanwhile, deformation sensors monitor continuous rock deformation data, and deformation sequences are obtained by sampling at equal intervals. ,in ; Deformation sequence Written in vector form, denoted as ; where vector The subscript indicates the sequence number of the time window. This indicates that the corresponding time window is the time window currently being monitored; Once a micro-seismic event occurs Microseismic characterization is converted from microseismic coding blocks. Characterization using microseismic events and deformation vector Predict the deformation vector for several subsequent time windows. .

6. The intelligent prediction method for mine roadway deformation according to claim 5, characterized in that, Once a micro-seismic event occurs Microseismic characterization is converted from microseismic coding blocks. Characterization using microseismic events and deformation vector Predict the deformation vector for several subsequent time windows. Specifically as follows: The detected deformation vector Along with microseismic characterization As input, the modified sequence for the next time window is obtained from the Long Short-Term Memory neural network layer. and hidden state ; Long Short-Term Memory (LSTM) neural networks iteratively generate prediction time windows. Deformation vector in and hidden state and along with microseismic characterization Input the long short-term memory neural network to predict the first... Deformation vector of each time window and hidden state The process continues until the Long Short-Term Memory (LSTM) neural network determines that it cannot predict the information in the next time window based on the current information and outputs an end marker. .

7. An electronic device, characterized in that, include: Memory and at least one processor; The memory contains computer programs; The at least one processor executes the computer program stored in the memory, causing the at least one processor to perform the intelligent prediction method for mine roadway deformation as described in any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that can be executed by a processor to implement the intelligent prediction method for mine roadway deformation as described in any one of claims 1 to 6.

Citation Information

Patent Citations

  • Intelligent prediction method and system for ground pressure disasters of wall rock affected by mining

    CN111472840A