Dynamic stress field induced rock burst intelligent identification and impact risk prediction method
By integrating multi-source sensor monitoring technology and artificial intelligence technology, a real-time monitoring and analysis system for dynamic stress fields is built, and the impact ground pressure risk identification and risk prediction is used to use DBN models, which solves the problem that traditional methods are difficult to predict and judge impact ground pressure under complex geological conditions, and achieves efficient and accurate impact ground pressure prediction and risk prediction.
Patent Information
- Application Number
- CN202411757047.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-03
- Publication Date
- 2025-03-28
AI Technical Summary
Traditional methods are difficult to effectively predict and identify shock ground-pressing disasters under complex geological conditions and mining environments, and there are great limitations and uncertainties.
By integrating multi-source sensor monitoring technology, signal processing and data analysis technology, as well as machine learning and deep learning, a real-time monitoring and accurate analysis system for dynamic stress fields is built, and a deep belief network (DBN) model is used to determine impact ground pressure risk and predict risk.
Real-time and comprehensive monitoring of dynamic stress fields is achieved, the accuracy and timeliness of impact ground pressure prediction is improved, potentially dangerous areas can be identified in advance, and the frequency and degree of loss of impact ground pressure disasters are reduced.
Smart Images
Figure CN119712226A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of coal mine safety, and in particular to a method for intelligently identifying rock burst induced by a dynamic stress field and predicting rock burst danger. Background Art
[0002] In the field of coal mining and underground engineering, rock burst is a highly destructive dynamic disaster. Its suddenness and unpredictability have always been a major problem affecting production safety and stability. The occurrence of rock burst is often closely related to the stress state of the coal rock mass, especially the changes in the dynamic stress field, such as stress concentration and energy release caused by mining disturbance, overburden fracture, geological tectonic movement, seismic waves, etc., which can easily induce rock burst disasters.
[0003] Traditionally, the identification and prediction of rock burst mainly rely on empirical judgment and simplified mechanical model analysis, which have great limitations and uncertainties in complex geological conditions and mining environments. With the rapid development of technologies such as the Internet of Things, big data, and artificial intelligence, new approaches have been provided for real-time monitoring of dynamic stress fields and intelligent identification and prediction of rock burst.
[0004] The intelligent identification of rock burst induced by dynamic stress field and the prediction method of rock burst danger aim to achieve real-time monitoring and accurate analysis of dynamic stress field in coal mining process by integrating multi-source sensor monitoring technology, signal processing and data analysis technology, and artificial intelligence technologies such as machine learning and deep learning. By deeply mining the potential information and features in stress data and building an efficient and accurate rock burst prediction model, it can warn potential dangerous areas in advance, provide scientific decision-making support for coal mining enterprises, and effectively reduce the frequency and degree of loss of rock burst disasters.
[0005] This technical background emphasizes the importance and necessity of intelligent identification of rock burst induced by dynamic stress field and prediction of rock burst danger under the background of limitations of traditional methods. It can not only improve the accuracy and timeliness of rock burst prediction, but also provide more comprehensive and systematic guarantee for coal mine safety production, and has broad application prospects and social and economic value. Summary of the invention
[0006] In view of this, the purpose of the present invention is to provide a method for intelligently identifying rock burst induced by a dynamic stress field and predicting its impact hazard, including: determining the dynamic stress field evolution area; intelligently identifying whether there is a change in the dynamic stress field caused by overburden rupture or mining in a given area to be predicted; assuming that there is a change in the dynamic stress field in the given area to be predicted, judging by intelligent calculation whether there is a risk of rock burst induced by dynamic stress evolution in the area; proposing an impact hazard prediction index for rock burst induced by dynamic stress, and at the same time classifying the impact hazard level induced by dynamic stress evolution; calculating the range of the prediction index for the impact hazard induced by the dynamic stress field evolution in the area to be predicted, classifying the area into impact hazard levels according to the proposed impact hazard level classification method, and predicting the impact hazard of the area. This is to solve the technical problems in the intelligent identification of rock burst induced by dynamic stress field and the prediction of hazard levels. The present invention specifically includes:
[0007] A method for intelligently identifying rock burst induced by a dynamic stress field and predicting rock burst hazard, comprising: determining a dynamic stress field evolution region; intelligently identifying whether there is a change in the dynamic stress field caused by overburden fracture or mining in a given area to be predicted; assuming that there is a change in the dynamic stress field in the given area to be predicted, identifying by intelligent calculation whether there is a risk of rock burst induced by dynamic stress evolution in the area; proposing a rock burst hazard prediction index for dynamic stress induced rock burst, and classifying the rock burst hazard induced by dynamic stress evolution; calculating a range of values of the rock burst hazard prediction index for the area to be predicted induced by the dynamic stress field evolution, classifying the area into rock burst hazard levels according to the proposed rock burst hazard level classification method, and predicting the rock burst hazard of the area.
[0008] Furthermore, the determination of the dynamic stress field evolution area specifically includes: collecting data on site, mainly collecting data on pressure step distance, mining stress change characteristics, surrounding rock deformation, etc., and indoor coal rock strength testing, sampling the area, performing uniaxial and triaxial tests indoors, and obtaining the uniaxial compressive strength σ of the coal rock mass. b , tensile strength σ t , elastic modulus E, Poisson's ratio and other mechanical parameters; numerical simulation calculation, numerical modeling of the area through discrete element, simulated and calculated the changes in the plastic zone and the distribution characteristics of the maximum principal stress in the area, and at the same time analyzed the dynamic stress evolution law under the conditions of dynamic load generated by overburden fracture, and then analyzed whether the dynamic stress changes in the area are likely to induce impact rock pressure.
[0009] Furthermore, the intelligent judgment of whether there is overburden fracture or mining causing dynamic stress field changes in a given predicted area includes a dynamic stress wave monitoring sensor, an intelligent recognition program, and a wireless transmission and receiving device. The wireless transmission and receiving device is mainly used to transmit or receive dynamic stress wave sensing signals.
[0010] Furthermore, the stress waves generated by overburden fracture or mining are monitored and converted into electrical signals.
[0011] Further, including:
[0012] Ⅰ) Collect stress wave data monitored by dynamic stress sensors in real time, integrate the collected data into a dynamic stress sample data set according to the time series, normalize, standardize and denoise the dynamic stress data set, and divide it into a training set, a test set and a validation set according to a certain time interval;
[0013] Ⅱ) The training set, test set and validation set of the dynamic stress dataset are used to perform unsupervised pre-training, layer-by-layer construction and validation of the deep belief network (DBN) prediction model to obtain a deep belief network (DBN) model for predicting dynamic stress.
[0014] Further, including:
[0015] III) Determine the number of layers of the DBN model. Each layer is a restricted Boltzmann machine (RBM). Set the appropriate number of neurons in the visible and hidden layers for each layer of RBM.
[0016] Ⅶ) Input the sample data set into the DBN model through position encoding for pre-training, and obtain the initial data of the DBN model at the same time. The DBN model is trained layer by layer in a bottom-up order. Each layer is an RBM. When the first layer training is completed, the next layer of RBM is trained through maximum likelihood estimation. The accuracy of the DBN model is improved through repeated iterations;
[0017] VIII) Improve the accuracy and generalization ability of the DBN prediction model through training with the back-propagation algorithm, perform fine-tuning tests on the trained DBN model, select the appropriate number of iterations to improve the stability and convergence speed of the prediction model, and obtain the decoded data input of the DBN prediction model.
[0018] Furthermore, if there is a dynamic stress field change in the given area to be predicted, determining whether there is a risk of rock burst induced by dynamic stress evolution in the area through intelligent calculation includes the following steps:
[0019] 1) The dynamic stress output value t is calculated by the DBN prediction model, and the dynamic stress change and rock burst pressure are Weibull distributed by analysis;
[0020] 2) The distribution function of t and rock burst probability f is as follows:
[0021]
[0022] When t≥0, there is a risk of rock burst in the area, and when t<0, there is no risk of rock burst in the area;
[0023] 3) The historical data of rock burst accidents induced by overburden fracture were selected, the t value was calculated by DBN prediction model, the confidence level of rock burst probability was set to 95%, and the calculated value of λ was 8.55.
[0024] Furthermore, the proposed dynamic stress-induced rock burst hazard prediction index and the classification of dynamic stress evolution-induced rock burst hazard levels specifically include:
[0025] 1) Combined with the dynamic stress field generated by mining or overburden fracture that induces rock burst in the area, the rock burst hazard prediction index D formed by dynamic stress t is proposed. The calculation of D is as follows:
[0026]
[0027] Where D is 0-1. The larger the D value is, the greater the risk of surface impact. k is the rock expansion coefficient. The rock expansion coefficient refers to the degree of cracking and fragmentation of the rock under the action of external force. It is one of the important indicators to measure the compressive strength and impact resistance of the rock. It is usually expressed by the formula: k = (V1-V0) / V0×100%, V1 is the volume after force; V0 is the volume before force.
[0028] 2) The D values are graded and classified as follows:
[0029] When the value of D is 0-0.25, the risk level of rock burst induced by the dynamic stress field is low intensity; when the value of D is 0.25-0.5, the risk level of rock burst induced by the dynamic stress field is medium intensity; when the value of D is 0.5-0.75, the risk level of rock burst induced by the dynamic stress field is high intensity; when the value of D is 0.75-1, the risk level of rock burst induced by the dynamic stress field is super intensity.
[0030] Furthermore, the calculation of the range of impact hazard prediction indicators induced by the dynamic stress field evolution of the area to be predicted is carried out, the area is classified into impact hazard levels according to the proposed impact hazard level classification method, and the impact hazard of the area is predicted, including: first calculating the dynamic stress value t of the area to be predicted, then calculating the impact rock pressure hazard prediction indicator D, and taking the impact hazard level corresponding to the maximum value of the D value calculation result within a certain time range as the impact hazard level of the area.
[0031] Beneficial effects of the present invention:
[0032] 1. The present invention provides an intelligent identification of rock burst induced by a dynamic stress field, which uses a deep belief network (DBN) model to identify and process the data of the dynamic stress field, and obtains whether there is a rock burst risk in the detection area. Through advanced data analysis algorithms, it is possible to achieve comprehensive and real-time monitoring of the dynamic stress field and capture the subtle features of stress changes. Combined with the deep belief network (DBN) model, it is possible to deeply explore the potential laws in the stress data and improve the accuracy and reliability of rock burst prediction.
[0033] 2. The present invention provides a method for predicting the impact hazard of rock burst induced by a dynamic stress field. The method can identify in advance the key areas in the dynamic stress field that may induce rock burst, such as stress concentration areas, abnormal energy release rates, etc., so as to achieve timely level prediction of potential dangerous areas and improve the basic guarantee for rock burst induced by dynamic stress fields. This helps coal mining enterprises to take preventive measures in advance and reduce the probability of occurrence and degree of loss of rock burst disasters. Decision support based on intelligent identification and prediction results can enable coal mining enterprises to be more scientific and reasonable in formulating production plans, optimizing mining layouts, and strengthening safety supervision. This not only helps to improve the safety and stability of coal mine production, but also improves production efficiency and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] Figure 1 A flow chart of a method for intelligently identifying rock burst induced by a dynamic stress field and predicting rock burst risk provided in an embodiment of the present invention.
[0035] Figure 2 This is a diagram of the DBN model prediction method architecture provided by an embodiment of the invention.
[0036] Figure 3 A dynamic stress evolution-induced rock burst risk level prediction process and risk measure response diagram are provided for an embodiment of the invention. DETAILED DESCRIPTION
[0037] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.
[0038] The method for intelligently identifying rock burst induced by dynamic stress field and predicting rock burst risk in this embodiment includes:
[0039] 1. Through on-site data collection, mainly collect data such as pressure step distance, mining stress change characteristics, surrounding rock deformation, etc. Indoor coal rock strength test, by sampling the area, performing uniaxial and triaxial tests indoors, obtain the uniaxial compressive strength σ of the coal rock mass b , tensile strength σ t, elastic modulus E, Poisson's ratio and other mechanical parameters. Numerical simulation calculation, numerical modeling of the region through discrete elements, simulation calculation of the changes in the plastic zone and the distribution characteristics of the maximum principal stress in the region, and analysis of the dynamic stress evolution law under the dynamic load conditions generated by the overburden fracture. Then analyze whether the dynamic stress changes in the region are likely to induce rock burst.
[0040] 2. It includes intelligent identification of whether there is a change in dynamic stress field caused by overburden breaking or mining in a given predicted area, including a dynamic stress wave monitoring sensor for monitoring stress waves generated by overburden breaking or mining, and converting stress waves into electrical signals. , intelligent identification program, wireless transmission and receiving device, the wireless transmission and receiving device is mainly used to transmit or receive dynamic stress wave sensor signals.
[0041] 3. Intelligent recognition program includes:
[0042] 3.1 Collect stress wave data monitored by dynamic stress sensors in real time, integrate the collected data into a dynamic stress sample data set according to the time series, normalize, standardize and denoise the dynamic stress data set, and divide it into training set, test set and validation set according to a certain time interval.
[0043] 3.2 The training set, test set and validation set of the dynamic stress dataset are used to perform unsupervised pre-training, layer-by-layer construction and validation of the deep belief network (DBN) prediction model to obtain a deep belief network (DBN) model for predicting dynamic stress.
[0044] 4. DBN model training, testing and validation sets include:
[0045] 4.1 Determine the number of layers of the DBN model. Each layer is a restricted Boltzmann machine (RBM). Set the appropriate number of neurons in the visible layer and hidden layer for each layer of RBM.
[0046] 4.2 Input the sample data set into the DBN model through position encoding for pre-training, and obtain the initial data of the DBN model. The DBN model is trained layer by layer in a bottom-up order, and each layer is an RBM. When the first layer training is completed, the next layer of RBM is trained through maximum likelihood estimation, and the accuracy of the DBN model is improved through repeated iterations.
[0047] 4.3 Improve the accuracy and generalization ability of the DBN prediction model through training with the back-propagation algorithm, fine-tune the trained DBN model, select the appropriate number of iterations to improve the stability and convergence speed of the prediction model, and obtain the decoded data input of the DBN prediction model.
[0048] 5. Given that there are dynamic stress field changes in the area to be predicted, whether there is a risk of rock burst induced by dynamic stress evolution in the area is determined through intelligent calculation, including the following steps:
[0049] 5.1 The dynamic stress output value is calculated as t through the DBN prediction model, and the dynamic stress change and rock burst pressure are analyzed to present a Weibull distribution.
[0050] The distribution function of 5.2t and rock burst probability f is as follows:
[0051]
[0052] When t≥0, there is a risk of rock burst in the area, and when t<0, there is no risk of rock burst in the area.
[0053] 5.3 The historical data of rock burst accidents induced by overburden fracture were selected, and the t value was calculated by DBN prediction model. The confidence level of rock burst probability was set to 95%, and the calculated value of λ was 8.55.
[0054] 6. Propose the prediction index of rock burst induced by dynamic stress, and classify the rock burst hazard level induced by dynamic stress evolution, including:
[0055] 6.1 Combined with the dynamic stress field generated by mining or overburden fracture that induces rock burst in the area, the rock burst hazard prediction index D formed by dynamic stress t is proposed. The calculation of D is as follows:
[0056]
[0057] Among them, D ranges from 0 to 1. The larger the D value, the greater the risk of surface impact. k is the rock expansion coefficient. The rock expansion coefficient refers to the degree of cracking and fragmentation of the rock under the action of external force. It is one of the important indicators to measure the compressive strength and impact resistance of rock. It is usually expressed by the formula: k = (V1-V0) / V0×100%, V1 is the volume after force; V0 is the volume before force.
[0058] 6.2 The D value is graded and classified as follows:
[0059] When the value of D is 0-0.25, the risk level of rock burst induced by the dynamic stress field is low intensity; when the value of D is 0.25-0.5, the risk level of rock burst induced by the dynamic stress field is medium intensity; when the value of D is 0.5-0.75, the risk level of rock burst induced by the dynamic stress field is high intensity; when the value of D is 0.75-1, the risk level of rock burst induced by the dynamic stress field is super intensity.
[0060] 7. It also includes calculating the range of values of the impact hazard prediction index induced by the dynamic stress field evolution of the area to be predicted, classifying the impact hazard level of the area according to the proposed impact hazard level classification method, and predicting the impact hazard of the area, including: calculating the dynamic stress value t of the area to be predicted according to claims 3 to 6, calculating the impact rock pressure hazard prediction index D according to claim 8, and taking the impact hazard level corresponding to the maximum value of the D value calculation result within a certain time range as the impact hazard level of the area.
[0061] Finally, it should be noted that the above description is only an embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A method for intelligently identifying rock burst induced by a dynamic stress field and predicting rock burst risk, comprising: Determine the dynamic stress field evolution area; Intelligently identify whether there is overburden fracture or dynamic stress field change caused by mining in a given predicted area; Assuming that there are dynamic stress field changes in the given area to be predicted, it is determined through intelligent calculation whether there is a risk of rock burst induced by dynamic stress evolution in the area; an impact hazard prediction index for dynamic stress induced rock burst is proposed, and the dynamic stress evolution induced rock burst hazard level is divided; the range value of the impact hazard prediction index induced by dynamic stress field evolution in the area to be predicted is calculated, and the area is divided into impact hazard levels according to the proposed impact hazard level division method, and the impact hazard of the area is predicted.
2. The method for intelligently identifying rock burst induced by dynamic stress field and predicting rock burst risk according to claim 1 is characterized in that: The determination of the dynamic stress field evolution area specifically includes: collecting data on site, mainly collecting data on pressure step distance, mining stress change characteristics, surrounding rock deformation, etc., and indoor coal rock strength testing, sampling the area, performing uniaxial and triaxial tests indoors, and obtaining the uniaxial compressive strength σ of the coal rock mass. b , tensile strength σ t , elastic modulus E, Poisson's ratio and other mechanical parameters; numerical simulation calculation, numerical modeling of the area through discrete element, simulated and calculated the changes in the plastic zone and the distribution characteristics of the maximum principal stress in the area, and at the same time analyzed the dynamic stress evolution law under the conditions of dynamic load generated by overburden fracture, and then analyzed whether the dynamic stress changes in the area are likely to induce impact rock pressure.
3. The method for intelligently identifying rock burst induced by dynamic stress field and predicting rock burst risk according to claim 1, characterized in that: The intelligent identification system for determining whether there is overburden fracture or mining-induced dynamic stress field changes in a given predicted area includes a dynamic stress wave monitoring sensor, an intelligent identification program, and a wireless transmission and receiving device. The wireless transmission and receiving device is mainly used to transmit or receive dynamic stress wave sensing signals.
4. The method for intelligently identifying rock burst induced by dynamic stress field and predicting rock burst risk according to claim 3 is characterized in that: Monitor the stress waves generated by overburden fracture or mining, and convert the stress waves into electrical signals.
5. The method for intelligently identifying rock burst induced by dynamic stress field and predicting rock burst risk according to claim 3, characterized in that: include: Ⅰ) Collect stress wave data monitored by dynamic stress sensors in real time, integrate the collected data into a dynamic stress sample data set according to the time series, normalize, standardize and denoise the dynamic stress data set, and divide it into a training set, a test set and a validation set according to a certain time interval; Ⅱ) The training set, test set and validation set of the dynamic stress dataset are used to perform unsupervised pre-training, layer-by-layer construction and validation of the deep belief network (DBN) prediction model to obtain a deep belief network (DBN) model for predicting dynamic stress.
6. The method for intelligently identifying rock burst induced by dynamic stress field and predicting rock burst risk according to claim 5, characterized in that: include: III) Determine the number of layers of the DBN model. Each layer is a restricted Boltzmann machine (RBM). Set the appropriate number of neurons in the visible and hidden layers for each layer of RBM. Ⅶ) Input the sample data set into the DBN model through position encoding for pre-training, and obtain the initial data of the DBN model at the same time. The DBN model is trained layer by layer in a bottom-up order. Each layer is an RBM. When the first layer training is completed, the next layer of RBM is trained through maximum likelihood estimation. The accuracy of the DBN model is improved through repeated iterations; VIII) Improve the accuracy and generalization ability of the DBN prediction model through training with the back-propagation algorithm, perform fine-tuning tests on the trained DBN model, select the appropriate number of iterations to improve the stability and convergence speed of the prediction model, and obtain the decoded data input of the DBN prediction model.
7. The method for intelligently identifying rock burst induced by dynamic stress field and predicting rock burst risk according to claim 1, characterized in that: The given area to be predicted has dynamic stress field changes, and judging whether there is a risk of rock burst induced by dynamic stress evolution in the area through intelligent calculation includes the following steps: 1) The dynamic stress output value t is calculated by the DBN prediction model, and the dynamic stress change and rock burst pressure are Weibull distributed by analysis; 2) The distribution function of t and rock burst probability f is as follows: When t≥0, there is a risk of rock burst in the area, and when t<0, there is no risk of rock burst in the area; 3) The historical data of rock burst accidents induced by overburden fracture were selected, the t value was calculated by DBN prediction model, the confidence level of rock burst probability was set to 95%, and the calculated value of λ was 8.
55.
8. The method for intelligently identifying rock burst induced by dynamic stress field and predicting rock burst risk according to claim 1, characterized in that: The proposed dynamic stress-induced rock burst hazard prediction index and the classification of dynamic stress evolution-induced rock burst hazard levels specifically include: 1) Combined with the dynamic stress field generated by mining or overburden fracture that induces rock burst in the area, the rock burst hazard prediction index D formed by dynamic stress t is proposed. The calculation of D is as follows: Where D is 0-1. The larger the D value is, the greater the risk of surface impact. k is the rock expansion coefficient. The rock expansion coefficient refers to the degree of cracking and fragmentation of the rock under the action of external force. It is one of the important indicators to measure the compressive strength and impact resistance of the rock. It is usually expressed by the formula: k = (V1-V0) / V0×100%, V1 is the volume after force; V0 is the volume before force. 2) The D values are graded and classified as follows: When the value of D is 0-0.25, the risk level of rock burst induced by the dynamic stress field is low intensity; when the value of D is 0.25-0.5, the risk level of rock burst induced by the dynamic stress field is medium intensity; when the value of D is 0.5-0.75, the risk level of rock burst induced by the dynamic stress field is high intensity; when the value of D is 0.75-1, the risk level of rock burst induced by the dynamic stress field is super intensity.
9. The method for intelligently identifying rock burst induced by dynamic stress field and predicting rock burst risk according to claim 8, characterized in that: The method of calculating the range of values of the impact hazard prediction index induced by the dynamic stress field evolution of the area to be predicted, classifying the area into impact hazard levels according to the proposed impact hazard level classification method, and predicting the impact hazard of the area comprises: first calculating the dynamic stress value t of the area to be predicted, then calculating the impact rock pressure hazard prediction index D, and taking the impact hazard level corresponding to the maximum value of the D value calculation result within a certain time range as the impact hazard level of the area.