Mine roadway surrounding rock deformation prediction method based on neural network

Abstract

The invention relates to the technical field of roadway surrounding rock prediction, in particular to a mine roadway surrounding rock deformation prediction method based on a neural network, and the method comprises the steps: selecting at least one target roadway section in a mined region in a roadway being mined; inputting the roadway width, the roadway height, the mining progress diagram, the material tensor and the position code of the target roadway section into a prediction network to predict the maximum stress, the maximum strain and the plastic region area of each target roadway section; and according to the maximum stress, the maximum strain and the plastic region area of each target roadway section, early warning is carried out on surrounding rock deformation. By means of the technical scheme, accurate prediction of mine roadway surrounding rock deformation in the mining process can be achieved, and therefore early warning can be given out in time.

Description

Technical Field

[0001] The present application relates to the field of roadway surrounding rock prediction technology, and particularly to a method for predicting the deformation of the surrounding rock of a mine roadway based on a neural network. Background Art

[0002] The drift filling mining method is a mining method often used in gold mining. After a section is mined, the mined - out area is backfilled with cemented tailings, paste or other filling materials. The filling body can bear part of the rock pressure, reduce the roof subsidence and rib displacement, so as to maintain the stability of the roadway surrounding rock. During the mining process, the mining progress will affect the stress condition of the roadway, and thus affect the stability of the roadway.

[0003] At present, the patent application document with the publication number of CN105260575A discloses a method for predicting the deformation of the roadway surrounding rock based on a neural network. The method includes: obtaining the key influencing factors of the surrounding rock through the analytic hierarchy process, establishing a BP neural network model, inputting the training samples formed by the data groups monitored under different geological conditions in the early stage. The data structure is a multi - dimensional matrix, and the training sample data is used to train the neural network system through the trainlm function; using the trained neural network to predict the deformation amount at the initial stage of roadway excavation. According to the input index parameters, the roof subsidence amount, floor uplift amount, rib strain amount and the maximum plastic zone failure depth of the surrounding rock are obtained. The neural network will predict the deformation of the roadway at the initial stage of excavation according to the prediction request, select appropriate support parameters to control the roadway support, and prevent safety accidents caused by the instability of the rock mass.

[0004] The above - mentioned method trains the BP neural network model based on the data groups monitored under different geological conditions, so as to predict the deformation of the roadway at the initial stage of excavation. However, the above - mentioned method ignores the influence of the mining progress on the stress condition of the roadway during the mining process, and cannot accurately predict the deformation of the surrounding rock of the mine roadway during the mining process. Summary of the Invention

[0005] In order to solve the technical problem that the accurate prediction of the deformation of the surrounding rock of the mine roadway during the mining process cannot be achieved, the present application provides a method for predicting the deformation of the surrounding rock of a mine roadway based on a neural network, which can accurately predict the deformation of the surrounding rock of the mine roadway during the mining process, and thus issue an early warning in a timely manner.

[0006] In the first aspect of the present application, a method for predicting the deformation of the surrounding rock of a mine roadway based on a neural network is provided. The prediction method includes: selecting at least one target roadway cross-section in the mined area of the roadway being mined; inputting the mining progress map, the material tensor, and the position encoding of the target roadway cross-section into a prediction network to predict the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section, where the mining progress map includes the mining status of each position point in all roadways, and the mining status includes filled, mined, and unmined; the position encoding includes the roadway ID of the roadway where it is located and the horizontal distance between the target roadway cross-section and the stope; the material tensor includes the mechanical parameters of the hanging wall surrounding rock, the footwall surrounding rock, the ore body, and the filling material; judging whether the roadway being mined is deformed based on the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section, and if it is deformed, an early warning signal is issued.

[0007] During the mining of a gold mine using the drift filling mining method, as the mining process progresses, the stress condition of the roadway being mined will also change accordingly. To timely judge whether the roadway being mined is deformed, at least one target roadway cross-section is selected in the mined area of the roadway being mined, and the roadway width, roadway height, mining progress map, material tensor, and the position encoding of the target roadway cross-section are input into the prediction network to predict the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section. The roadway width and roadway height can reflect the size information of the target roadway cross-section, the mining progress map and the material tensor can reflect the mechanical parameters of the materials around the target roadway cross-section, and the position encoding of the target roadway cross-section is used to determine the position information of the target roadway cross-section. By comprehensively considering the size information, position information, and mechanical parameters of the materials around the target roadway cross-section, the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section can be accurately predicted, and then it can be judged whether the roadway being mined is deformed. If it is deformed, an early warning signal is issued.

[0008] Preferably, the mechanical parameters include elastic modulus, Poisson's ratio, internal friction angle, density, tensile strength, and cohesion.

[0009] Preferably, selecting at least one target roadway cross-section includes: calculating the horizontal distance from each roadway cross-section to the stope in the mined area of the roadway being mined; selecting at least one target roadway cross-section based on the horizontal distance.

[0010] Preferably, selecting at least one target roadway cross-section based on the horizontal distance includes: the interval between any two roadway cross-sections is the absolute value of the difference in horizontal distance, and multiple target roadway cross-sections are selected from each roadway cross-section at a preset interval.

[0011] Preferably, selecting at least one target roadway cross-section according to the horizontal distance includes: selecting the roadway cross-section with a horizontal distance of 0 as the target roadway cross-section, and calculating a selection coefficient, where the selection coefficient is positively correlated with the horizontal distance of the target roadway cross-section; taking the product of the selection coefficient and the minimum interval as the selection interval, and taking the roadway cross-section corresponding to the sum of the horizontal distance of the target roadway cross-section and the selection interval as the next target roadway cross-section; successively selecting target roadway cross-sections until no next target roadway cross-section can be obtained and then stopping.

[0012] The stope is the active area of the mine exploitation and is the main disturbance source affecting the stress conditions of each roadway cross-section. Therefore, the closer to the stope, the faster the change of its stress conditions. Selecting dense cross-sections in the area close to the stope and sparse cross-sections in the area far from the stope can not only accurately capture the change trend of the stress conditions to accurately obtain the deformation prediction results, but also reduce the calculation amount.

[0013] Preferably, the prediction network includes a convolutional sub-network and a regression sub-network; the convolutional sub-network is used to extract features from the mining progress map to obtain progress features; after splicing the progress features, the material tensor and the position encoding of the target roadway cross-section, the splicing result is input into the regression sub-network to predict the maximum stress, maximum strain and plastic zone area of the target roadway cross-section.

[0014] Provide an end-to-end neural network structure that can predict the maximum stress, maximum strain and plastic zone area of any target roadway cross-section.

[0015] Preferably, the training method of the prediction network includes: numerically simulating the mine exploitation process according to the material tensor, obtaining the stress nephogram, strain nephogram and plastic zone distribution map of the roadway cross-section with any position encoding under any mining progress map, and then obtaining the stress label, strain label and plastic zone label; inputting the material tensor, the mining progress map and the position encoding into the prediction network to predict the maximum stress, maximum strain and plastic zone area, and performing weighted summation on the first deviation between the maximum stress and the stress label, the second deviation between the maximum strain and the strain label, and the third deviation between the plastic zone area and the plastic zone label to obtain a loss function, so as to perform iterative training on the prediction network; in response to the loss function being less than a preset loss or the number of iterations being greater than a preset number, complete the training of the prediction network.

[0016] Use FLAC3D software to numerically simulate the exploitation process to obtain training data for training the prediction network, and constrain the prediction network to learn the mapping relationship between the maximum stress, maximum strain, plastic zone area and input information.

[0017] Preferably, the loss function satisfies the relational expression: ; , and are the maximum stress, maximum strain, and plastic zone area predicted by the prediction network, respectively. , and are the stress label, strain label, and plastic zone label, respectively. , and are the first weight, second weight, and third weight, respectively, and satisfy .

[0018] Preferably, the method for obtaining the first weight includes: during the numerical simulation process, statistically calculate the maximum stress of any roadway section at different mining progress to obtain a stress curve, and use the mean variance of the stress curves of multiple roadway sections as the stress variance; use the ratio of the stress variance to the sum of variances as the first weight, where the sum of variances is the total of the stress variance, strain variance, and plastic zone area variance.

[0019] Due to the different degrees of influence of the maximum stress, maximum strain, and plastic zone area on the mining progress, the first weight is assigned to the maximum stress according to the change of the maximum stress with the mining progress during the numerical simulation. If the change of the maximum stress with the mining progress is greater, the first weight is also greater, ensuring the prediction accuracy of the maximum stress.

[0020] Preferably, determining whether the roadway being mined is deformed includes: taking the horizontal distance between the target roadway section and the stope as the abscissa and the maximum stress of the target roadway section as the ordinate to obtain a stress scatter curve, using the interpolation algorithm to interpolate the stress scatter curve to obtain the maximum stress of each roadway section in the mined area of the roadway being mined; obtaining the maximum strain and plastic zone area of each roadway section in the mined area of the roadway being mined in the same way; in response to the maximum stress of any roadway section being greater than the maximum allowable stress, or the maximum strain of any roadway section being greater than the maximum allowable strain, or the plastic zone area of any roadway section being greater than the maximum allowable plastic zone area, determining that the roadway being mined is deformed.

[0021] The technical solution of this application has the following beneficial technical effects: During the process of gold mine exploitation using the drift filling mining method, as the exploitation progresses, the stress conditions of the drift being exploited will also change accordingly. To timely determine whether the drift being exploited is deformed, at least one target drift cross-section is selected within the exploited area of the drift being exploited. The drift width, drift height, exploitation progress map, material tensor, and position encoding of the target drift cross-section are input into the prediction network to predict the maximum stress, maximum strain, and plastic zone area of each target drift cross-section. Among them, the drift width and drift height can reflect the size information of the target drift cross-section, the exploitation progress map and material tensor can reflect the mechanical parameters of the materials around the target drift cross-section, and the position encoding of the target drift cross-section is used to determine the position information of the target drift cross-section. By comprehensively considering the size information, position information, and mechanical parameters of the materials around the target drift cross-section, the maximum stress, maximum strain, and plastic zone area of each target drift cross-section are accurately predicted, and then it is determined whether the drift being exploited is deformed. If deformation occurs, a warning signal is issued.

[0022] Furthermore, when it is monitored that the exploitation progress increases by a preset length, at least one target drift cross-section is selected again to determine whether the drift being exploited is deformed after the change in exploitation progress, so as to realize real-time warning of the deformation of the surrounding rock of the mine drift during the exploitation process. Brief Description of the Drawings

[0023] Figure 1 is a flowchart of a method for predicting the deformation of the surrounding rock of a mine drift based on a neural network according to an embodiment of the present application.

[0024] Figure 2 is a schematic diagram of a mine model according to an embodiment of the present application. Detailed Embodiments

[0025] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0026] According to the first aspect of the present application, the present application provides a method for predicting the deformation of the surrounding rock of a mine drift based on a neural network. During the process of gold mine exploitation using the drift filling mining method, it is predicted whether the surrounding rock of the drift is deformed, and a warning message is issued in a timely manner.

[0027] For ease of understanding, the operation process of the drift filling mining method is introduced herein. The drift filling mining method divides the ore body into multiple layers according to different heights, and each layer corresponds to a drift roadway. According to the mining sequence of multiple drift roadways, the drift filling mining method can be divided into upward drift mining and downward drift mining. Among them, upward drift mining starts from the lowest drift roadway and mines upward in sequence, while downward drift mining starts from the uppermost drift roadway and mines downward in sequence.

[0028] In the drift filling mining method, the ore body is divided into 4 layers according to different heights, and each layer corresponds to a drift roadway. The drift roadways are sequentially denoted as roadway 1, roadway 2, roadway 3, and roadway 4 from top to bottom. If upward drift mining is used to operate on the ore body with a fixed roadway width and roadway height, the mining of roadway 4 is preferentially carried out. At this time, the surrounding of roadway 4 is all ore body materials. After the mining of roadway 4 is completed, the filling material is first used to backfill roadway 4, and then the mining process of roadway 3 is started. At this time, the material below roadway 3 is the filling material, and the materials in other positions are still ore body materials. The mining of each layer of drift roadway is completed in the same way in sequence.

[0029] Among them, the roadway width and roadway height can determine the size information of the roadway cross-section.

[0030] Figure 1 is a flowchart of the method for predicting the deformation of the surrounding rock of a mine roadway based on a neural network according to an embodiment of the present application. As Figure 1 shown, the method for predicting the deformation of the surrounding rock of a mine roadway based on a neural network includes steps S101 to S103, which are described in detail below.

[0031] S101, select at least one target roadway cross-section in the mined area of the roadway being mined.

[0032] In one embodiment, the target roadway cross-section is any cross-section in the mined area of the roadway being mined. The deformation of the roadway being mined is judged by predicting the maximum stress, maximum strain, and plastic zone area in the target roadway cross-section.

[0033] Specifically, selecting at least one target roadway cross-section includes: calculating the horizontal distance from each roadway cross-section to the stope in the mined area of the roadway being mined; selecting at least one target roadway cross-section based on the horizontal distance.

[0034] Understandably, the mined area in the roadway being mined contains multiple roadway cross-sections. If the maximum stress, maximum strain, and plastic zone area are predicted for all roadway cross-sections, although the accuracy of surrounding rock deformation prediction can be ensured, it undoubtedly requires a huge amount of computational effort. Therefore, it is necessary to select multiple target roadway cross-sections from the mined area in the roadway being mined. By predicting the maximum stress, maximum strain, and plastic zone area for multiple target roadway cross-sections, while accurately obtaining the deformation prediction results, the computational effort can be reduced.

[0035] In one embodiment, selecting at least one target roadway cross-section based on the horizontal distance includes: the interval between any two roadway cross-sections is the absolute value of the difference in horizontal distance, and multiple target roadway cross-sections are selected from each roadway cross-section at a preset interval.

[0036] That is to say, the absolute value of the difference in horizontal distance between adjacent target roadway cross-sections is equal to the preset interval. In the embodiment of the present application, the preset interval is 5 meters.

[0037] In another embodiment, the stope is the active area of mine mining and is the main disturbance source affecting the stress conditions of each roadway cross-section. Therefore, the closer to the stope, the faster the change in its stress conditions. To accurately capture the change trend of the stress conditions, it is necessary to arrange dense target roadway cross-sections. Relatively, as the horizontal distance between the cross-section and the stope increases, the disturbance intensity weakens, and the change trend of the stress conditions tends to be stable. Sparse target roadway cross-sections are sufficient to reflect the change trend of the stress conditions. At this time, sparse target roadway cross-sections can be arranged to avoid redundant sampling. Therefore, dense cross-sections are selected in the area close to the stope, and sparse cross-sections are selected in the area far from the stope, which can not only accurately capture the change trend of the stress conditions to accurately obtain the deformation prediction results, but also reduce the computational effort; wherein, the stress conditions include the maximum stress, maximum strain, and plastic zone area.

[0038] Specifically, selecting at least one target roadway cross-section based on the horizontal distance includes: selecting the roadway cross-section with a horizontal distance of 0 as the target roadway cross-section, and calculating the selection coefficient, where the selection coefficient is positively correlated with the horizontal distance of the target roadway cross-section; taking the product of the selection coefficient and the minimum interval as the selection interval, and taking the roadway cross-section corresponding to the sum of the horizontal distance of the target roadway cross-section and the selection interval as the next target roadway cross-section; sequentially selecting target roadway cross-sections until no next target roadway cross-section can be obtained and then stop.

[0039] Among them, the horizontal distance is of the selection coefficient of the target roadway cross-section is: , is the adjustment coefficient, which is used to control the speed at which the selection interval increases with the horizontal distance. In the embodiment of the present application, the value of the adjustment coefficient is 0.1; the value of the minimum interval is 2 meters.

[0040] Exemplarily, first select the roadway cross-section with a horizontal distance of 0 as the target roadway cross-section, then the value of the selection coefficient is 1, and the selection interval is the product of the minimum interval and the selection coefficient, that is, the value of the selection interval is 2 meters, and the next target roadway cross-section is the roadway cross-section with a horizontal distance of 2 meters; further calculate that the selection coefficient of the roadway cross-section with a horizontal distance of 2 meters is 1.2, and the value of the selection interval is 2.4 meters, then the next target roadway cross-section is the roadway cross-section with a horizontal distance of 4.4 meters; successively select multiple target roadway cross-sections.

[0041] In this way, by selecting multiple target roadway cross-sections in the mined area of the roadway being mined and predicting the maximum stress, maximum strain, and plastic zone area of all target roadway cross-sections, the distribution of the maximum stress, maximum strain, and plastic zone area in the mined area of the roadway being mined can be accurately obtained.

[0042] S102, input the roadway width, roadway height, mining progress map, material tensor, and position encoding of the target roadway cross-section into the prediction network to predict the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section.

[0043] In one embodiment, the roadway width, roadway height, mining progress map, material tensor, and position encoding of the target roadway cross-section are input into the prediction network to predict the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section; the prediction network is used to predict the maximum stress, maximum strain, and plastic zone area of the target roadway cross-section. Among them, the roadway width and roadway height can determine the dimension information of the roadway cross-section.

[0044] It should be noted that since there are differences in the dimensions among the roadway width, roadway height, and material tensor in the input data, the method of standardization processing should be adopted to eliminate the dimensions and avoid the situation of large numerical differences caused by different dimensions.

[0045] Among them, the mining progress map is a single-channel two-dimensional image, including the mining status of each position point in all roadways, and the mining status includes filled, mined, and unmined. Specifically, the method for obtaining the mining progress map is: project all roadways in the mine vertically onto a horizontal plane to form a top view, and map the mining status of each position point in all roadways to this top view to obtain the mining progress map. While the mining progress map can well reflect the roadway distribution and relative positions, it can intuitively reflect the mining status of each position point in all roadways.

[0046] Exemplarily, there are 4 stratified roadways in the mine exploitation plan, and each stratified roadway corresponds to a roadway ID, which are sequentially recorded as roadway 1, roadway 2, roadway 3, and roadway 4. If the exploitation status of each position point in roadway 1 is filled, it means that the exploitation and filling of roadway 1 have been completed, and the mechanical parameters of each position point in roadway 1 are the mechanical parameters of the filling material. If the exploitation status of each position point in roadway 3 and roadway 4 is unexploited, it means that roadway 3 and roadway 4 have not started exploitation yet, and the mechanical parameters of each position point in roadway 3 and roadway 4 are the mechanical parameters of the ore body material. If the exploitation status of a part of the position points in roadway 2 is exploited and the exploitation status of another part of the position points is unexploited, it means that roadway 2 is the roadway under exploitation, and all the position points with the exploitation status of exploited form the exploited area in the roadway under exploitation, and all the position points with the exploitation status of unexploited form the unexploited area in the roadway under exploitation. There is no material in the exploited area of roadway 2, and the mechanical parameters in the unexploited area of roadway 2 are the mechanical parameters of the ore body material.

[0047] Among them, the material tensor includes the mechanical parameters of the hanging wall surrounding rock, the footwall surrounding rock, the ore body, and the filling material, and the mechanical parameters include elastic modulus, Poisson's ratio, internal friction angle, density, tensile strength, and cohesion.

[0048] It can be understood that the comprehensive exploitation progress map and the material tensor can accurately reflect the material distribution of any cross-section in the exploited area of the roadway under exploitation.

[0049] Among them, the position coding of the target roadway cross-section includes the roadway ID of the roadway where it is located and the horizontal distance between the target roadway cross-section and the stope. The target roadway cross-section can be uniquely and accurately located through the roadway ID of the roadway where it is located and the horizontal distance from the stope. Among them, the stope is the active area for gold mine exploitation, and the position information of the stope can be judged based on the exploitation progress map. The stope position is the boundary between the exploited area and the unexploited area in roadway 2.

[0050] Taking the roadway width, roadway height, exploitation progress map, material tensor, and the position coding of the target roadway cross-section as the input of the prediction network, the prediction network can learn the material distribution at the position of the target roadway cross-section under the exploitation progress map, and then predict the maximum stress, maximum strain, and plastic zone area of the target roadway cross-section.

[0051] It can be understood that under the condition that the exploitation progress map and the material tensor remain unchanged, by changing the position coding of the target roadway cross-section, the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section under this exploitation progress map can be obtained.

[0052] In one embodiment, the prediction network includes a convolutional sub-network and a regression sub-network. The convolutional sub-network is used to extract features from the mining progress map to obtain progress features. After concatenating the roadway width, roadway height, the progress features, the material tensor, and the position encoding of the target roadway section, the concatenated result is input into the regression sub-network to predict the maximum stress, maximum strain, and plastic zone area of the target roadway section.

[0053] Among them, the convolutional sub-network can adopt existing convolutional neural networks such as ResNet or VGGNet; the regression sub-network is a BP neural network. The progress feature is a column vector of 1 column in A rows, and the value of A is related to the network structure of the convolutional sub-network; the material tensor includes 6 mechanical parameters of the hanging wall surrounding rock, footwall surrounding rock, ore body, and filling material respectively, so the material tensor can be regarded as a column vector of 24 rows in 1 column; the position encoding of the target roadway section can be regarded as a column vector of 2 rows in 1 column, and both the roadway width and roadway height are specific values. Therefore, the input layer of the regression sub-network includes 24 + 2 + 2 + A neurons, and the output layer of the regression sub-network includes 3 neurons, corresponding to the maximum stress, maximum strain, and plastic zone area of the target roadway section respectively.

[0054] In one embodiment, in order to ensure that the prediction network can accurately predict the maximum stress, maximum strain, and plastic zone area of the target roadway section, it is necessary to train the prediction network to constrain the prediction network to learn the mapping relationship between the maximum stress, maximum strain, plastic zone area, and input information.

[0055] Specifically, the training method of the prediction network includes: establishing a mine model based on the material tensor, and numerically simulating the mining process with the roadway width and roadway height to obtain the stress cloud diagram, strain cloud diagram, and plastic zone distribution diagram of the roadway section with any position encoding under any mining progress map, and obtaining the stress labels, strain labels, and plastic zone labels of multiple input samples; inputting the input samples into the prediction network to predict the maximum stress, maximum strain, and plastic zone area, and performing weighted summation on the first deviation between the maximum stress and the stress label, the second deviation between the maximum strain and the strain label, and the third deviation between the plastic zone area and the plastic zone label to obtain a loss function, and performing iterative training on the prediction network; in response to the loss function being less than the preset loss or the number of iterations being greater than the preset number, the training of the prediction network is completed.

[0056] Among them, the preset loss is 0.01; the preset number of times is 300.

[0057] Among them, the process of numerical simulation is implemented using the FLAC3D software (Fast Lagrangian Analysis of Continua). The FLAC3D software is suitable for simulating the deformation of materials and complex deformation states, as well as the deformation of highly non-linear geological materials. It can obtain the stress nephogram, strain nephogram, and plastic zone distribution map of any roadway cross-section. Further, the maximum stress value in the stress nephogram is standardized to obtain a stress label, the maximum strain in the strain nephogram is standardized to obtain a strain label, and the area of the plastic zone in the plastic zone distribution map is standardized to obtain a plastic zone label. That is, the strain label, stress label, and plastic zone label are all dimensionless values.

[0058] Please refer to Figure 2 , which is a schematic diagram of the mine model according to the embodiment of the present application. In this mine model, the mechanical parameters of the ore body, the hanging wall surrounding rock, the footwall surrounding rock, and the filling material used are shown in Table 1.

[0059] Table 1 Mechanical parameters of the ore body, the hanging wall surrounding rock, the footwall surrounding rock, and the filling material used

[0060] Specifically, the loss function satisfies the relationship: ; , and are the maximum stress, maximum strain, and plastic zone area predicted by the prediction network respectively, , and are the stress label, strain label, and plastic zone label respectively, , and are the first weight, second weight, and third weight respectively, and satisfy .

[0061] Among them, the first weight, second weight, and third weight are used to control the degree of attention paid to the maximum stress, maximum strain, and plastic zone area during the training process of the prediction network. For example, when the value of the first weight is relatively large, it means that a relatively large degree of attention is allocated to the maximum stress during the training process of the prediction network. When there is a prediction error in the maximum stress, it can significantly change the value of the loss function, thereby ensuring that the prediction network can output an accurate maximum stress.

[0062] In one embodiment, due to the different degrees of influence of the maximum stress, maximum strain, and the mining progress of the plastic zone area, during the numerical simulation, the change of the maximum stress with the mining progress is analyzed. If the change is greater, it indicates that the maximum stress is more affected by the mining progress. To accurately predict whether the surrounding rock of the mine roadway deforms, the prediction accuracy of the maximum stress needs to be focused on.

[0063] Specifically, the method for obtaining the first weight includes: during the numerical simulation, the maximum stress of any roadway cross-section at different mining progress is statistically analyzed to obtain a stress curve, and the average variance of the stress curves of multiple roadway cross-sections is used as the stress variance; the ratio of the stress variance to the sum of variances is used as the first weight, and the sum of variances is the total of the stress variance, strain variance, and plastic zone area variance.

[0064] Among them, the methods for obtaining the strain variance and the plastic zone area variance are the same as those for obtaining the stress variance. And the value of the second weight is the ratio of the strain variance to the sum of variances; the value of the third weight is the ratio of the plastic zone area variance to the sum of variances.

[0065] Thus, the training of the prediction network is completed, and the trained prediction network can output the maximum stress, maximum strain, and plastic zone area of the target roadway cross-section according to the mining progress map, material tensor, and the position encoding of the target roadway cross-section.

[0066] S103, determine whether the roadway being mined deforms according to the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section. If it deforms, an early warning signal is issued.

[0067] In one embodiment, determining whether the roadway being mined deforms includes: taking the horizontal distance between the target roadway cross-section and the stope as the abscissa and the maximum stress of the target roadway cross-section as the ordinate to obtain a stress scatter curve, using an interpolation algorithm to interpolate the stress scatter curve to obtain the maximum stress of each roadway cross-section in the mined area of the roadway being mined; obtaining the maximum strain and plastic zone area of each roadway cross-section in the mined area of the roadway being mined in the same way; in response to the maximum stress of any roadway cross-section being greater than the maximum allowable stress, or the maximum strain of any roadway cross-section being greater than the maximum allowable strain, or the plastic zone area of any roadway cross-section being greater than the maximum allowable plastic zone area, it is determined that the roadway being mined deforms.

[0068] Among them, the interpolation algorithm uses the radial basis interpolation method or the spatial Kriging interpolation method.

[0069] Among them, the maximum allowable stress, maximum allowable strain, and maximum allowable plastic zone area are set by those skilled in the art according to the Hoek-Brown strength criterion. Taking the maximum stress as an example, the maximum stress of each target roadway section can reflect the change of the maximum stress in the mined area under the mining progress corresponding to the mining progress map. Interpolating the maximum stress of each target roadway section, the maximum stress of each roadway section in the mined area can be obtained. As long as the maximum stress of one roadway section is greater than the maximum allowable stress, it means that the roadway being mined has a deformation risk, and then an early warning is issued.

[0070] It should be noted that the mining progress is collected in real time. When it is monitored that the mining progress increases by a preset length, at least one target roadway section is selected again in the mined area of the roadway being mined to determine whether the roadway being mined deforms after the mining progress changes.

[0071] Among them, the preset length is 5 meters. A laser ranging sensor can be used to monitor the change of the length of the mined area in the roadway being mined. When the mining progress increases by the preset length, it means that the stress condition in the roadway being mined will change. At this time, it is necessary to select the target roadway section again to realize the real-time early warning of the deformation of the surrounding rock of the mine roadway during the mining process, so as to remind the staff to adjust the support parameters in time and ensure the stability of the roadway being mined.

[0072] It should be pointed out that for those of ordinary skill in the art, without departing from the concept of this application, several deformations and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of this patent application shall be subject to the appended claims.

Claims

1. A method for predicting the deformation of surrounding rock in mine roadways based on a neural network, characterized in that, The prediction method includes: selecting at least one target roadway cross-section in the mined area of the roadway being mined; Inputting the roadway width, roadway height, mining progress map, material tensor, and the position encoding of the target roadway cross-section into a prediction network to predict the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section. Among them, the mining progress map includes the mining status of each position point in all roadways, and the mining status includes filled, mined, and unmined; the position encoding includes the roadway ID of the roadway where it is located and the horizontal distance between the target roadway cross-section and the stope; the material tensor includes the mechanical parameters of the hanging wall surrounding rock, footwall surrounding rock, ore body, and filling material; Judge whether the roadway being mined is deformed based on the maximum stress, maximum strain, and plastic zone area of each target roadway cross-section. If deformation occurs, an early warning signal is issued.

2. The method for predicting the deformation of surrounding rock in mine roadways based on a neural network according to claim 1, wherein The mechanical parameters include elastic modulus, Poisson's ratio, internal friction angle, density, tensile strength, and cohesion.

3. The method for predicting the deformation of the surrounding rock of a mine roadway based on a neural network according to claim 1, wherein Selecting at least one target roadway cross-section includes: In the mined area of the roadway being mined, calculate the horizontal distance from each roadway cross-section to the stope; select at least one target roadway cross-section based on the horizontal distance.

4. The method for predicting the deformation of the surrounding rock of a mine roadway based on a neural network according to claim 3, wherein, Selecting at least one target roadway cross-section based on the horizontal distance includes: The interval between any two roadway cross-sections is the absolute value of the difference in horizontal distance, and multiple target roadway cross-sections are selected from each roadway cross-section at a preset interval.

5. The method for predicting the deformation of surrounding rock in a mine roadway based on a neural network according to claim 3, characterized in that, Selecting at least one target roadway cross-section based on the horizontal distance includes: Select the roadway cross-section with a horizontal distance of 0 as the target roadway cross-section, and calculate the selection coefficient, where the selection coefficient is positively correlated with the horizontal distance of the target roadway cross-section; Take the product of the selection coefficient and the minimum interval as the selection interval, and take the roadway cross-section corresponding to the sum of the horizontal distance of the target roadway cross-section and the selection interval as the next target roadway cross-section; Select target roadway cross-sections in sequence until no next target roadway cross-section can be obtained and then stop.

6. The method for predicting the deformation of surrounding rock in mine roadways based on a neural network according to claim 1, characterized in that, The prediction network includes a convolutional sub-network and a regression sub-network; The convolutional sub-network is used to extract features from the mining progress map to obtain progress features; After splicing the roadway width, roadway height, the progress features, material tensor, and the position encoding of the target roadway cross-section, input the splicing result into the regression sub-network to predict the maximum stress, maximum strain, and plastic zone area of the target roadway cross-section.

7. The method for predicting the deformation of the surrounding rock of a mine roadway based on a neural network according to claim 1, wherein The training method of the prediction network includes: establishing a mine model based on the material tensor, and numerically simulating the mining process with the roadway width and roadway height to obtain the stress nephogram, strain nephogram, and plastic zone distribution map of the roadway cross-section with any position encoding under any mining progress map, and obtaining the stress labels, strain labels, and plastic zone labels of multiple input samples; inputting the input samples into the prediction network to predict the maximum stress, maximum strain, and plastic zone area, performing weighted summation on the first deviation between the maximum stress and the stress label, the second deviation between the maximum strain and the strain label, and the third deviation between the plastic zone area and the plastic zone label to obtain a loss function, and performing iterative training on the prediction network; in response to the loss function being less than a preset loss or the number of iterations being greater than a preset number, complete the training of the prediction network.

8. The method for predicting the deformation of surrounding rock in mine roadways based on a neural network according to claim 7, characterized in that The loss function satisfies the relation: ; , and are the maximum stress, maximum strain and plastic zone area predicted by the prediction network, respectively. , and are the stress label, strain label and plastic zone label, respectively. , and are the first weight, second weight and third weight, respectively, and satisfy .

9. The method for predicting the deformation of surrounding rock in a mine roadway based on a neural network according to claim 7, characterized in that, The method for obtaining the first weight includes: During the numerical simulation of the mining process, the maximum stress of any roadway cross-section at different mining progress is statistically analyzed to obtain a stress curve, and the mean variance of the stress curves of multiple roadway cross-sections is used as the stress variance; The ratio of the stress variance to the sum of variances is used as the first weight, and the sum of variances is the total of the stress variance, the strain variance, and the plastic zone area variance.

10. The method for predicting the deformation of surrounding rock in mine roadways based on a neural network according to claim 1, characterized in that Judging whether the roadway being mined is deformed includes: Taking the horizontal distance between the target roadway cross-section and the stope as the abscissa and the maximum stress of the target roadway cross-section as the ordinate to obtain a stress scatter curve, and using an interpolation algorithm to interpolate the stress scatter curve to obtain the maximum stress of each roadway cross-section in the mined area of the roadway being mined; Obtaining the maximum strain and plastic zone area of each roadway cross-section in the mined area of the roadway being mined according to the same steps; In response to the maximum stress of any roadway cross-section being greater than the maximum allowable stress, or the maximum strain of any roadway cross-section being greater than the maximum allowable strain, or the plastic zone area of any roadway cross-section being greater than the maximum allowable plastic zone area, it is determined that the roadway being mined is deformed.

Images