A coal mine tunneling water inflow prediction method and device

Abstract

The present application provides a kind of coal mine roadway excavation water inflow prediction method and device, comprising the following steps: step 1, the geological key parameters of coal mine roadway excavation section are explored to obtain, the water inflow in the excavation roadway is collected by collection device;Step 2, establish roadway excavation water inflow prediction data sample;Step 3, extract the main features of water inflow prediction data;Step 4, based on the training of characteristic vector sample, construct multiple prediction models;Step 5, the optimal value of the parameters of the predicted model is obtained;Step 6, construct multiple optimized prediction models;Step 7, input the key parameter survey value of roadway unexcavated area, obtain the prediction result of coal mine roadway excavation water inflow.The present application can quickly obtain water inflow value based on water inflow collection device, effectively improve the prediction accuracy of coal mine roadway excavation water inflow by considering the water inflow prediction result of multiple characteristic parameters.

Description

Technical Field

[0001] This invention belongs to the field of tunnel construction technology, specifically relating to a method and device for predicting water inflow during coal mine tunnel excavation. Background Technology

[0002] Water inrush is a common occurrence during coal mine roadway excavation. Excessive water inrush can easily lead to water inrush disasters. Therefore, it is necessary to accurately predict water inrush during coal mine roadway excavation and take effective countermeasures to ensure the safety of roadway construction.

[0003] Currently, analytical, empirical, and numerical methods are commonly used to estimate tunnel water inflow. However, analytical solutions rely on given hydrogeological assumptions and cannot predict water flow under complex hydrogeological conditions, such as fractured rock strata. Furthermore, tunnel water inflow problems under various complex geological conditions can also be analyzed through numerical simulations. However, numerical models require a realistic generalized model based on the input geological data. Moreover, they either fail to consider the assessment of water inrush risk or do not deeply reveal the quantitative relationship between water inrush and influencing factors. Therefore, due to the given hydrogeological assumptions and the simplification of these heterogeneous media, predicting potential tunnel water inflow using analytical solutions and numerical tools often fails. Simultaneously, the above methods suffer from slow computational efficiency in water inflow calculations; in reality, the entire geological condition can be considered a chaotic system. Therefore, general exploration methods cannot accurately detect hydrogeological features.

[0004] Based on this, a method and device for predicting water inflow during coal mine roadway excavation are proposed. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method and apparatus for predicting water inflow during coal mine roadway excavation, in order to address the shortcomings of the prior art mentioned above.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0007] In a first aspect, a device for predicting water inflow during coal mine roadway excavation includes a first collection slot and a second collection slot, a movable partition and a telescopic third collection slot.

[0008] The first collection tank consists of a partition, a top steel mesh, and a front steel mesh. The top steel mesh is located on the top surface of the partition and is used to catch larger-diameter falling rocks while ensuring that all the water flowing in the tunnel can flow into the water volume collection device. The front steel mesh is located at the front end of the partition and is used to filter sand and gravel from the water flowing in the tunnel.

[0009] The second collection slot is equipped with a depth scale, and a buoyancy ball is also provided inside the second collection slot.

[0010] The movable partition is provided with a groove, and a tension rod, a rotating shaft, a rotating rod, a compression rod, and a compression U-shaped rod are provided inside the groove. A first spring and a second spring are provided at the connection between the movable partition and the second collection slot.

[0011] The third collection slot is provided with a groove depth. The third sliding groove on the first partition of the third collection slot is a convex sliding groove, which is located inside the first partition. The first sliding groove of the second collection slot is a concave sliding groove, which is located outside the second collection slot. The third sliding groove and the first sliding groove are fastened together to realize the connection and sliding of the first partition. The fourth sliding groove on the second partition is a convex sliding groove, which is located inside the second partition. The second sliding groove of the second collection slot is a concave sliding groove, which is located outside the second collection slot. The fourth sliding groove and the second sliding groove are fastened together to realize the connection and sliding of the second partition. The volume of the third collection slot can be changed by extending and retracting the first partition and the second partition.

[0012] The buoyancy ball in the second collection tank is connected to the tension rod via a connecting rod. When the water flow in the second collection tank reaches its capacity, the flow continues to increase, causing the buoyancy ball to rise above its allowable height. The rising buoyancy ball pulls the tension rod, which in turn drives the movement of the compression rod and the U-shaped compression rod to complete the sliding operation of the movable partition, creating an opening. This allows the water in the second collection tank to flow into the third collection tank. The movable partition moves, compressing the first and second springs. When the water flow in the second collection tank returns to the allowable height, the buoyancy ball falls back down, and the first and second springs close the movable partition, preventing the water in the second collection tank from flowing into the third collection tank.

[0013] Secondly, a prediction method for a coal mine roadway excavation water inflow prediction device includes the following steps:

[0014] Step 1: Explore and obtain key geological parameters of the coal mine roadway excavation section, and collect the water inflow in the coal mine roadway using a prediction device.

[0015] Step 2: Establish a sample of predicted water inflow data for coal mine roadway excavation;

[0016] Step 3: Extract the main features of the water inflow prediction data;

[0017] Step 4: Train multiple prediction models based on feature vector samples, and establish a nonlinear mapping relationship between input parameters and output parameters through multiple prediction models;

[0018] The mapping relationship is expressed as S = TR(Z).

[0019] In the formula, S is the predicted output parameter, which is the inflow data.

[0020] TR is the expression form of the established mapping relationship;

[0021] Z is the parameter for prediction input, Z = [z1, z2, z3, z4, z5, z6], z1 is the degree of development of surrounding rock fractures, z2 is the water permeability of fractured surrounding rock, z3 is the distance between the overlying fractured zone and the roadway roof, z4 is the water-bearing characteristics of the surrounding rock aquifer, z5 is the pore water pressure of the overlying strata, and z6 is the roadway depth.

[0022] Step 5: Use an improved machine optimization algorithm to obtain the optimal values ​​for the parameters of the prediction model, the number of prediction models Q, and the maximum number of features L.

[0023] Step 6: Assign the optimal number of prediction models Q and the maximum number of features L obtained by the improved machine optimization algorithm to the aforementioned prediction model, train the established feature vector samples, and obtain the optimized prediction model.

[0024] Step 7: Input the key parameter survey values ​​of the unconstructed area of ​​the roadway, predict the water inflow in the unconstructed area of ​​the roadway through the optimized prediction model, obtain the prediction results of the water inflow during the subsequent roadway excavation, and formulate corresponding treatment measures based on the water inflow prediction results.

[0025] As a further explanation of the present invention, the key geological parameters mentioned in step 1 include the degree of development of surrounding rock fissures, the permeability of fractured surrounding rock, the distance between the overlying fractured zone and the roadway roof, the water-bearing characteristics of the surrounding rock aquifer, the pore water pressure of the overlying strata, and the roadway depth.

[0026] As a further explanation of the present invention, the specific data for collecting the water inflow in step 1 are as follows:

[0027] The stated water inflow ;

[0028] In the formula: A is the width of the prediction device;

[0029] The width of the second acquisition slot B1;

[0030] H1 is the water level in the second collection tank;

[0031] B2 is the width of the third acquisition slot;

[0032] H2 is the water level in the third collection tank.

[0033] As a further explanation of the present invention, the data sample mentioned in step 2 includes input parameters and output parameters;

[0034] The input parameters include the degree of development of surrounding rock fractures, the permeability of fractured surrounding rock, the distance between the overlying fractured zone and the roadway roof, the water-bearing characteristics of the surrounding rock aquifer, the pore water pressure of the overlying strata, and the roadway depth.

[0035] The output parameter is the water inflow during coal mine roadway excavation.

[0036] As a further explanation of the present invention, the main feature of extracting the water inflow prediction data in step 3 is to use convolutional pooling operation to extract key parameter feature values ​​from the aforementioned collected prediction data samples to form feature vector samples. The formation of feature vector samples includes the following steps.

[0037] The collected data, including the degree of development of surrounding rock fissures, permeability of fractured surrounding rock, distance between the overlying fractured zone and the roadway roof, water-bearing characteristics of the surrounding rock aquifer, pore water pressure of the overlying strata, roadway depth, and water inflow during roadway excavation, are divided and standardized. Through convolutional pooling, a one-dimensional feature sequence is output, thereby forming a feature vector sample.

[0038] As a further explanation of the present invention, step 5 includes the following steps:

[0039] Initialize the population positions, set the population size to n, and have n seekers. Each position is assigned a position with a position of n. This indicates that the fitness value is set, and the fitness value function is set as follows:

[0040] ;

[0041] In the formula: S i This represents the actual water inflow rate.

[0042] This is the predicted water inflow value;

[0043] n represents the nth sample;

[0044] Individuals with high fitness values ​​are selected as discoverers to search for results. The search range of discoverers is larger than that of joiners. During each iteration of optimization, the position is updated according to the following formula:

[0045] ;

[0046] In the formula: For the (t+1)th iteration, the... i The position of the optimizer in the k-th dimension;

[0047] For the first t During the nth iteration i The first seeker in the first k The position of the dimension;

[0048] V is a random number that follows a normal distribution;

[0049] H is a 1×d-dimensional sequence with all elements being 1;

[0050] Y Z This is the warning value, and its range is [0, 1].

[0051] A Z For a safe value, A Z The value range is [0.5, 1].

[0052] ;

[0053] Participants are judged and acquire energy based on the quality of the result and their position. When a participant notices that the discoverer has obtained a better result, it will try to compete for it. If the participant succeeds, it can replace the discoverer's position. The participant is updated according to the following formula:

[0054] ;

[0055] In the formula: This represents the globally optimal position at the t-th iteration.

[0056] These are the step size and direction control parameters, which are random numbers that follow a standard normal distribution;

[0057] It is a random number, and its value ranges from [-1, 1].

[0058] The fitness value represents the globally optimal position.

[0059] This represents the fitness value for the worst-case position globally. It is the smallest constant.

[0060] As a further explanation of the present invention, the formation process of the prediction model after optimization in step 6 is as follows:

[0061] Q distinctive training sets are generated from the established feature vector samples by sampling with replacement, and Q distinctive prediction models are established.

[0062] The total dimension of the features is set to L. The nodes are split based on the optimal splitting strategy to maximize the use of information and minimize the error, and then multiple optimized prediction models are formed.

[0063] A weighted average method is used to achieve prediction for a single optimized prediction model. By averaging the prediction results of multiple optimized prediction models, the mapping relationship between key geological parameters of the excavation section and water inflow during tunnel excavation is finally established.

[0064] As a further explanation of the present invention, the process by which the prediction model optimized in step 7 predicts the water inflow in the unexcavated area of ​​a coal mine roadway is as follows:

[0065] Key parameters for the unexcavated sections of coal mine roadways, including the degree of development of surrounding rock fissures, the permeability of fractured surrounding rock, the distance between the overlying fractured zone and the roadway roof, the water-bearing characteristics of the surrounding rock aquifer, the pore water pressure of the overlying strata, and the roadway depth, are surveyed in advance. These survey parameters are then input into an optimized prediction model, which outputs the predicted water inflow value for the unexcavated sections. Based on the predicted water inflow results, corresponding disposal measures are formulated in advance to ensure the smooth and safe construction of the roadway.

[0066] Compared with the prior art, the present invention has the following advantages:

[0067] This invention enables rapid and relatively accurate acquisition of water inflow in unexcavated sections of coal mine roadways, which can effectively guide the formulation of excavation support and drainage measures, and is of great significance for promoting safe production. Attached Figure Description

[0068] Figure 1 This is a flowchart of the method of the present invention;

[0069] Figure 2 This is a schematic diagram of the prediction device of the present invention;

[0070] Figure 3 This is a schematic diagram of the internal structure connection of the movable partition of the present invention;

[0071] Figure 4 This is a schematic diagram of the buoyancy sphere connection structure of the present invention;

[0072] Figure 5 A graph is created for the feature vector samples of this invention;

[0073] Figure 6 A graph is created for the prediction model of this invention;

[0074] Figure 7 This is a comparison chart of the prediction results and monitoring results of this invention.

[0075] Explanation of reference numerals in the attached figures:

[0076] First collection slot 1, second collection slot 2, movable partition 3, telescopic third collection slot 4, partition 1-1, top steel mesh 1-2, front steel mesh 1-3, slot depth scale 2-1, buoyancy ball 2-2, groove 3-1, tension rod 3-2, rotating shaft 3-3, rotating rod 3-4, compression rod 3-5, compression U-shaped rod 3-6, first spring 3-7, second spring 3-8, slot depth scale 4-1, third slide 4-2, fourth slide 4-3; first partition 4-4; second partition 4-5. Detailed Implementation

[0077] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0078] like Figure 1-7 As shown, the present invention provides a technical solution: a coal mine roadway excavation water inflow prediction device, comprising a first collection tank 1 and a second collection tank 2, a movable partition 3 and a telescopic third collection tank 4;

[0079] The first collection tank 1 consists of a partition plate 1-1, a top reinforcing mesh 1-2, and a front reinforcing mesh 1-3. The top reinforcing mesh 1-2 is located on the top surface of the partition plate 1-1, and the front reinforcing mesh 1-3 is located at the front end of the partition plate 1-1. The top reinforcing mesh 1-2 is disassembled and replaced according to the actual particle size in the coal mine roadway. The top reinforcing mesh 1-2 is used to catch larger-sized falling rocks from the top, while ensuring that all the water flowing in the roadway can flow into the water flow collection device. The front reinforcing mesh 1-3 is disassembled and replaced according to the particle size of the sand and gravel in the first collection tank. The front reinforcing mesh is used to filter the sand and gravel in the roadway water flow, ensuring the filtration of sand and gravel in the roadway water flow and ensuring the accuracy of the collected water flow value.

[0080] The second collection slot 2 is provided with a slot depth height scale 2-1 inside, and a buoyancy ball 2-2 is also provided inside the second collection slot 2;

[0081] The movable partition 3 is provided with a groove 3-1. The groove 3-1 is provided with a tension rod 3-2, a rotating shaft 3-3, a rotating rod 3-4, a compression rod 3-5, and a compression U-shaped rod 3-6. The movable partition 3 is provided with a first spring 3-7 and a second spring 3-8 at the connection between it and the second collection groove 2.

[0082] The third collection slot 4 is provided with a groove depth 4-1. The third sliding groove 4-2 on the first partition 4-4 of the third collection slot 4 is a convex sliding groove, and the third sliding groove 4-2 is located inside the first partition 4-4. The first sliding groove 2-4 of the second collection slot 2 is a concave sliding groove, and the first sliding groove 2-4 is located outside the second collection slot 2. The third sliding groove 4-2 and the first sliding groove 2-4 are fastened together to realize the connection and sliding of the first partition 4-4. The fourth sliding groove 4-3 on the second partition 4-5 is a convex sliding groove, and the fourth sliding groove 4-3 is located inside the second partition 4-5. The second sliding groove 2-5 of the second collection slot 2 is a concave sliding groove, and the second sliding groove 2-5 is located outside the second collection slot 2. The fourth sliding groove 4-3 and the second sliding groove 2-5 are fastened together to realize the connection and sliding of the second partition 4-5. The volume of the third collection slot 4 can be changed by extending and retracting the first partition 4-4 and the second partition 4-5.

[0083] The buoyancy ball 2-2 in the second collection tank 2 is connected to the tension rod 3-2 via the connecting rod 2-3. When the water flow in the second collection tank 2 reaches its capacity, the water flow continues to increase, causing the buoyancy ball 2-2 to rise above its allowable height. The rise of the buoyancy ball 2-2 causes the tension rod 3-2 to pull, and the movement of the rotating rod 3-4 drives the compression rod 3-5 and the compression U-shaped rod 3-6 to complete the sliding operation of the moving partition 3, creating an opening so that the water in the second collection tank 2 flows into the third collection tank 4. The moving partition moves, compressing the first spring 3-7 and the second spring 3-8. When the water flow in the second collection tank 2 returns to the allowable height, the buoyancy ball 2-2 falls back, and the first spring 3-7 and the second spring 3-8 drive the moving partition 3 to close, and the water in the second collection tank 2 no longer flows into the third collection tank 4.

[0084] The prediction method of the above-mentioned coal mine roadway excavation water inflow prediction device includes the following steps:

[0085] Step 1: Explore and obtain key geological parameters of the coal mine roadway excavation section, and collect the water inflow in the coal mine roadway using a prediction device.

[0086] The key geological parameters include the degree of development of surrounding rock fractures, permeability of fractured surrounding rock, distance between the overlying fractured zone and the roadway roof, water-bearing characteristics of the surrounding rock aquifer, pore water pressure of the overlying strata, and roadway depth.

[0087] The specific data on water inflow are as follows:

[0088] The stated water inflow ;

[0089] In the formula: A is the width of the prediction device;

[0090] The width of the second acquisition slot B1;

[0091] H1 is the water level in the second collection tank;

[0092] B2 is the width of the third acquisition slot;

[0093] H2 is the water level in the third collection tank.

[0094] Step 2: Establish a sample of predicted water inflow data for coal mine roadway excavation;

[0095] The data sample includes input parameters and output parameters;

[0096] The input parameters include the degree of development of surrounding rock fractures, the permeability of fractured surrounding rock, the distance between the overlying fractured zone and the roadway roof, the water-bearing characteristics of the surrounding rock aquifer, the pore water pressure of the overlying strata, and the roadway depth.

[0097] The output parameter is the water inflow during coal mine roadway excavation.

[0098] Step 3: Extract the main features of the water inflow prediction data;

[0099] The main feature of extracting the water inflow prediction data is to use convolutional pooling operation to extract key parameter feature values ​​from the aforementioned collected prediction data samples to form feature vector samples. The formation of feature vector samples includes the following steps:

[0100] The collected data, including the degree of development of surrounding rock fissures, permeability of fractured surrounding rock, distance between the overlying fractured zone and the roadway roof, water-bearing characteristics of the surrounding rock aquifer, pore water pressure of the overlying strata, roadway depth, and water inflow during roadway excavation, are divided and standardized. Through convolutional pooling, a one-dimensional feature sequence is output, thereby forming a feature vector sample.

[0101] Step 4: Train multiple prediction models based on feature vector samples, and establish a nonlinear mapping relationship between input parameters and output parameters through multiple prediction models;

[0102] The mapping relationship is expressed as S = TR(Z).

[0103] In the formula, S is the predicted output parameter, which is the inflow data;

[0104] TR is the expression form of the established mapping relationship;

[0105] Z is the parameter for prediction input, Z = [z1, z2, z3, z4, z5, z6], z1 is the degree of development of surrounding rock fractures, z2 is the water permeability of fractured surrounding rock, z3 is the distance between the overlying fractured zone and the roadway roof, z4 is the water-bearing characteristics of the surrounding rock aquifer, z5 is the pore water pressure of the overlying strata, and z6 is the roadway depth.

[0106] Step 5: Use an improved machine optimization algorithm to obtain the optimal values ​​for the parameters of the prediction model, the number of prediction models Q, and the maximum number of features L.

[0107] Includes the following steps:

[0108] Initialize the population positions, set the population size to n, and have n seekers. Each position is assigned a position with a position of n. This indicates that the fitness value is set, and the fitness value function is set as follows:

[0109] ;

[0110] In the formula: S i This represents the actual water inflow rate.

[0111] This is the predicted water inflow value;

[0112] n represents the nth sample;

[0113] Individuals with high fitness values ​​are selected as discoverers to search for results. The search range of discoverers is larger than that of joiners. During each iteration of optimization, the position is updated according to the following formula:

[0114] ;

[0115] In the formula: For the (t+1)th iteration, the... i The position of the optimizer in the k-th dimension;

[0116] For the first t During the nth iteration i The first seeker in the first k The position of the dimension;

[0117] V is a random number that follows a normal distribution;

[0118] H is a 1×d-dimensional sequence with all elements being 1;

[0119] Y Z This is the warning value, and its range is [0, 1].

[0120] A Z For a safe value, A Z The value range is [0.5, 1].

[0121] ;

[0122] Participants are judged and acquire energy based on the quality of the result and their position. When a participant notices that the discoverer has obtained a better result, it will try to compete for it. If the participant succeeds, it can replace the discoverer's position. The participant is updated according to the following formula:

[0123] ;

[0124] In the formula: This represents the globally optimal position at the t-th iteration.

[0125] These are the step size and direction control parameters, which are random numbers that follow a standard normal distribution;

[0126] It is a random number, and its value ranges from [-1, 1].

[0127] The fitness value represents the globally optimal position.

[0128] This represents the fitness value for the worst-case position globally. It is the smallest constant.

[0129] Step 6: Assign the optimal number of prediction models Q and the maximum number of features L obtained by the improved machine optimization algorithm to the aforementioned prediction model, train the established feature vector samples, and obtain the optimized prediction model.

[0130] The process of forming the prediction model after optimization in step 6 is as follows:

[0131] Q distinctive training sets are generated from the established feature vector samples by sampling with replacement, and Q distinctive prediction models are established.

[0132] The total dimension of the features is set to L. The nodes are split based on the optimal splitting strategy to maximize the use of information and minimize the error, and then multiple optimized prediction models are formed.

[0133] A weighted average method is used to achieve predictions for a single optimized prediction model. By averaging the prediction results of multiple optimized prediction models, a mapping relationship between key geological parameters of the excavation section and water inflow during tunnel excavation is finally established.

[0134] Step 7: Input the key parameter survey values ​​of the unconstructed area of ​​the roadway, predict the water inflow in the unconstructed area of ​​the roadway through the optimized prediction model, obtain the prediction results of the water inflow in the later roadway excavation, and formulate corresponding treatment measures based on the water inflow prediction results.

[0135] Key parameters for the unexcavated sections of coal mine roadways, including the degree of development of surrounding rock fissures, the permeability of fractured surrounding rock, the distance between the overlying fractured zone and the roadway roof, the water-bearing characteristics of the surrounding rock aquifer, the pore water pressure of the overlying strata, and the roadway depth, are surveyed in advance. These survey parameters are then input into an optimized prediction model, which outputs the predicted water inflow value for the unexcavated sections. Based on the predicted water inflow results, corresponding disposal measures are formulated in advance to ensure the smooth and safe construction of the roadway.

[0136] This invention enables rapid and relatively accurate acquisition of water inflow in unexcavated sections of coal mine roadways, which can effectively guide the formulation of excavation support and drainage measures, and is of great significance for promoting safe production.

[0137] In an experimental example, during the tunnel excavation process at Wanfu Coal Mine in Shandong Province, key geological parameters of the excavated tunnel were surveyed and collected: z1 (surrounding rock fissure development degree), z2 (fractured surrounding rock permeability), z3 (distance between the overlying fractured zone and the tunnel roof), z4 (water-bearing characteristics of the surrounding rock aquifer), z5 (pore water pressure of the overlying strata), and z6 (tunnel depth). The water inflow S at the corresponding collection points was also collected using a data acquisition device. The collected data are shown in the table below.

[0138] ;

[0139] Geological key parameters were surveyed and numbered 1 to 8 at eight cross-sections (ZK91+651~634) of the unexcavated section of the tunnel in the study area, and water inflow was predicted. The prediction results and the actual water inflow data collected after subsequent tunneling are as follows: Figure 7 As shown,

[0140] As can be seen from the figure, the predicted water inflow at different cross-sections in the unexcavated section is quite close to the actual water inflow collected after excavation, which verifies the feasibility and accuracy of the method. This prediction method can effectively guide the preparation of water inflow plans in the later excavation process and effectively avoid the occurrence of water inrush disasters during construction.

[0141] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0142] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A device for predicting water inflow during coal mine roadway excavation, characterized in that: It includes a first collection slot (1), a second collection slot (2), a movable partition (3), and a telescopic third collection slot (4); The first collection tank (1) consists of a partition (1-1), a top steel mesh (1-2), and a front steel mesh (1-3). The top steel mesh (1-2) is located on the top surface of the partition (1-1) and is used to receive falling rocks with a top particle size larger than the mesh aperture, while ensuring that all the water flowing in the tunnel can flow into the water flow collection device. The front steel mesh (1-3) is located at the front end of the partition (1-1) and is used to filter sand and gravel from the water flowing in the tunnel. The second collection slot (2) is provided with a slot depth height scale (2-1), and a buoyancy ball (2-2) is also provided inside the second collection slot (2). The movable partition (3) is provided with a groove (3-1), and a tension rod (3-2), a rotating shaft (3-3), a rotating rod (3-4), a compression rod (3-5), and a compression U-shaped rod (3-6) are provided inside the groove (3-1). A first spring (3-7) and a second spring (3-8) are provided at the connection between the movable partition (3) and the second collection groove (2). The third collection groove (4) is provided with a groove depth (4-1). The third sliding groove (4-2) on the first partition (4-4) of the third collection groove (4) is a convex sliding groove. The third sliding groove (4-2) is located inside the first partition (4-4). The first sliding groove (2-4) of the second collection groove (2) is a concave sliding groove. The first sliding groove (2-4) is located outside the second collection groove (2). The third sliding groove (4-2) and the first sliding groove (2-4) are fastened together to realize the connection and sliding of the first partition (4-4). The fourth sliding groove (4-3) on the second partition (4-5) inside the groove (4) is a convex sliding groove. The fourth sliding groove (4-3) is located inside the second partition (4-5). The second sliding groove (2-5) of the second collection groove (2) is a concave sliding groove. The second sliding groove (2-5) is located outside the second collection groove (2). The fourth sliding groove (4-3) and the second sliding groove (2-5) are fastened together to realize the connection and sliding of the second partition (4-5). The volume of the third collection groove (4) is changed by extending and retracting the first partition (4-4) and the second partition (4-5). The buoyancy ball (2-2) in the second collection tank (2) is connected to the tension rod (3-2) through the connecting rod (2-3). When the water flow in the second collection tank (2) reaches its capacity value, the water flow continues to increase, causing the buoyancy ball (2-2) to rise above its allowable height. The rise of the buoyancy ball (2-2) causes the tension rod (3-2) to pull. Through the rotation rod (3-4), the compression rod (3-5) and the compression U-shaped rod (3-6) are driven to complete the sliding operation of the moving partition (3), creating an opening so that the water flow in the second collection tank (2) flows into the third collection tank (4). The moving partition moves, compressing the first spring (3-7) and the second spring (3-8). When the water flow in the second collection tank (2) returns to the allowable height, the buoyancy ball (2-2) falls back. The first spring (3-7) and the second spring (3-8) drive the moving partition (3) to close, and the water flow in the second collection tank (2) no longer flows into the third collection tank (4).

2. The prediction method for a coal mine roadway excavation water inflow prediction device according to claim 1, characterized in that, Includes the following steps: Step 1: Explore and obtain key geological parameters of the coal mine roadway excavation section, and collect the water inflow in the coal mine roadway using a prediction device. Step 2: Establish a sample of predicted water inflow data for coal mine roadway excavation; Step 3: Extract the main features of the water inflow prediction data; Step 4: Train multiple prediction models based on feature vector samples, and establish a nonlinear mapping relationship between input parameters and output parameters through multiple prediction models; The mapping relationship is expressed as S = TR(Z). In the formula, S is the predicted output parameter, which is the inflow data. TR is the expression form of the established mapping relationship; Z is the parameter for prediction input, Z = [z1, z2, z3, z4, z5, z6], z1 is the degree of development of surrounding rock fractures, z2 is the water permeability of fractured surrounding rock, z3 is the distance between the overlying fractured zone and the roadway roof, z4 is the water-bearing characteristics of the surrounding rock aquifer, z5 is the pore water pressure of the overlying strata, and z6 is the roadway depth. Step 5: Use an improved machine optimization algorithm to obtain the optimal values ​​for the parameters of the prediction model, the number of prediction models Q, and the maximum number of features L. Step 6: Assign the optimal number of prediction models Q and the maximum number of features L obtained by the improved machine optimization algorithm to the aforementioned prediction model, train the established feature vector samples, and obtain the optimized prediction model. Step 7: Input the key parameter survey values ​​of the unconstructed area of ​​the roadway, predict the water inflow in the unconstructed area of ​​the roadway through the optimized prediction model, obtain the prediction results of the water inflow during the subsequent roadway excavation, and formulate corresponding treatment measures based on the water inflow prediction results.

3. The prediction method for a coal mine roadway excavation water inflow prediction device according to claim 2, characterized in that, The key geological parameters mentioned in step 1 include the degree of development of surrounding rock fissures, the permeability of fractured surrounding rock, the distance between the overlying fractured zone and the tunnel roof, the water-bearing characteristics of the surrounding rock aquifer, the pore water pressure of the overlying strata, and the tunnel depth.

4. The prediction method for a coal mine roadway water inflow prediction device according to claim 2, characterized in that, The specific data for water inflow collection in step 1 are as follows: The stated water inflow ; In the formula: A is the width of the prediction device; The width of the second acquisition slot B1; H1 is the water level in the second collection tank; B2 is the width of the third acquisition slot; H2 is the water level in the third collection tank.

5. The prediction method for a coal mine roadway water inflow prediction device according to claim 2, characterized in that, The data sample mentioned in step 2 includes input parameters and output parameters; The input parameters include the degree of development of surrounding rock fractures, the permeability of fractured surrounding rock, the distance between the overlying fractured zone and the roadway roof, the water-bearing characteristics of the surrounding rock aquifer, the pore water pressure of the overlying strata, and the roadway depth. The output parameter is the water inflow during coal mine roadway excavation.

6. The prediction method for a coal mine roadway excavation water inflow prediction device according to claim 5, characterized in that, The main feature of extracting the water inflow prediction data in step 3 is to use convolutional pooling to extract key parameter feature values ​​from the aforementioned collected prediction data samples to form feature vector samples. The formation of feature vector samples includes the following steps: The collected data, including the degree of development of surrounding rock fissures, permeability of fractured surrounding rock, distance between the overlying fractured zone and the roadway roof, water-bearing characteristics of the surrounding rock aquifer, pore water pressure of the overlying strata, roadway depth, and water inflow during roadway excavation, are divided and standardized. Through convolutional pooling, a one-dimensional feature sequence is output, thereby forming a feature vector sample.

7. The prediction method for a coal mine roadway excavation water inflow prediction device according to claim 2, characterized in that, Step 5 includes the following steps: Initialize the population positions, set the population size to n, and have n seekers. Each position is assigned a position with a position of n. This indicates that the fitness value is set, and the fitness value function is set as follows: ； In the formula: S i This represents the actual water inflow rate. This is the predicted water inflow value; n represents the nth sample; Individuals with high fitness values ​​are selected as discoverers to search for results. The search range of discoverers is larger than that of joiners. During each iteration of optimization, the position is updated according to the following formula: ； In the formula: For the (t+1)th iteration, the... i The position of the optimizer in the k-th dimension; For the first t During the nth iteration i The first seeker in the first k The position of the dimension; V is a random number that follows a normal distribution; H is a 1×d-dimensional sequence with all elements being 1; Y Z This is the warning value, and its range is [0, 1]. A Z For a safe value, A Z The value range is [0.5, 1]; ； Participants are judged and acquire energy based on the quality of the result and their position. When a participant notices that the discoverer has obtained a better result, it will try to compete for it. If the participant succeeds, it can replace the discoverer's position. The participant is updated according to the following formula: ； In the formula: This represents the globally optimal position at the t-th iteration. These are the step size and direction control parameters, which are random numbers that follow a standard normal distribution; It is a random number, and its value ranges from [-1, 1]. The fitness value represents the globally optimal position. This represents the fitness value for the worst-case position globally. It is the smallest constant.

8. The prediction method for a coal mine roadway water inflow prediction device according to claim 2, characterized in that, The process of forming the prediction model after optimization in step 6 is as follows: Q distinctive training sets are generated from the established feature vector samples by sampling with replacement, and Q distinctive prediction models are established. The total dimension of the features is set to L. The nodes are split based on the optimal splitting strategy to maximize the use of information and minimize the error, and then multiple optimized prediction models are formed. A weighted average method is used to achieve prediction for a single optimized prediction model. By averaging the prediction results of multiple optimized prediction models, the mapping relationship between key geological parameters of the excavation section and water inflow during tunnel excavation is finally established.

9. The prediction method for a coal mine roadway excavation water inflow prediction device according to claim 2, characterized in that, Step 7, after optimization, describes the prediction process for the water inflow in the unexplored areas of coal mine roadways: Key parameters for the unexcavated sections of coal mine roadways, including the degree of development of surrounding rock fissures, the permeability of fractured surrounding rock, the distance between the overlying fractured zone and the roadway roof, the water-bearing characteristics of the surrounding rock aquifer, the pore water pressure of the overlying strata, and the roadway depth, are surveyed in advance. These survey parameters are then input into an optimized prediction model, which outputs the predicted water inflow value for the unexcavated sections. Based on the predicted water inflow results, corresponding disposal measures are formulated in advance to ensure the smooth and safe construction of the roadway.

Images