A method for evaluating the correlation between water level in long-hole observation holes and mining dynamics at working faces

Abstract

The present invention discloses a method for evaluating the correlation between the water level of a long observation hole and mining activity on a working face, which belongs to the field of mining engineering technology and includes the following steps: Step 1, collecting water level response data as a basic data set; Step 2, drawing a daily variation curve of the water level of the long observation hole of the basic data set, and constructing a deep learning training total data set; Step 3, screening the deep learning training total data set to obtain a deep learning training data set; Step 4, constructing a feature learning deep neural network model, and completing the training using the deep learning training data set; Step 5, during the mining process of the working face to be evaluated, using the feature learning deep neural network model to classify the daily variation curves of the water level of the surrounding long observation holes for correlation; Step 6, evaluating the correlation between the water level of the long observation hole and mining activity on the working face to be evaluated in stages. The present invention can select a long observation hole of a confined aquifer with better early warning applicability by evaluating the correlation between the water level fluctuation of the long observation hole and mining activity on the working face to be evaluated.

Description

Technical Field

[0001] The invention belongs to the technical field of mining engineering, and in particular relates to a method for evaluating the correlation between a long-observation hole water level and working face mining. Background Art

[0002] Currently, some mineable Permian coal seams are hidden beneath thick, unconsolidated Cenozoic layers. These unconsolidated, porous, confined aquifers (primarily the Quaternary Aquifer, hereinafter referred to as the "Quaternary Aquifer") are thick and have high water pressure. Under certain mining conditions, coal mining faces present a significant risk of water inrush, posing a serious threat to miners' lives and the normal operation of the coal mine. Prior to coal mining operations, establishing a water inrush monitoring and early warning system and a response plan for working faces at risk of water inrush from the "Quaternary Aquifer" is crucial for ensuring safe production.

[0003] Engineering practice has shown that changes in the water level of the "four-containment" confined aquifer can reflect the characteristics of overburden movement caused by mining at the working face. When the working face has developed water-conducting fracture zones that do not affect the confined aquifer, the linkage between mining at the working face and the "four-containment" water level is manifested as follows: the water level in the long observation hole exhibits a significant daily fluctuation and recovery phenomenon with mining at the working face. That is, when the production team resumes mining, the roof depressurizes and the water level drops. When the maintenance team stops mining, the "four-containment" is replenished and the water level rises. The amplitude of this fluctuation varies depending on the aquifer characteristics and mining status. In other words, during the safe mining process of the working face, the water level of the "four-containment" above it changes in tandem with the mining operations at the working face. When it suddenly deviates from this linkage pattern, it indicates the possibility of a disaster. Based on this change, personnel implement water inrush warnings. For example, when the water level of the confined aquifer continues to decline and does not rise, it indicates the possibility of a sudden and concentrated influx of large amounts of groundwater into the wells and tunnels. A water inrush warning is issued, and personnel conduct relevant investigations to prevent water inrush disasters.

[0004] The water levels of the "four containments" are measured through hydrological observation holes located within the mining area. During on-site mining operations, the water levels in the observation holes serve as a key indicator for warning of water inrush risks from the "four containments" support. Under complex hydrogeological conditions, the water levels in the multiple observation holes surrounding the working face vary in their response to overburden movement. When the water levels in the observation holes are strongly correlated with mining activity at the working face, their changes provide greater value in providing early warning of water inrush. Otherwise, false alarms and missed warnings can occur, impacting normal production processes or even rendering the early warning function ineffective.

[0005] However, there is currently no evaluation method for evaluating the strength of the correlation between the water level in the long observation hole and coal mining. As a result, many mines do not evaluate the strength of the correlation between the water level in the long observation hole and the mining of the working face when constructing the water inrush warning system. It is easy for the long observation hole to be not applicable, resulting in the failure to fully utilize its warning function of water level changes. Summary of the Invention

[0006] The purpose of the present invention is to overcome the deficiencies of the above-mentioned prior art and to provide a method for evaluating the correlation between the water level of a long observation hole and the mining dynamics of a working face.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for evaluating the correlation between the water level of a long-observation borehole and mining dynamics of a working face comprises the following steps:

[0009] Step 1: Before the working face to be evaluated is mined, the operating process and long-hole water level response monitoring data of the surrounding working faces with similar conditions are collected as the basic data set;

[0010] Step 2: Draw the daily variation curve of the water level in the long observation hole of the basic data set. Combined with the operation process of the recovered working face, mark the daily variation curve of the water level in the long observation hole of the basic data set as the strong correlation response curve and the weak correlation response curve of the working face mining to construct the total dataset for deep learning training;

[0011] Step 3: Screen the total deep learning training data set to obtain a deep learning training data set;

[0012] Step 4: Build a feature learning deep neural network model and complete the training using the deep learning training dataset;

[0013] Step 5: During the mining process of the working face to be evaluated, the feature learning deep neural network model is used to perform correlation classification on the daily variation curves of the water level in the surrounding long observation holes;

[0014] Step 6: Based on the correlation classification results, the correlation between the water level in the long observation hole and the mining activity of the working face to be evaluated is evaluated in stages.

[0015] Preferably, in step 1, the mined working face with similar surrounding conditions is a working face that is located in the same coal mine or the same mining area as the working face to be evaluated, has the same mined coal seam and the same pressurized aquifer, and has been safely mined; the operation process refers to the daily production and maintenance shift times during the mining period of the mined working face, and the operation process of the working face to be evaluated is consistent with the operation process of the mined working face.

[0016] Preferably, in step 2, the method for marking the daily variation curve of the long observation hole water level of the basic data set as the strong correlation response curve and the weak correlation response curve of the working face mining is as follows: during the process of coal mining by the single-day production shift and suspension of mining by the maintenance shift, when the daily variation curve of the long observation hole water level of the confined aquifer conforms to the single-day fluctuation recovery change law, the daily variation curve is marked as the strong correlation response curve of the working face mining; otherwise, it is marked as the weak correlation response curve of the working face mining;

[0017] The daily fluctuation and recovery pattern is as follows: the water level in the underground long observation hole drops when the production team is mining, and rises when the maintenance team stops mining.

[0018] Preferably, in step 3, in the deep learning training data set, the proportion of working face mining-related response curve data is 70%, and the proportion of working face mining-related response curve data is 30%.

[0019] Optimized, in step 4, the feature learning deep neural network model structure includes an input layer, a convolutional layer A, a pooling layer A, a convolutional layer B, a pooling layer B, and a fully connected layer;

[0020] The input end of the input layer is used to input the long-term observation aperture daily variation curve graph, the output end of the input layer is communicatively connected to the input end of the convolution layer A, the output end of the convolution layer A is communicatively connected to the input end of the pooling layer A, the output end of the pooling layer A is communicatively connected to the input end of the convolution layer B, the output end of the convolution layer B is communicatively connected to the input end of the pooling layer B, and the output end of the pooling layer B is communicatively connected to the input end of the fully connected layer;

[0021] The fully connected layer is used to output two labels, namely strong correlation and weak correlation.

[0022] Preferably, the long aperture daily variation curve graph input in the input layer is 224×224 pixels;

[0023] The convolution layer A is configured with 16 convolution kernels of size 5×5 and a stride of 1;

[0024] The pooling window size in the pooling layer A is 2×2, and the step size is 2;

[0025] The convolution layer B is configured with 32 convolution kernels of size 3×3 and a stride of 1;

[0026] The pooling window size in the pooling layer B is 1×1 and the step size is 1.

[0027] Preferably, in step 4, when the deep learning training data set is used to complete the training, the deep learning training data set is divided into a training set and a validation set in proportion;

[0028] The maximum number of training times is 500;

[0029] Conditions for achieving model training: the training set fitting error is less than 5%, the validation set fitting error is less than 5%, and 20% of the data in the deep learning training data set is randomly selected as the test set, and the test set fitting error is less than 5%.

[0030] Preferably, in step 5, after the mining of the working face to be evaluated begins, the feature learning deep neural network model in step 4 is used to perform correlation classification on the daily water level curves of the surrounding long observation holes every day, and the classification results are marked as water level fluctuations strongly correlated with working face mining or water level fluctuations weakly correlated with working face mining.

[0031] Preferably, step 6 includes the following sub-steps:

[0032] Step 61, determining the mining distance value of a stage when evaluating the working face to be evaluated in stages according to the variation characteristics of the coal seam overburden;

[0033] Step 62, after completing the jth (j≥1) stage of mining, the correlation between the water level of the long observation hole and the mining dynamics during the jth stage of mining of the working face to be evaluated is evaluated in stages, using the following method:

[0034] For each long observation hole, the number of days in which the water level fluctuation has a strong correlation with the mining movement of the working face during the mining process in the jth stage is counted. and the number of days when water level fluctuations are weakly correlated with mining at the working face ;

[0035] For each long observation hole, the proportion of days with strong correlation between water level fluctuation and working face mining in the normal mining days of the working face in the jth stage is counted. ;

[0036] ×100% (1)

[0037] The proportion of days with strong correlation between water level fluctuation and working face mining in the number of days of mining in the jth stage of the working face The correlation between the water level in the long observation hole and the mining activity during the j-th stage of the working face to be evaluated is divided into strong and weak.

[0038] Preferably, in step 62, the proportion ≥70% of the long observation holes are strongly correlated with the mining of the working face to be evaluated and are defined as level I correlation holes; 40%≤ <70% of the long observation holes have a medium correlation with the mining of the working face to be evaluated, and are defined as level II correlation holes; Long observation holes with a correlation rate of less than 40% are weakly correlated with the mining of the working face to be evaluated and are defined as level III correlation holes.

[0039] The beneficial effects of the present invention are:

[0040] The present invention is based on the water level monitoring data of the long observation hole of the pressurized aquifer during the mining process of the working face. By evaluating the correlation between the water level fluctuation of the long observation hole and the mining of the working face to be evaluated, the long observation hole of the pressurized aquifer with better early warning applicability can be selected. In the subsequent mining process of the working face, it can provide a more accurate early warning of water inrush disasters. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The drawings in the specification, which constitute a part of this application, are used to provide further understanding of this application. The illustrative embodiments of this application and their descriptions are used to explain this application and do not constitute improper limitations on this application.

[0042] Figure 1 It is a flow chart of the method for evaluating the correlation between the water level of a long observation hole and mining dynamics of a working face according to the present invention;

[0043] Figure 2 This is the strong correlation response curve between the water level of the long observation hole of the confined aquifer and the mining operation of the working face in the present invention;

[0044] Figure 3 This is the weak correlation response curve between the water level of the long observation hole of the confined aquifer and the mining operation of the working face in the present invention;

[0045] Figure 4 It is the stage evaluation result of the correlation between the water level of the long observation hole and the mining dynamics of the working face in the embodiment. DETAILED DESCRIPTION

[0046] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present application belongs.

[0047] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0048] In the present invention, the directions or positional relationships indicated by terms such as "upper", "lower", "bottom", and "top" are based on the directions or positional relationships shown in the accompanying drawings. They are relational words determined only for the convenience of describing the structural relationships of the various parts or elements of the present invention. They do not specifically refer to any part or element in the present invention and cannot be understood as limitations on the present invention.

[0049] In the present invention, terms such as "connected" and "connection" should be interpreted broadly to mean a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediary. Relevant researchers or technicians in this field may determine the specific meanings of these terms in the present invention based on specific circumstances, and they should not be construed as limitations on the present invention.

[0050] The present invention will be further described below with reference to the accompanying drawings and examples.

[0051] like Figure 1 As shown, a method for evaluating the correlation between the water level of a long-observation hole and mining dynamics of a working face includes the following steps:

[0052] Step 1: Before the working face to be evaluated is mined, the operating process and long-hole water level response monitoring data of the surrounding working faces with similar conditions are collected as the basic data set;

[0053] Step 2: Draw the daily variation curve of the water level in the long observation hole of the basic data set. Combined with the operation process of the recovered working face, mark the daily variation curve of the water level in the long observation hole of the basic data set into two categories: strong correlation response curve and weak correlation response curve of the working face mining, and construct the total dataset for deep learning training;

[0054] Step 3: Screen the total deep learning training data set to obtain a deep learning training data set;

[0055] Step 4: Build a feature learning deep neural network model and complete the training using the deep learning training dataset;

[0056] Step 5: During the mining process of the working face to be evaluated, the feature learning deep neural network model is used to perform correlation classification on the daily variation curves of the water level in the surrounding long observation holes;

[0057] Step 6: Based on the correlation classification results, the correlation between the water level in the long observation hole and the mining activity of the working face to be evaluated is evaluated in stages.

[0058] Preferably, in step 1, the mined working face with similar surrounding conditions is a working face that is located in the same coal mine or the same mining area as the working face to be evaluated, has the same mined coal seam and the same pressurized aquifer, and has been safely mined; the operation process refers to the daily production and maintenance shift times during the mining period of the mined working face, and the operation process of the working face to be evaluated is consistent with the operation process of the mined working face.

[0059] Preferably, in step 2, the method for marking the daily variation curve of the long observation hole water level of the basic data set as the strong correlation response curve and the weak correlation response curve of the working face mining is as follows: during the process of coal mining by the single-day production shift and suspension of mining by the maintenance shift, when the daily variation curve of the long observation hole water level of the confined aquifer conforms to the single-day fluctuation recovery change law, the daily variation curve is marked as the strong correlation response curve of the working face mining; otherwise, it is marked as the weak correlation response curve of the working face mining;

[0060] The daily fluctuation and recovery pattern is as follows: the water level in the underground long observation hole drops when the production team is mining, and rises when the maintenance team stops mining.

[0061] The response curve of the strong correlation between the water level of the long observation hole of the confined aquifer and the mining operation of the working face is as follows: Figure 2 As shown in the figure, the water level of the long observation hole in the confined aquifer is weakly correlated with the mining operation of the working face. Figure 3 shown.

[0062] Preferably, in step 3, the deep learning training data set is screened to make it easier for feature training. In the deep learning training data set, the proportion of working face mining-strongly correlated response curve data is 70%, and the proportion of working face mining-weakly correlated response curve data is 30%.

[0063] Optimized, in step 4, the feature learning deep neural network model structure includes an input layer, a convolutional layer A, a pooling layer A, a convolutional layer B, a pooling layer B, and a fully connected layer;

[0064] The input end of the input layer is used to input the long-term observation aperture daily variation curve graph, the output end of the input layer is communicatively connected to the input end of the convolution layer A, the output end of the convolution layer A is communicatively connected to the input end of the pooling layer A, the output end of the pooling layer A is communicatively connected to the input end of the convolution layer B, the output end of the convolution layer B is communicatively connected to the input end of the pooling layer B, and the output end of the pooling layer B is communicatively connected to the input end of the fully connected layer;

[0065] The fully connected layer is used to output two labels, namely strong correlation and weak correlation.

[0066] Preferably, the long aperture daily variation curve graph input in the input layer is 224×224 pixels;

[0067] The convolution layer A is configured with 16 convolution kernels of size 5×5 and a stride of 1;

[0068] The pooling window size in the pooling layer A is 2×2, and the step size is 2, that is, it moves 2 pixels each time, thereby reducing the dimension of the feature map;

[0069] The convolution layer B is configured with 32 convolution kernels of size 3×3 and a stride of 1;

[0070] The pooling window size in the pooling layer B is 1×1 and the step size is 1.

[0071] Preferably, in step 4, when the deep learning training data set is used to complete the task, the deep learning training data set is divided into a training set and a validation set in proportion; wherein the ratio of the number of data in the training set to the number of data in the validation set is 4:1.

[0072] The maximum number of training times is 500;

[0073] Conditions for achieving model training: the training set fitting error is less than 5%, the validation set fitting error is less than 5%, and 20% of the data in the deep learning training data set is randomly selected as the test set, and the test set fitting error is less than 5%.

[0074] Regarding the fitting error, the long-viewing hole daily variation curve graph in the training set, validation set, or test set is input into the feature learning deep neural network model, and the fitted label result is output. The number N of the fitted label results in the corresponding data set that are inconsistent with the long-viewing hole daily variation curve graph in the data set according to the situation marked in step 2 is counted. The fitting error is the proportion of the number N to the number of data in the corresponding data set. For example, the long-viewing hole daily variation curve graphs in the training set are input into the feature learning deep neural network model, and the fitted label result is output. The number N1 of the fitted label results in the training set that are inconsistent with the long-viewing hole daily variation curve graph in the training set according to the situation marked in step 2 is counted. The training set fitting error is the proportion of the number N1 to the number of data in the training set.

[0075] The inconsistency between the fitted label result and the long-view hole daily variation curve chart marked according to step 2 means: the fitted label result is strongly correlated, and the long-view hole daily variation curve chart marked according to step 2 is a weakly correlated response curve of the working face mining; or the fitted label result is weakly correlated, and the long-view hole daily variation curve chart marked according to step 2 is a strongly correlated response curve of the working face mining.

[0076] Preferably, in step 5, after the mining of the working face to be evaluated begins, the feature learning deep neural network model in step 4 is used to perform correlation classification on the daily water level curves of the surrounding long observation holes every day, and the classification results are marked as water level fluctuations strongly correlated with working face mining or water level fluctuations weakly correlated with working face mining.

[0077] Preferably, step 6 includes the following sub-steps:

[0078] Step 61, determining the mining distance value of a stage when evaluating the working face to be evaluated in stages according to the variation characteristics of the coal seam overburden, wherein the mining distance value may be 100 meters;

[0079] Step 62, after completing the jth (j≥1) stage of mining, the correlation between the water level of the long observation hole and the mining dynamics during the jth stage of mining of the working face to be evaluated is evaluated in stages, using the following method:

[0080] For each long observation hole, the number of days in which the water level fluctuation has a strong correlation with the mining movement of the working face during the mining process in the jth stage is counted. and the number of days when water level fluctuations are weakly correlated with mining at the working face ;

[0081] For each long observation hole, the proportion of days with strong correlation between water level fluctuation and working face mining in the normal mining days of the working face in the jth stage is counted. ;

[0082] ×100% (1)

[0083] The proportion of days with strong correlation between water level fluctuation and working face mining in the number of days of mining in the jth stage of the working face The correlation between the water level in the long observation hole and the mining activity during the j-th stage of the working face to be evaluated is divided into strong and weak.

[0084] Preferably, in step 62, the proportion ≥70% of the long observation holes are strongly correlated with the mining of the working face to be evaluated and are defined as level I correlation holes; 40%≤ <70% of the long observation holes have a medium correlation with the mining of the working face to be evaluated, and are defined as level II correlation holes; Long observation holes with a correlation rate of less than 40% are weakly correlated with the mining of the working face to be evaluated and are defined as level III correlation holes.

[0085] Example:

[0086] Taking the 01 working face mined under the "four-containing" confined aquifer in a coal mine as an example, there are four "four-containing" water level long observation holes around the 01 working face, namely Guan 1, Guan 2, Guan 3 and Guan 4. The working face operation cycle during the mining period of the 01 working face is: the maintenance shift is from 8:00 to 15:00 every day, and the production shift is from 15:00 to 8:00 the next day. The designed mining length is 900 meters.

[0087] Before the 01 working face was mined, daily water level curves from four long observation holes ("four-containment" holes) were collected from similar working faces in the region. 560 curves with strong correlations between water levels in these long observation holes and mining activity in the mined working face, along with 240 curves with weak correlations, were selected to construct a deep learning training dataset. Based on this deep learning training dataset, the feature learning deep neural network model of the present invention was trained.

[0088] During the mining phase of the 01 working face, the water levels in the four long observation holes (Guan 1, Guan 2, Guan 3, and Guan 4) surrounding the working face were monitored and daily water level curves were plotted. A feature-based deep neural network model was used to classify the correlation between water level fluctuations in the four long observation holes and working face mining as either strongly or weakly correlated.

[0089] During the mining process of the 01 working face, the correlation between the water level of the long observation hole and the mining movement of the working face was evaluated in stages with 100 meters as a stage. The evaluation results are shown in Figure 4Among them, during the first 100 meters of mining at the 01 working face, the strong correlation ratios for Guan 1, Guan 2, Guan 3, and Guan 4 were 68%, 10%, 52%, and 55%, respectively. According to the correlation evaluation method of the present invention, Guan 1, Guan 3, and Guan 4 long observation holes are level II correlation holes, and their water level fluctuations can play a good role in water inrush warning. Guan 2 long observation hole is a level III correlation hole, and its role in water inrush warning at the 01 working face is relatively poor. With the continued mining of the 01 working face, the correlation between the water levels of Guan 1, Guan 3, and Guan 4 long observation holes and the mining of the 01 working face shows a relatively coherent dynamic adjustment. As shown in Table 1, the Guan 2 long observation hole has always had a low correlation with the mining of the 01 working face. The overburden tends to be stable during the mining completion stage, and the daily variation curves of all long observation holes no longer reflect the mining characteristics.

[0090] Table 1 Statistics of periodic evaluation results of correlation

[0091]

[0092] In the actual operation water inrush warning plan of the 01 working face, the daily rapid drop value of the water level in the "four-contained" long observation holes is used as an important warning parameter. The current mining stage refers to the mining correlation evaluation and grading of the observation holes 1, 2, 3, and 4 in the previous stage, and defines different warning weights, which greatly reduces the false alarm rate and realizes the safe and efficient mining of the 01 working face.

[0093] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not a limitation of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A method for evaluating the correlation between the water level of a long-range borehole and mining dynamics at a working face, characterized in that: The following steps are involved: Step 1: Before the working face to be evaluated is mined, the operating process and long-hole water level response monitoring data of the surrounding working faces with similar conditions are collected as the basic data set; Step 2: Draw the daily variation curve of the water level in the long observation hole of the basic data set. Combined with the operation process of the recovered working face, mark the daily variation curve of the water level in the long observation hole of the basic data set as the strong correlation response curve and the weak correlation response curve of the working face mining to construct the total dataset for deep learning training; Step 3: Screen the total deep learning training data set to obtain a deep learning training data set; Step 4: Build a feature learning deep neural network model and complete the training using the deep learning training dataset; Step 5: During the mining process of the working face to be evaluated, the feature learning deep neural network model is used to perform correlation classification on the daily variation curves of the water level in the surrounding long observation holes; Step 6: Based on the correlation classification results, the correlation between the water level of the long observation hole and the mining operation of the working face to be evaluated is evaluated in stages; In step 1, the mined working face with similar surrounding conditions is a working face that is located in the same coal mine or within the same mining area as the working face to be evaluated, has the same mined coal seam and the same confined aquifer, and has been safely mined; the operating process refers to the daily production and maintenance shift times of the mined working face during the mining period, and the operating process of the working face to be evaluated is consistent with the operating process of the mined working face; In step 2, the method for marking the daily variation curve of the long observation hole water level in the basic data set as the strong correlation response curve and the weak correlation response curve of the working face mining is as follows: during the process of coal mining by the single-day production shift and suspension of mining by the maintenance shift, when the daily variation curve of the long observation hole water level in the confined aquifer conforms to the single-day fluctuation and recovery variation law, the daily variation curve is marked as the strong correlation response curve of the working face mining; otherwise, it is marked as the weak correlation response curve of the working face mining; The daily fluctuation and recovery pattern is as follows: the water level in the underground long observation hole drops when the production team mines coal, and rises when the maintenance team stops mining; Step 6 includes the following sub-steps: Step 61, determining the mining distance value of a stage when evaluating the working face to be evaluated in stages according to the variation characteristics of the coal seam overburden; Step 62, after completing the j-th stage of mining, evaluate the correlation between the water level of the long observation hole and the mining dynamics during the j-th stage of mining in the working face to be evaluated, where j ≥ 1, by the following method: For each long observation hole, the number of days in which the water level fluctuation has a strong correlation with the mining movement of the working face during the mining process in the jth stage is counted. and the number of days when water level fluctuations are weakly correlated with mining at the working face ; For each long observation hole, the proportion of days with strong correlation between water level fluctuation and working face mining in the normal mining days of the working face in the jth stage is counted. ; （1） The proportion of days with strong correlation between water level fluctuation and working face mining in the number of days of mining in the jth stage of the working face Divide the correlation between the water level of the long observation hole and the mining activity during the j-th stage of the working face to be evaluated; In step 62, the proportion ≥70% of the long observation holes are strongly correlated with the mining of the working face to be evaluated and are defined as level I correlation holes; 40%≤ <70% of the long observation holes have a medium correlation with the mining of the working face to be evaluated, and are defined as level II correlation holes; Long observation holes with a correlation rate of less than 40% are weakly correlated with the mining of the working face to be evaluated and are defined as level III correlation holes.

2. The method for evaluating the correlation between the water level of a long observation hole and mining dynamics of a working face according to claim 1, characterized in that: In step 3, in the deep learning training data set, the proportion of working face mining-related response curve data is 70%, and the proportion of working face mining-related response curve data is 30%.

3. The method for evaluating the correlation between the water level of a long observation hole and mining dynamics of a working face according to claim 1, characterized in that: In step 4, the feature learning deep neural network model structure includes an input layer, a convolutional layer A, a pooling layer A, a convolutional layer B, a pooling layer B, and a fully connected layer; The input end of the input layer is used to input the long-term observation aperture daily variation curve graph, the output end of the input layer is communicatively connected to the input end of the convolution layer A, the output end of the convolution layer A is communicatively connected to the input end of the pooling layer A, the output end of the pooling layer A is communicatively connected to the input end of the convolution layer B, the output end of the convolution layer B is communicatively connected to the input end of the pooling layer B, and the output end of the pooling layer B is communicatively connected to the input end of the fully connected layer; The fully connected layer is used to output two labels, namely strong correlation and weak correlation.

4. The method for evaluating the correlation between the water level of a long observation hole and mining dynamics of a working face according to claim 3, characterized in that: The long aperture daily variation curve image input in the input layer is 224×224 pixels; The convolution layer A is configured with 16 convolution kernels of size 5×5 and a stride of 1; The pooling window size in the pooling layer A is 2×2, and the step size is 2; The convolution layer B is configured with 32 convolution kernels of size 3×3 and a stride of 1; The pooling window size in the pooling layer B is 1×1 and the step size is 1.

5. The method for evaluating the correlation between the water level of a long observation hole and mining dynamics of a working face according to claim 4, characterized in that: In step 4, when the deep learning training data set is used to complete the training, the deep learning training data set is divided into a training set and a validation set in proportion; The maximum number of training times is 500; Conditions for achieving model training: the training set fitting error is less than 5%, the validation set fitting error is less than 5%, and 20% of the data in the deep learning training data set is randomly selected as the test set, and the test set fitting error is less than 5%.

6. The method for evaluating the correlation between the water level of a long observation hole and mining dynamics of a working face according to claim 1, characterized in that: In step 5, after the mining of the working face to be evaluated begins, the feature learning deep neural network model in step 4 is used to perform correlation classification on the daily water level variation curves of the surrounding long observation holes every day, and the classification results are marked as strong correlation between water level fluctuations and working face mining or weak correlation between water level fluctuations and working face mining.

Images