Fully mechanized coal mining face support pressure intelligent prediction method and system

Abstract

The invention discloses a fully mechanized coal mining face support pressure intelligent prediction method and system, and the method comprises the steps: enabling pressure data to correspond to a coal mining cycle through building a support pressure space coordinate system, and carrying out the space positioning. According to pressure data characteristics, the pressure sensor is divided into a constant resistance type, a slow resistance increasing type, a rapid resistance increasing type, a resistance reducing type and a composite type. And typical pressure values (such as initial supporting force and end resistance) of each work period are screened, and prediction is carried out by using a fitting function or a machine learning algorithm. And establishing a pressure grade classification standard, dividing prediction data into three grades of weak pressure, medium pressure and strong pressure, and displaying the motion state of the top plate and the pressure grade in real time. The system comprises a data processing module, a classification module, a prediction module and a display module, data extraction, classification, prediction and real-time display are achieved, the accuracy and real-time performance of roof pressure prediction are improved, and safe production of a fully mechanized coal mining face is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of coal mine working face safety management technology, and particularly relates to an intelligent prediction method and system for support pressure in fully mechanized mining faces. Background Technology

[0002] With the advancement of intelligent equipment in coal mining, monitoring equipment for various conditions at coal faces has become increasingly sophisticated. Monitoring methods for roof subsidence, micro-seismic events within the rock strata, and support pressure have become more diversified, facilitating production managers' assessment of the face's safety status. Safety management of fully mechanized mining faces is crucial during coal mining, requiring managers to promptly grasp the roof pressure status. In recent years, numerous roof pressure prediction software programs based on hydraulic support pressure data from fully mechanized mining faces have emerged for intelligent face management.

[0003] Currently, most software for predicting support pressure in longwall mining faces relies on extracting patterns from large amounts of support pressure data for prediction. While support pressure is typically recorded at time intervals during coal mining, the pressure can exhibit various irregularities due to factors such as support movement and the lithology of the working face roof. This reduces the accuracy of pressure prediction based on support pressure patterns, and the real-time prediction and display functions for future pressure during mining also need optimization. Therefore, this paper proposes an intelligent prediction method and system for support pressure in fully mechanized mining faces. This method addresses the problem of filtering effective support pressure data during roof pressure prediction, saves computing power, improves the accuracy of real-time roof pressure early warning, and enables real-time display of roof pressure prediction and movement status, effectively guiding safe production at the working face. Summary of the Invention

[0004] This invention proposes an intelligent prediction method and system for support pressure in fully mechanized mining faces to solve the problems existing in the prior art.

[0005] To achieve the above objectives, this invention provides an intelligent prediction method for support pressure in fully mechanized mining faces, comprising the following steps:

[0006] Establish a spatial coordinate system for support pressure data, correlate support pressure data with each coal mining cycle, and spatially locate the support pressure data of the fully mechanized mining face.

[0007] Stent pressure categories are classified based on the characteristics of pressure data during the stent's working cycle.

[0008] The pressure category of the support is correlated with the movement state of the roof in the mining area, and typical pressure values ​​of a single support in each working cycle are selected.

[0009] Based on the typical pressure values, obtain the typical pressure data pattern characteristics, and perform data prediction based on the typical pressure data pattern characteristics.

[0010] The pressure levels are classified based on the predicted data, and the predicted pressure data categories are mapped to the corresponding pressure levels to output the prediction results.

[0011] Optionally, the establishment of the spatial coordinate system for stent pressure data includes:

[0012] The dip length of the working face is recorded as the x-axis. The x-axis is divided equally according to the number of working face supports to determine the position of each support on the x-axis.

[0013] The working face length coordinates are marked as the y-axis, and the position of the support pressure data on the y-axis is determined by the coal mining cycle advance.

[0014] Optionally, the working cycle of the support is divided into:

[0015] Monitor the changing trend of the working resistance value of the stent, and determine the start and end of the cycle based on the descent threshold, rise threshold, stent descent threshold, stent elevation threshold and time threshold;

[0016] If the cycle time span does not reach the time threshold, it is judged as an abnormal cycle and is merged or corrected with the adjacent cycle.

[0017] Optionally, the pressure type of the support includes constant resistance type, slow resistance increase type, rapid resistance increase type, resistance decrease type, and composite type.

[0018] Optionally, the movement state of the mining roof includes a stable state, a bending and subsidence state, and a fracture and rotation state;

[0019] Among them, the constant resistance type corresponds to the stable state, the slowly increasing resistance type corresponds to the bending and sinking state, and the rapidly increasing resistance type corresponds to the fracture and rotation state.

[0020] Optionally, the typical pressure values ​​for each working cycle of a single support are selected, including:

[0021] Extract the initial support force and final resistance values ​​within the working cycle of a single support;

[0022] Initial support force is selected using the time-rate of change method, and final resistance force is selected using the time-maximum-minimum method.

[0023] Optionally, the data prediction based on the regular characteristics of typical pressure data includes:

[0024] The initial support force and final resistance data are fitted using a fitting function, and predictions are made based on the fitting function.

[0025] By using deep learning algorithms and training a model with historical pressure data, future changes in mine pressure at the working face can be predicted.

[0026] Optionally, the pressure levels include weak pressure, medium pressure, and strong pressure; wherein, weak pressure corresponds to a stable state, medium pressure corresponds to a bending and sinking state, and strong pressure corresponds to a fracture and rotation state.

[0027] This invention also proposes a real-time early warning system for roof pressure in fully mechanized mining faces, comprising:

[0028] The data processing module is used to extract and store stent pressure data, classify it by stent number, and divide the work cycle.

[0029] The data classification module is used to classify the support pressure data and correlate it with the top plate movement status;

[0030] The data prediction module is used to filter typical pressure values, perform data prediction, and classify pressure levels.

[0031] The data display module is used to display the support pressure category, top plate movement status, and pressure level in real time.

[0032] Compared with the prior art, the present invention has the following advantages and technical effects:

[0033] This invention establishes a spatial coordinate system for support pressure, achieving precise correspondence and spatial positioning between pressure data and the coal mining cycle. By extracting typical pressure values ​​based on data screening principles and using fitting functions or machine learning algorithms for prediction, it significantly improves pressure prediction accuracy. Simultaneously, it optimizes computing resources, reduces computational complexity, and enhances system operating efficiency. Furthermore, this invention constructs a pressure level classification standard, displaying the roof movement status and pressure level in real time, providing intuitive data for production management. Its dynamic monitoring and real-time early warning functions can promptly detect and warn of abnormal roof pressure, effectively preventing accidents and ensuring safe production in fully mechanized mining faces. This invention has a wide range of applications and strong adaptability, allowing for flexible adjustments to meet the production conditions of different working faces, providing strong support for coal mine safety management. Attached Figure Description

[0034] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0035] Figure 1 This is a flowchart of a method according to an embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of a constant resistance type according to an embodiment of the present invention;

[0037] Figure 3 This is a schematic diagram of a slow resistance-increasing type according to an embodiment of the present invention;

[0038] Figure 4 This is a schematic diagram of a fast resistance-increasing type according to an embodiment of the present invention;

[0039] Figure 5 This is a schematic diagram of a drag-reducing type according to an embodiment of the present invention;

[0040] Figure 6 This is a schematic diagram of a composite embodiment of the present invention;

[0041] Figure 7 This is an unfolded diagram of the LSTM neural network according to an embodiment of the present invention;

[0042] Figure 8 This is a system structure diagram of an embodiment of the present invention. Detailed Implementation

[0043] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0044] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0045] Example 1:

[0046] like Figure 1 As shown, this embodiment provides a method for intelligent prediction of support pressure in a fully mechanized mining face, including the following steps:

[0047] Establish a spatial coordinate system for support pressure data, correlate support pressure data with each coal mining cycle, and spatially locate the support pressure data of the fully mechanized mining face.

[0048] Stent pressure categories are classified based on the characteristics of pressure data during the stent's working cycle.

[0049] The pressure category of the support is correlated with the movement state of the roof in the mining area, and typical pressure values ​​of a single support in each working cycle are selected.

[0050] Based on the typical pressure values, obtain the typical pressure data pattern characteristics, and perform data prediction based on the typical pressure data pattern characteristics.

[0051] The pressure levels are classified based on the predicted data, and the predicted pressure data categories are mapped to the corresponding pressure levels to output the prediction results.

[0052] Furthermore, establishing a spatial coordinate system for stent pressure data includes:

[0053] The dip length of the working face is recorded as the x-axis. The x-axis is divided equally according to the number of working face supports to determine the position of each support on the x-axis.

[0054] The working face length coordinates are marked as the y-axis, and the position of the support pressure data on the y-axis is determined by the coal mining cycle advance.

[0055] This embodiment also includes: dividing the working surface inclination length equally according to the number of working surface supports, and determining the position of each support on the x-axis.

[0056] Extract the timing record pressure data of a single hydraulic support and match the hydraulic support data of the working face with the support number in sequence.

[0057] The computer divides the working cycle of a single support and corresponds it to the coal mining cycle advance in sequence to determine the y-axis position of the support pressure data.

[0058] As a further implementation method, the working cycle of a single stent is divided as follows:

[0059] The slope of the curve alternates between positive and negative at the end of one cycle (decompression and lowering stage) and the beginning of the next cycle (initial pressurization stage), and the working resistance value of the support is generally very low at the boundary between two cycles. Based on the characteristics of the support's working cycle, a computer-automated method for dividing the support's working cycle is proposed. The specific implementation steps are as follows:

[0060] The first step is to process the original mine pressure data into a Datetime format, and then create left and right column timestamp data lists respectively, and calculate and record the time difference in a loop.

[0061] The second step is to set five cycle judgment thresholds: descent threshold, ascent threshold, lowering threshold, raising threshold, and time threshold (which can be adjusted according to different production conditions on different work surfaces). The descent threshold is used to determine whether the support is currently undergoing a lowering operation or whether the support's working resistance is decreasing due to the support's operation. The ascent threshold is used to determine whether the support is currently undergoing a raising operation or whether the support's working resistance is increasing due to the support's operation. The lowering threshold is used to determine whether the support has completed the lowering operation. The raising threshold is used to determine whether the support has started the raising operation. The time threshold is used to determine whether the cycle is a normal working cycle.

[0062] The third step is to monitor the stent's working resistance value to reach the stent lowering threshold, and when the trend of change changes to increase after cycling to the next timestamp data, record the previous timestamp data as the end of the previous cycle, and record this time point as the beginning of the next cycle.

[0063] The fourth step is to determine if the time span of a cycle does not meet the time threshold standard when it is detected that the cycle is an abnormal cycle. Record the cycle and continue to calculate the next cycle. If the combined time of two consecutive cycles is close to the time threshold, the two cycles are merged into one cycle. The start time of the first cycle is recorded as the start time of the entire cycle, and the end time of the second cycle is recorded as the end time of the entire cycle. If the combined time span of the two cycles is still less than the time threshold, record the cycle and calculate the next cycle. When the combined time span of several cycles meets the time threshold standard, integrate all the recorded cycles, set the start time of the first cycle as the start time of the entire cycle, and set the end time of the last cycle as the end time of the entire cycle.

[0064] Furthermore, when dividing the working cycle of the support into the coal cutting cycle, if the relationship between the working cycle of the support and the coal cutting cycle is uncertain, it can be determined by comparing adjacent supports.

[0065] Furthermore, if the adjacent supports have not undergone a shifting operation, but the intermediate support shows a shifting operation in the original data, it is determined that the shifting operation does not actually exist, and the working cycle of the intermediate support is corrected by the working cycle of the adjacent supports.

[0066] Furthermore, if there are long or ultra-long cycles that exceed the time of one coal cutting cycle, the computer needs to separate these ultra-long cycles (without coal cutting operations).

[0067] Furthermore, each coal mining cycle advance corresponds to one support working cycle. The data within the support working cycle is correlated with the coal mining cycle advance, and the position of the pressure data on the y-axis of the mining space within the support working cycle is determined.

[0068] Furthermore, the stent's work cycle is divided into the following categories:

[0069] Monitor the changing trend of the working resistance value of the stent, and determine the start and end of the cycle based on the descent threshold, rise threshold, stent descent threshold, stent elevation threshold and time threshold;

[0070] If the cycle time span does not reach the time threshold, it is judged as an abnormal cycle and is merged or corrected with the adjacent cycle.

[0071] Furthermore, the pressure data types during the stent's working cycle are classified into five types: constant resistance type, slow resistance increase type, rapid resistance increase type, resistance decrease type, and composite type.

[0072] The types of changes in the working resistance of hydraulic supports can generally be classified into five categories: constant resistance, slow resistance increase, rapid resistance increase, resistance decrease, and composite type (the above five types are based on the classification of multiple issues in the "Journal of Coal Science and Technology"). These classification methods have been widely used in roof condition identification and mine pressure prediction research.

[0073] Constant resistance type ( Figure 2 After the support reaches its initial support force, the working resistance of the support does not fluctuate significantly, showing an almost linear shape in the pressure curve. The initial support force and the final resistance value are basically equal, with the fluctuation between the two values ​​not exceeding 5%. The pressure of a constant resistance type support will be less than the rated working resistance of the support.

[0074] Slowly increasing resistance type ( Figure 3 After the working resistance of the support reaches the initial support force, the support pressure shows a slow increasing trend, and the change in support pressure within the cycle is small. The final resistance value is slightly greater than the initial support force value, and the final resistance is 5%-20% greater than the initial support force.

[0075] Fast resistance increase type ( Figure 4 The working resistance of the support continues to rise after reaching the initial support force, and the growth rate is relatively large. The working resistance value of the support changes significantly within the working cycle. The final resistance value is significantly greater than the initial support force value, and the final resistance is more than 25% greater than the initial support force.

[0076] Resistance reduction type ( Figure 5 ): After the working resistance of the support reaches the initial support force, the pressure begins to drop, which is judged as an abnormal state.

[0077] Composite ( Figure 6 The stent exhibits a mixture of various conditions within a single cycle, and after maintaining a relatively stable state for a period of time, the working cycle ends.

[0078] Furthermore, the motion state of the roof plate is divided into three types: stable state, bending and sinking state, and fracture and rotation state.

[0079] The pressure data characteristics during the working cycle of the stent are as follows: constant resistance type corresponds to a stable state, slow resistance increase type corresponds to a bending and sinking state, and rapid resistance increase type corresponds to a fracture and rotation state.

[0080] A method for extracting support pressure was established to extract the initial support force and final resistance of a single coal mining cycle from massive support pressure data.

[0081] There are two initial support processes for the support system: single initial support and multiple initial support.

[0082] Initial support refers to a single continuous pressurization of the support until it contacts the top plate and forms a relatively balanced state. The duration is relatively short, the pressure change trend is almost vertical with a large slope, and the pressure change slows down instantly after reaching the initial support force, the slope decreases, and then the working resistance changes slowly under the interaction with the top plate.

[0083] Multiple initial supports refer to a brief pause in pressurization during the initial pressurization phase of the support, resulting in a trapezoidal upward trend in the pressure change curve. After reaching the final initial support force, the working resistance changes slowly due to the interaction with the top plate.

[0084] Furthermore, typical pressure values ​​for a single stent during each working cycle were selected, including:

[0085] Extract the initial support force and final resistance values ​​within the working cycle of a single support;

[0086] Initial support force is selected using the time-rate of change method, and final resistance force is selected using the time-maximum-minimum method.

[0087] Specifically, the first step is to set a time threshold, within which the initial pressurization stage of the stent is defined. The second step is to calculate the pressure-time change rate for every five adjacent pressure values ​​within the time threshold. If the change rate decreases and there is no sharp increase in the change rate within a short period of time (due to the different time intervals of different acquisition methods, the time span of three adjacent pressure value points is taken), then the value at the point where the change rate changes is recorded as the initial support force value. If the change rate increases sharply, it is determined to be multiple initial supports, and the pressure value at the point where the last change rate changes within the time threshold is recorded as the initial support force value.

[0088] The characteristic of stent end-cycle resistance is that it is the maximum resistance value within a certain period at the end of the stent's working cycle, and the stent pressure shows a significant downward trend after the end-cycle resistance is generated. Based on this characteristic, a time-maximum / minimum judgment method is designed to enable automatic computer screening of end-cycle resistance.

[0089] By setting a time threshold, in a normal stent cycle, the maximum value is selected within a certain period before the pressure is released and the pressure recorded on the stent after that time point shows a significant downward trend. The value at that time point is recorded as the end-of-cycle resistance value. In a multi-cycle integrated stent cycle, the maximum value within a certain period before the end of the last cycle is selected as the end-of-cycle resistance value.

[0090] The cycle division and data screening calculations are carried out until the division of all working cycles of the stent and the screening of the initial support force and the resistance at the end of the cycle are completed.

[0091] A partial code snippet is shown below, demonstrating an algorithm for automatically dividing the scaffold's work cycle using Python.

[0092] #Analyze the pressure data of each support

[0093] defanalyse_press(all data,configs):

[0094] allpressure time=[

[0095] allpress index = [] # Get time

[0096] forx in all data:

[0097] iflen(x)! = 0:

[0098] time = x[:,0]

[0099] time=[datetime.datetime.strptime(x,"%Y-%m-%d%H:%M:%5")for x intime]break

[0100] #Get the index for each cut

[0101] for iin range(len(all data)):

[0102] data = all data[i]

[0103] iflen(data) ! = 0:

[0104] forx in data:

[0105] ifx[1]is None: # Prevents the initial mineral pressure value from being 0

[0106] x[1]=25

[0107] data = datal:,1]

[0108] data=np.array([float(x)forx in data])

[0109] for i in range(len(data)):

[0110] if data[i] <= 20:

[0111] data[i] = 20

[0112] sign=0

[0113] index = []

[0114] all index = []

[0115] for i in range(len(data)):

[0116] if data[i] < 20:

[0117] data[i] = 20

[0118] ifsign == 0 and data[i] < 25:

[0119] sign=1

[0120] index.append(i-1)

[0121] elifsign==l and data[i]>25:

[0122] sign=0

[0123] index.append(i)

[0124] all index.append(index)

[0125] index = []

[0126] #Obtain the initial support force and final resistance for each cut

[0127] index_press=[]#init: initial support force

[0128] final press = []#final_press: final resistance

[0129] limit=configs['preprocess']['time_range']

[0130] forx in all indexes:

[0131] ifx[0] <limit:

[0132] final_press.append(np.max(data[x[0]]))

[0133] init_press.append(np.max(data[x[1]:x[1]+limit]))

[0134] else:

[0135] final_press.append(np.max(data[x[0]-limit:x[0]]))

[0136] initpress.append(np.max(data[x[1]:x[1]+limit]))

[0137] Furthermore, data prediction based on the regular characteristics of typical pressure data includes:

[0138] The initial support force and final resistance data are fitted using a fitting function, and predictions are made based on the fitting function.

[0139] By using deep learning algorithms and training a model with historical pressure data, future changes in mine pressure at the working face can be predicted.

[0140] Specifically:

[0141] Method 1: Prediction using fitting functions. Based on the selected initial support force and final resistance data, software such as MATLAB is used for fitting to obtain the fitting functions of the initial support force and final resistance relative to the mining area space. The obtained fitting functions are then used to predict the values ​​of the initial support force and final resistance at future locations.

[0142] Method 2: Deep learning algorithm prediction. By designing and establishing a calculation model that matches historical pressure data and progress, deep learning algorithms are used to discover the changing patterns of mine pressure data from a time series perspective. By analyzing historical pressure trends, the changes in mine pressure in the working face over a future period of time can be predicted.

[0143] Construction of a top plate pressure prediction model based on deep learning;

[0144] The algorithm model is based on the Keras artificial neural network library written in Python, and uses the TensorFlow deep learning framework based on data flow graphs as the backend to build a prediction model.

[0145] The prediction model consists of three stacked LSTM networks, including an input layer, an LSTM hidden layer, a flattened layer, and a fully connected layer. It uses the sigmoid function and the tanh function as activation functions, and the adaptive Adam optimization algorithm as the model optimizer to calculate the loss gradient and the mean squared error as the loss function to compile the model.

[0146] Expanding the above LSTM neural network along the time dimension, as follows: Figure 7As shown in the diagram, during model training, data from each training cycle is input through the input layer, where network units in hidden layer 1 extract data information and selectively pass it to the network units of the next hidden layer, continuing until hidden layer m, and finally output by the output layer. The arrows parallel to the time axis in the diagram represent the transmission of training information from each hidden layer along the time dimension; that is, the hidden layer information trained in the previous time step can be passed along the time axis T for training the hidden layer information in the next time step. This transmission of training information ensures the memory function of the LSTM neural network in the time dimension.

[0147] Establish a support pressure zoning standard to effectively zon the predicted support pressure data of the fully mechanized mining face, and provide production management suggestions for the fully mechanized mining face for different zoning.

[0148] Establish a pressure level classification standard and classify the predicted data into pressure levels.

[0149] Furthermore, the pressure levels are divided into three types: low pressure, medium pressure, and high pressure.

[0150] Based on the predicted data, the initial support force and final resistance of the support structure exhibit the following characteristics: weak pressure corresponds to a stable state, medium pressure corresponds to a bending and sinking state, and strong pressure corresponds to a fracture and rotation state.

[0151] Furthermore, the predicted pressure data categories are mapped to corresponding pressure levels and roof movement states. The prediction results are output and displayed in real time to guide safe production and support management at the working face.

[0152] Example 2:

[0153] like Figure 8 As shown in the figure, this embodiment also proposes a real-time early warning system for roof pressure in fully mechanized mining faces, including:

[0154] The data processing module extracts and stores the pressure data from each support, and then labels and classifies the extracted data according to the support number. Based on the support pressure cycle classification standard described in the first aspect, the cycle is divided and correlated with the coal cutting cycle.

[0155] The data classification module categorizes the pressure data characteristics within the working cycle of the support into the five types described in the first aspect, and, based on the established roof movement state zoning standard, maps the support pressure category to the roof movement state of the mining area.

[0156] The data prediction module, following the established stent pressure data filtering principles, selects typical pressure values ​​for each stent within each working cycle and performs data prediction based on the established data processing methods. According to the established pressure level classification standards, the predicted data is categorized into pressure levels, and the predicted pressure data categories are mapped to the corresponding pressure levels before the prediction results are output.

[0157] The data display module displays the corresponding pressure levels in real time for the aforementioned support pressure categories, the movement status of the mining roof, and the predicted pressure data categories, ensuring the production safety of the fully mechanized mining face.

[0158] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent prediction of support pressure in a fully mechanized mining face, characterized in that, Includes the following steps: Establish a spatial coordinate system for support pressure data, correlate support pressure data with each coal mining cycle, and spatially locate the support pressure data of the fully mechanized mining face. Stent pressure categories are classified based on the characteristics of pressure data during the stent's working cycle. The pressure category of the support is correlated with the movement state of the roof in the mining area, and typical pressure values ​​of a single support in each working cycle are selected. Based on the typical pressure values, obtain the typical pressure data pattern characteristics, and perform data prediction based on the typical pressure data pattern characteristics. The pressure levels are classified based on the predicted data, and the predicted pressure data categories are mapped to the corresponding pressure levels to output the prediction results.

2. The method according to claim 1, characterized in that, The established spatial coordinate system for stent pressure data includes: The dip length of the working face is recorded as the x-axis. The x-axis is divided equally according to the number of working face supports to determine the position of each support on the x-axis. The working face length coordinates are marked as the y-axis, and the position of the support pressure data on the y-axis is determined by the coal mining cycle advance.

3. The method according to claim 1, characterized in that, The working cycle of the stent is divided into: Monitor the changing trend of the working resistance value of the stent, and determine the start and end of the cycle based on the descent threshold, rise threshold, stent descent threshold, stent elevation threshold and time threshold; If the cycle time span does not reach the time threshold, it is judged as an abnormal cycle and is merged or corrected with the adjacent cycle.

4. The method according to claim 1, characterized in that, The pressure categories of the stents include constant resistance type, slow resistance increase type, rapid resistance increase type, resistance decrease type, and composite type.

5. The method according to claim 1, characterized in that, The movement states of the mining roof include stable state, bending and subsidence state, and fracture and rotation state. Among them, the constant resistance type corresponds to the stable state, the slowly increasing resistance type corresponds to the bending and sinking state, and the rapidly increasing resistance type corresponds to the fracture and rotation state.

6. The method according to claim 1, characterized in that, The typical pressure values ​​for each working cycle of a single stent selected include: Extract the initial support force and final resistance values ​​within the working cycle of a single support; Initial support force is selected using the time-rate of change method, and final resistance force is selected using the time-maximum-minimum method.

7. The method according to claim 1, characterized in that, The data prediction based on the regularity and characteristics of typical pressure data includes: The initial support force and final resistance data are fitted using a fitting function, and predictions are made based on the fitting function. By using deep learning algorithms and training a model with historical pressure data, future changes in mine pressure at the working face can be predicted.

8. The method according to claim 1, characterized in that, The pressure levels include weak pressure, medium pressure, and strong pressure; where weak pressure corresponds to a stable state, medium pressure corresponds to a bending and sinking state, and strong pressure corresponds to a fracture and rotation state.

9. A real-time early warning system for roof pressure in a fully mechanized mining face, characterized in that, include: The data processing module is used to extract and store stent pressure data, classify it by stent number, and divide the work cycle. The data classification module is used to classify the support pressure data and correlate it with the top plate movement status; The data prediction module is used to filter typical pressure values, perform data prediction, and classify pressure levels. The data display module is used to display the support pressure category, top plate movement status, and pressure level in real time.

Images