Coal mine geological structure intelligent analysis system based on artificial intelligence

Abstract

The present application provides an artificial intelligence-based intelligent analysis system for coal mine geological structures. Applied to the field of coal mine monitoring technology, the system includes: a sensing device and a monitoring device. The monitoring device is used to receive real-time status data of a coal mine with a folded geological structure sent by the sensing device, and based on the real-time status data collected multiple times, obtain a geological feature matrix, input the geological feature matrix into a pre-acquired settlement prediction model for prediction processing, and obtain a predicted settlement after a preset time period. Based on the predicted settlement and a preset settlement threshold, the safety status of the coal mine can be determined. The above system can be used to predict the settlement of a coal mine with a folded geological structure, improve the timeliness of determining the settlement, and ensure the safety of coal mine production.

Description

Technical Field

[0001] The present application relates to the field of coal mine monitoring technology, and in particular to an artificial intelligence-based intelligent analysis system for coal mine geological structures. Background Art

[0002] Folded geological structures are common in coal mines. They cause coal seams to bend and deform, resulting in uneven inclination and curvature. During mining, these folded geological structures often lead to further deformation and subsidence. Therefore, to ensure the safety and reliability of coal mining, it is necessary to determine the amount of subsidence caused by mining.

[0003] In the related art, for coal mines with folded geological structures, the method for determining the settlement of coal mines mainly relies on professional surveyors using professional measuring equipment to conduct field measurements and obtain measurement data to determine the settlement of the coal mine. However, due to the long measurement time and safety risks, surveyors do not conduct measurements frequently, but instead conduct measurements according to the measurement cycle (for example, once every month) to obtain measurement data. Due to the long measurement cycle, the timeliness of determining the settlement based on the measurement data of each measurement cycle is poor. Summary of the Invention

[0004] The present application provides an artificial intelligence-based intelligent analysis system for coal mine geological structures, which is used to solve the problem of poor timeliness in determining the amount of settlement based on the measurement data of each measurement cycle due to the long measurement cycle.

[0005] In a first aspect, the present application provides an artificial intelligence-based intelligent analysis system for coal mine geological structures, including a sensing device and a monitoring device;

[0006] The monitoring equipment is used to:

[0007] receiving real-time status data of the coal mine sent by the sensing device, and acquiring a geological feature matrix based on the real-time status data collected multiple times, wherein the coal mine has a fold structure geological structure;

[0008] Inputting the geological feature matrix into a pre-acquired settlement prediction model for prediction processing to obtain a predicted settlement after a preset time period, wherein the settlement prediction model is a data set constructed based on pre-acquired historical status data and historical settlement of a coal mine with a folded geological structure, and an algorithm model for predicting coal mine settlement is obtained by optimizing and training a relevance vector machine (RVM) model using a particle swarm optimization (PSO) algorithm;

[0009] The safety status of the coal mine is determined based on the predicted settlement amount and a preset settlement amount threshold.

[0010] In a possible embodiment, the sensing device includes multiple sensors, which are used to collect real-time status data of the coal mine in real time. The real-time status data includes coal seam mining depth, coal seam temperature, coal seam humidity, coal seam inclination, goaf pressure and coal mine vibration frequency.

[0011] In a possible implementation, the sensing device includes a depth sensor, a temperature sensor, a humidity sensor, a coal seam inclination sensor, a pressure sensor, and a vibration sensor;

[0012] The depth sensor is used to measure the mining depth of the coal seam;

[0013] The temperature sensor is used to measure the temperature of the coal seam;

[0014] The humidity sensor is used to measure the humidity of the coal seam;

[0015] The coal seam inclination sensor is used to measure the coal seam inclination;

[0016] The pressure sensor is used to measure the pressure of the goaf;

[0017] The vibration sensor is used to measure the vibration frequency of the coal mine.

[0018] In a possible implementation, the settlement threshold is determined by analyzing historical accident data of coal mines with folded geological structures.

[0019] In a possible implementation, the monitoring device obtains a geological feature matrix based on real-time status data collected multiple times, including:

[0020] Determine the multidimensional geological characteristic vector corresponding to each moment in the real-time state data to obtain multiple geological characteristic vectors;

[0021] The multiple geological feature vectors are combined and processed according to the time sequence of multiple moments to obtain the geological feature matrix.

[0022] In a possible implementation, the system further includes a server configured to:

[0023] Based on the historical status data of at least one coal mine with a folded geological structure and the historical settlement amount of each coal mine uploaded by the sensing device, the data set is constructed, wherein the data set includes multiple groups of samples, each group of samples includes a geological feature matrix obtained by processing the historical status data within a preset time period, and the settlement amount after a preset time period after the preset time period;

[0024] According to the data set, the PSO algorithm is used to optimize the RVM model to obtain the settlement prediction model;

[0025] The settlement prediction model is configured in the monitoring equipment.

[0026] In a possible implementation, the server uses a PSO algorithm to optimize and train the RVM model based on the data set to obtain the settlement prediction model, including:

[0027] Dividing the samples in the data set into training samples and test samples according to a preset ratio;

[0028] Inputting the training samples in the data set into the RVM model, and optimizing the parameters of the RVM model using the PSO algorithm to obtain optimized parameters;

[0029] Configuring the optimization parameters in the RVM model to obtain an initial prediction model;

[0030] Performing prediction calculation on the initial prediction model according to the test samples in the data set to obtain an output result of the initial prediction model;

[0031] If the mean square error value of the output result of the initial prediction model is less than a preset mean square error threshold, the initial prediction model is determined as the settlement prediction model.

[0032] In a possible implementation, the server is further configured to:

[0033] If the mean square error value of the output result of the initial prediction model is greater than or equal to the preset mean square error threshold, the number of iterations corresponding to the PSO algorithm is increased according to the preset rules to obtain a new number of iterations;

[0034] Inputting the training samples in the data set into the RVM model, and optimizing the parameters of the RVM model again using the PSO algorithm according to the new number of iterations to obtain new optimized parameters;

[0035] Configuring the new optimization parameters in the RVM model to obtain a new prediction model;

[0036] Performing prediction calculation on the new prediction model according to the test samples in the data set to obtain an output result of the new prediction model;

[0037] If the mean square error value of the output result of the new prediction model is less than a preset mean square error threshold, the new prediction model is determined as the settlement prediction model.

[0038] In a possible implementation, the server inputs the training samples in the data set into the RVM model, and optimizes the parameters of the RVM model using the PSO algorithm to obtain optimized parameters, including:

[0039] Step a, randomly initializing the particle swarm of the PSO algorithm according to a preset number of particles to obtain an initial particle swarm, where the position of each particle in the initial particle swarm is the kernel width parameter of the Gaussian kernel function in the RVM model;

[0040] Step b, determining the objective function of the RVM model as the fitness function of each particle in the initial particle swarm, wherein the objective function is a mean square error function;

[0041] Step c, configuring the kernel width parameter corresponding to each particle in the initialized particle swarm in the RVM model, performing prediction calculation on the training samples in the data set, and calculating the fitness value corresponding to each particle using the fitness function;

[0042] Step d, updating the position and velocity of each particle in the initial particle swarm according to the fitness value corresponding to each particle, and recalculating the fitness value corresponding to each updated particle, repeating steps c to d until a preset number of iterations is reached to obtain a target particle swarm;

[0043] Step e: determining the kernel width parameter corresponding to the position of the particle with the smallest fitness value in the target particle group as the optimization parameter.

[0044] In a possible implementation, the system further includes: a terminal device;

[0045] The monitoring device is further configured to push an alarm message to the terminal device when the predicted settlement amount is greater than or equal to the settlement amount threshold, wherein the alarm message is configured to indicate that there is a safety risk in the coal mine;

[0046] The terminal device is used to receive the alarm information sent by the monitoring device.

[0047] In a second aspect, an embodiment of the present application provides a computer-readable storage medium, which stores computer execution instructions. When the computer execution instructions are executed by a processor, they are used to implement the artificial intelligence-based coal mine geological structure intelligent analysis system described in any one of the first aspects.

[0048] In a third aspect, an embodiment of the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the artificial intelligence-based coal mine geological structure intelligent analysis system as described in any one of the first aspects.

[0049] The present application provides an artificial intelligence-based intelligent analysis system for coal mine geological structures, including a sensing device and a monitoring device. The monitoring device can receive real-time status data of a coal mine with a folded geological structure, and obtain a geological feature matrix based on the real-time status data collected multiple times. The geological feature matrix can be input into a pre-acquired settlement prediction model for prediction processing to obtain a predicted settlement after a preset time period. The safety of the coal mine can be determined based on the predicted settlement and a preset settlement threshold. Through the above system, the settlement of a coal mine with a folded geological structure can be predicted, thereby improving the timeliness of determining the settlement. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0051] Figure 1 A schematic diagram of an application scenario provided in an embodiment of the present application;

[0052] Figure 2 This is a schematic diagram of the structure of the first embodiment of the artificial intelligence-based coal mine geological structure intelligent analysis system provided in this application;

[0053] Figure 3 A schematic diagram of the structure of a geological feature matrix provided in an embodiment of the present application;

[0054] Figure 4 This is a schematic diagram of the structure of the second embodiment of the artificial intelligence-based coal mine geological structure intelligent analysis system provided in this application;

[0055] Figure 5 A schematic diagram of the process of the first embodiment of the intelligent analysis method for coal mine geological structure based on artificial intelligence provided in this application;

[0056] Figure 6 This is a schematic diagram of the structure of the third embodiment of the artificial intelligence-based coal mine geological structure intelligent analysis system provided in this application;

[0057] Figure 7 A schematic diagram of the flow chart of the second embodiment of the intelligent analysis method for coal mine geological structure based on artificial intelligence provided in this application;

[0058] Figure 8 This is a system schematic diagram of the artificial intelligence-based coal mine geological structure intelligent analysis system provided in the embodiment of the present application.

[0059] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0060] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0061] It should be understood that in the various embodiments of the present application, the size of the serial number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0062] It should be understood that in this application, "include" and "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0063] First, let’s explain the terms involved in this application:

[0064] Coal mine geological structure: may include fold structure, fault structure, monocline structure, collapse column structure, magma intrusion structure, erosion structure, etc.

[0065] Particle Swarm Optimization (PSO): A swarm intelligence-based optimization algorithm used to solve optimization problems. This algorithm simulates the communication and cooperation mechanisms found in the social behavior of bird flocks, finding the optimal solution by simulating the movement of particles in the solution space and the exchange of information.

[0066] Relevance Vector Machine (RVM): is a machine learning method for supervised learning, which is a sparse probabilistic model.

[0067] The following specific embodiments are used to describe the technical solution of the present application in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0068] For ease of understanding, below, combined with Figure 1 , describes the application scenarios to which the embodiments of the present application are applicable.

[0069] Figure 1 This is a schematic diagram of the application scenario provided by the embodiment of this application. Figure 1 , including sensing devices and monitoring devices, which can communicate with each other through the fourth generation mobile communication technology 4G or the fifth generation mobile communication technology 5G network.

[0070] The sensing device can collect real-time status data of coal mines with folded geological structures and send the real-time status data to the monitoring device.

[0071] For example, the sensing device can collect the coal seam temperature, ..., and coal seam inclination of a coal mine with a folded geological structure in real time, and send the collected coal seam temperature, ..., and coal seam inclination to the monitoring device.

[0072] The monitoring equipment can obtain the real-time status data sent by the sensing equipment and predict the subsidence of the coal mine based on the real-time status data.

[0073] In the related art, when determining the settlement of coal mines with folded geological structures, professional surveyors mainly rely on professional surveying personnel to use professional surveying equipment to conduct field measurements and obtain measurement data to determine the settlement of the coal mine. However, due to the long measurement time and safety risks, surveyors do not conduct measurements frequently, but instead conduct surveys according to fixed cycles (for example, once every month) to obtain measurement data. Due to the long measurement cycle, the timeliness of determining the settlement based on the measurement data of each measurement cycle is poor.

[0074] To address the above issues, the inventors, while studying coal mines with folded geological structures, discovered that settlement measurements obtained based on measurement cycles are not only time-sensitive, but also, because settlement is obtained after actual settlement has occurred, cannot be used to determine the potential safety conditions of the coal mine in the future. Therefore, the inventors considered whether it would be possible to establish a prediction model that predicts settlement based on real-time status data collected by sensing equipment, thereby determining the potential settlement in the future. Based on this, after multiple experiments, the inventors discovered that an algorithmic model for predicting coal mine settlement can be obtained by optimizing and training the RVM model using a PSO algorithm based on a dataset constructed from historical status data and historical settlement of coal mines with folded geological structures. The real-time status data is then input into the pre-acquired settlement prediction model for prediction processing. This method can obtain predicted settlement after a preset time period, and the safety conditions of the coal mine can be determined based on the predicted settlement and a preset settlement threshold. Based on this, the present application proposes an artificial intelligence-based intelligent analysis system for coal mine geological structures, aiming to further improve the timeliness of determining coal mine settlement.

[0075] The following is an illustration of the artificial intelligence-based coal mine geological structure intelligent analysis system provided by this application through several specific embodiments.

[0076] Figure 2 This is a schematic diagram of the structure of the first embodiment of the artificial intelligence-based coal mine geological structure intelligent analysis system provided in this application. Figure 2 The artificial intelligence-based coal mine geological structure intelligent analysis system 10 includes a sensing device 11 and a monitoring device 12. During operation, the system 10 may include:

[0077] S201. The sensing device sends real-time status data of the coal mine to the monitoring device.

[0078] In this step, the sensing device 11 can collect real-time status data of the coal mine with the folded geological structure and send it to the monitoring device 12.

[0079] Optionally, the sensing device 11 may include multiple sensors, which are used to collect real-time status data of the coal mine. The real-time status data may include coal seam mining depth, coal seam temperature, coal seam humidity, coal seam inclination, goaf pressure and coal mine vibration frequency.

[0080] Coal seam mining depth refers to the vertical distance from the ground surface to the top of the coal seam, unit: meter (m).

[0081] Coal seam temperature refers to the temperature of underground coal seams, which is formed by the combined influence of multiple factors such as geothermal heat, surface temperature, coal seam depth and mining activities. The unit is: degrees Celsius (℃).

[0082] Coal seam humidity refers to the relative humidity of water in the coal seam. It is formed by the combined influence of multiple factors such as groundwater conditions, climatic conditions, the water content of the coal seam itself, and mine mining activities. The unit is percentage (%).

[0083] Coal seam dip refers to the angle of a coal seam relative to the horizontal plane. Due to crustal movement, previously horizontal coal seams can change, sometimes becoming steeper and sometimes flatter, taking on different shapes. Units are degrees (°). For example, a coal seam dip of 30° means the coal seam is tilted 30° upward relative to the horizontal plane.

[0084] Goaf pressure refers to the pressure state generated by the pressure of the overlying rock strata on the space formed by coal seam goaf during coal mining. The unit is megapascal (MPa).

[0085] Coal mine vibration frequency refers to the frequency of coal mine vibration caused by geological activities, blasting, mechanical equipment operation and other factors during coal mining and production, unit: Hertz (Hz).

[0086] For example, the sensing device 11 may send real-time status data of the coal mine to the monitoring device 12 , and the real-time status data may include data at three moments as shown in Table 1.

[0087] Table 1

[0088]

[0089] S202. The monitoring device receives the real-time status data of the coal mine sent by the sensing device, and obtains a geological feature matrix based on the real-time status data collected multiple times.

[0090] In this step, monitoring device 12 receives real-time status data of a coal mine with a folded address structure from sensing device 11 and obtains a geological characteristic matrix based on the real-time status data collected multiple times. In a specific implementation, a multidimensional geological characteristic vector corresponding to each moment can be determined in the real-time data to obtain multiple geological characteristic vectors. These multiple geological characteristic vectors are then combined and processed in chronological order across the multiple moments to obtain a geological characteristic matrix.

[0091] Figure 3 This is a schematic diagram of the structure of a geological feature matrix provided in the embodiment of this application. Figure 3 If time 1 corresponds to geological characteristic vector 1, time 2 can correspond to geological characteristic vector 2, and time 3 can correspond to geological characteristic vector 3. Then, according to the fact that time 1 occurs earlier than time 2, and time 2 occurs earlier than time 3, the three geological characteristic vectors can be combined to obtain a geological characteristic matrix.

[0092] S203: Input the geological characteristic matrix into the pre-acquired settlement prediction model for prediction processing to obtain the predicted settlement after a preset time period.

[0093] In this step, the monitoring equipment 12 can input the acquired geological feature matrix into a data set constructed based on the historical status data and historical settlement of the coal mine with a folded geological structure that have been acquired in advance, and use the PSO algorithm to optimize the RVM model to obtain the algorithm model for predicting the settlement of the coal mine, and perform predictive processing to obtain the predicted settlement after a preset time period.

[0094] Coal mine subsidence refers to the degree of underground subsidence caused by coal mining or related activities. During the coal mining process, due to the extraction of ore, underground cavities will collapse, resulting in a certain amount of underground subsidence.

[0095] Optionally, the data set may include multiple sets of historical feature matrices and multiple historical settlement amounts. Each set of historical feature matrices may correspond to a historical settlement amount. Each set of historical feature matrices is obtained based on historical status data within a preset time period, and the historical settlement amount is the settlement amount after a preset time period and a preset length of time.

[0096] For example, the preset time period 1 is August 1, 2022, and includes three moments: moment 1, moment 2, and moment 3. A multidimensional geological feature vector corresponding to each moment can be determined in the historical state data. By combining the multiple geological feature vectors in chronological order, a historical geological feature matrix 1 for August 1, 2022, can be obtained. Among them, moment 1 can be 8:00 on August 1, 2022, moment 2 can be 9:00 on August 1, 2022, and moment 3 can be 10:00 on August 1, 2022.

[0097] Optionally, the historical status data corresponding to time 1, time 2, and time 3 may be collected continuously, with time 1 and time 2 being adjacent times, and time 2 and time 3 being adjacent times.

[0098] Optionally, the preset duration may refer to the time interval between any moment in the preset time period, wherein the preset duration may be a value of any length of time such as 12 hours, 24 hours, or 72 hours (3 days).

[0099] For example, if moment 3 is 10:00 on August 1, 2022, and the preset duration is 24 hours, which is the time interval from moment 3, then the historical settlement 1 is the settlement 24 hours after the preset time period 1, and the corresponding specific moment can be 10:00 on August 2, 2022.

[0100] For example, the monitoring device 12 may input the acquired geological feature matrix into a pre-acquired settlement prediction model for prediction processing, and obtain a predicted settlement of the coal mine of 0.6 mm after 72 hours.

[0101] S204: Determine the safety status of the coal mine based on the predicted settlement amount and a preset settlement threshold.

[0102] In this step, the monitoring device 12 can evaluate the safety status of the coal mine by comparing the predicted settlement amount with a preset settlement threshold.

[0103] Optionally, the settlement threshold is determined by analyzing historical accident data of coal mines with folded geological structures. For example, by analyzing historical accident data of coal mines with folded geological structures, the settlement thresholds and the coal mine safety conditions corresponding to the settlement threshold ranges shown in Table 2 can be determined.

[0104] Table 2

[0105] Settlement (ΔY) Coal mine safety status ΔY≤1mm Safety 1mm<ΔY≤1.5mm Safer 1.5mm<ΔY≤2mm Danger 2mm≤ΔY Extremely dangerous

[0106] It should be noted that if the prediction result is safe, it means that the coal mine is in a safe state and can continue to work according to normal operating procedures; if the prediction result is relatively safe, it means that there are certain risks in the coal mine and the monitoring frequency and quality can be strengthened; if the prediction result is dangerous, it means that there are certain dangers in the coal mine and necessary measures should be taken, such as strengthening on-site management; if the prediction result is extremely dangerous, it means that there are extremely serious safety hazards in the coal mine and the emergency plan should be immediately activated and emergency measures should be taken, such as closing the dangerous area and notifying relevant departments.

[0107] For example, based on the predicted settlement of 0.6mm, which is less than the preset settlement threshold of 1mm, the safety of the coal mine can be determined to be safe, and mining work can continue according to normal operating procedures.

[0108] In an embodiment of the present application, an artificial intelligence-based intelligent analysis system for coal mine geological structures includes a sensing device and a monitoring device. The monitoring device can receive real-time status data of a coal mine with a folded structure geological structure sent by the sensing device, combine and process the real-time status data, obtain a geological feature matrix, input the geological feature matrix into a pre-acquired settlement prediction model for prediction processing, and obtain a settlement prediction result after a preset time period. The safety of the coal mine can be determined based on the predicted settlement and the preset settlement threshold. Through the above system, the prediction of coal mine settlement becomes more timely and accurate. In addition, by analyzing and comparing the predicted settlement results, the safety of the coal mine can be evaluated and the safety of the coal mine can be improved.

[0109] Figure 4 This is a schematic diagram of the structure of the second embodiment of the artificial intelligence-based coal mine geological structure intelligent analysis system provided in this application. Figure 4 On the basis of the above embodiment, the coal mine geological structure intelligent analysis system 10 based on artificial intelligence further includes a server 13. During operation, the server 13 may include:

[0110] S401. Construct a data set based on the historical status data of at least one coal mine with a folded geological structure and the historical settlement of each coal mine uploaded by the sensing device.

[0111] In this step, server 13 can obtain historical status data uploaded by sensing device 11 for at least one coal mine with a folded geological structure and the historical settlement of each coal mine, and process the historical data to ensure its quality and accuracy. This processed historical data can then be used to construct a data set, which includes multiple groups of samples. Each group of samples includes a geological feature matrix obtained by processing the historical status data within a preset time period, as well as settlement after a preset time period and a preset length of time.

[0112] Optionally, the sensing device 11 may include a depth sensor, a temperature sensor, a humidity sensor, a coal seam inclination sensor, a pressure sensor, and a vibration sensor. The depth sensor is used to measure the mining depth of the coal seam, the temperature sensor is used to measure the coal seam temperature, the humidity sensor is used to measure the coal seam humidity, the coal seam inclination sensor is used to measure the coal seam inclination, the pressure sensor is used to measure the pressure in the goaf, and the vibration sensor is used to measure the vibration frequency of the coal mine.

[0113] In an optional embodiment, the depth sensor and coal seam inclination sensor can be installed on the coal mining machine, the temperature sensor and humidity sensor can be installed at the coal mining face, and the pressure sensor and vibration sensor can be installed in the goaf. Specifically, the depth sensor can be installed on the top of the coal mining machine, and the coal seam inclination sensor can be installed on the support of the coal mining machine.

[0114] Optionally, processing historical data can include detecting and handling missing data and removing redundant data. Specifically, missing values ​​can be identified and filled using interpolation methods; redundant information can be found in the data, and identical records or unnecessary duplicate columns can be deleted.

[0115] In an optional embodiment, after the historical data is processed, the processed data can also be standardized using the minimum-maximum normalization method or the Z-score normalization method to eliminate dimensional differences, improve data quality, and accelerate algorithm convergence.

[0116] Optionally, the preset time period and the preset duration can be set according to specific needs and application scenarios. The preset time period of each geological feature matrix is ​​determined to be the same in length, each geological feature matrix includes the same number of moments, and the preset duration of each historical settlement is determined to be the same.

[0117] For example, if the preset time period 1 is August 1, 2022, the historical state data corresponding to three moments, namely, time 1, time 2, and time 3, may be included. Time 1 may be 8:00 AM on August 1, 2022, time 2 may be 9:00 AM on August 1, 2022, and time 3 may be 10:00 AM on August 1, 2022. The multidimensional geological feature vectors corresponding to time 1, time 2, and time 3 may be determined from the historical state data. By combining and processing the multiple geological feature vectors in the chronological order of the three moments, a historical geological feature matrix for the preset time period 1 may be obtained.

[0118] For example, if the preset time period 2 is August 2, 2022, it can include historical status data corresponding to three moments: moment 4, moment 5, and moment 6, where moment 4 is 15:00 on August 2, 2022, moment 5 is 16:00 on August 2, 2022, and moment 6 is 17:00 on August 2, 2022. The multidimensional geological feature vectors corresponding to moments 4, 5, and 6 can be determined in the historical status data. The multiple geological feature vectors are combined and processed according to the time sequence of the three moments to obtain the historical geological feature matrix within the preset time period 2.

[0119] For example, if the preset time period 1 is August 1, 2022, if the preset time period 2 is August 2, 2022, and if the preset duration is 3 days, then the historical settlement amount 1 corresponding to the preset time period 1 can be obtained, and the historical settlement amount 1 is the settlement amount 3 days after the preset time period 1; the historical settlement amount 2 corresponding to the historical preset time period 2 can be obtained, and the historical settlement amount 2 is the settlement amount 3 days after the preset time period 2.

[0120] The historical geological feature matrix within the preset time period 1 may be historical geological feature matrix 1, and the historical geological feature matrix within the preset time period 2 may be historical geological feature matrix 2. Accordingly, training sample 1 may be constructed based on historical geological feature matrix 1 and historical settlement amount 1; and training sample 2 may be constructed based on historical geological feature matrix 2 and historical settlement amount 2. Specific data for training sample 1 may be as shown in Table 3:

[0121] Table 3

[0122]

[0123] S402: Based on the data set, the PSO algorithm is used to optimize and train the RVM model to obtain a settlement prediction model.

[0124] In this step, the samples in the data set can be divided into training samples and test samples according to a preset ratio, the training samples in the data set are input into the RVM model, the parameters of the RVM model are optimized by the PSO algorithm to obtain the optimized parameters, the optimized parameters are configured in the RVM model to obtain an initial prediction model, the initial prediction model is predicted and calculated based on the test samples in the data set to obtain the output result of the initial prediction model, and the mean square error value of the output result of the initial prediction model is less than a preset mean square error threshold, the initial prediction model is determined as the settlement prediction model.

[0125] In a specific embodiment, the RVM model can capture the nonlinear relationship in the data through the Gaussian kernel function, which is defined as the distance from a point x in space to a certain center x. c It is a monotonic function of the Euclidean distance between the input data and the input data, which can map the input data into a high-dimensional feature space and realize the modeling of nonlinear relationships. Its expression is:

[0126]

[0127] Among them, σ is the kernel bandwidth parameter of the Gaussian kernel function, which is obtained by PSO algorithm optimization, and x c is the center of the Gaussian kernel function.

[0128] For example, 100 groups of samples in the data set can be divided into training samples and test samples in a ratio of 8:2, where there are 80 groups of training samples and 20 groups of test samples. The 80 groups of training samples in the data set are input into the RVM model, and the kernel bandwidth parameters of the RVM model are optimized by the PSO algorithm to obtain the optimized parameters. The optimized parameters are configured in the RVM model to obtain an initial prediction model. The initial prediction model is predicted and calculated based on the 20 groups of test samples in the data set to obtain the output result of the initial prediction model. If the mean square error value of the output result of the initial prediction model is less than a preset mean square error threshold, the initial prediction model is determined as the settlement prediction model.

[0129] In an alternative approach, new historical state data, such as water level, can be added to the dataset, and the PSO algorithm can be used to optimize and train the RVM model to obtain an algorithm model for predicting coal mine settlement. The water level can refer to the height of groundwater in coal mine tunnels or underground equipment.

[0130] For example, the sensing device may also include a water level sensor or a water level meter for real-time acquisition of the water level height. The water level height can then be added to the data set, and the geological feature matrix can be input into a pre-acquired settlement prediction model for prediction processing to obtain the predicted settlement after a preset time period. The historical status data may include coal seam mining depth, coal seam temperature, coal seam humidity, coal seam inclination, goaf pressure, coal mine vibration frequency, and water level height.

[0131] In another alternative, by adding new historical state data to the data set, such as coal seam thickness, the PSO algorithm can be used to optimize and train the RVM model to obtain an algorithm model for predicting coal mine subsidence. The coal seam thickness can be measured using a coal seam thickness meter.

[0132] S403: Configure the settlement prediction model into the monitoring equipment.

[0133] In this step, the server 13 can configure the determined settlement prediction model into the monitoring device 12. In a specific implementation, after determining the settlement prediction model, the server 13 can generate a model file and configure the generated model file into the monitoring device 12 via a network transmission protocol. The model file can include information such as model parameters and structure of the settlement prediction model.

[0134] Optionally, the network transmission protocols may include: Transmission Control Protocol / Internet Protocol (TCP / IP), Message Queuing Telemetry Transport Protocol (MQTT), and Hypertext Transfer Protocol (HTTP).

[0135] For example, after determining the settlement prediction model, the server 13 may generate a model file and configure the generated model file to the monitoring device 12 via the HTTP protocol.

[0136] In one optional solution, an AI-based intelligent analysis system for coal mine geological structures can not only obtain a settlement prediction model, but also add new historical status data to the data set, such as gas concentration, carbon dioxide concentration, or oxygen concentration, and use the PSO algorithm to optimize the RVM model to obtain an algorithm model for predicting the gas concentration hazard level of coal mines.

[0137] For example, the sensing device may also include a methane gas sensor for obtaining gas concentration in real time. The gas concentration can be added to the data set, and the geological feature matrix can be input into the pre-acquired gas concentration hazard level prediction model for predictive processing. The predicted gas concentration hazard level after a preset time period can be obtained. The gas concentration hazard level prediction model is a data set constructed based on the pre-acquired historical status data of coal mines with folded geological structures and historical gas concentration hazard levels. The PSO algorithm is used to optimize the RVM model to obtain an algorithm model for predicting the gas concentration hazard level of the coal mine. The historical status data may include coal seam mining depth, coal seam temperature, coal seam humidity, coal seam inclination, goaf pressure, coal mine vibration frequency and gas concentration.

[0138] In an embodiment of the present application, an artificial intelligence-based intelligent analysis system for coal mine geological structures includes a sensing device, a server, and a monitoring device. The server can construct a data set based on the historical status data of at least one coal mine with a folded geological structure and the historical settlement of each coal mine uploaded by the sensing device. Based on the data set, the RVM model is optimized and trained using the PSO algorithm to obtain a settlement prediction model, and the settlement prediction model is configured in the monitoring device so that the monitoring device can perform prediction processing on the real-time status data of the coal mine with a folded geological structure sent by the sensing device according to the settlement prediction model, and obtain the settlement prediction result after a preset time period. Through the above system, it is possible to obtain a prediction of the future settlement of a coal mine with a folded geological structure during real-time monitoring, making the prediction of coal mine settlement more timely and improving the safety of the coal mine.

[0139] Figure 5 This is a flow chart of the first embodiment of the intelligent analysis method for coal mine geological structure based on artificial intelligence provided by this application. Figure 5 The method is applied to the coal mine geological structure intelligent analysis system 10 based on artificial intelligence. On the basis of the above embodiments, step S402 specifically includes:

[0140] S501: Divide the samples in the data set into training samples and test samples according to a preset ratio.

[0141] For example, 100 groups of samples in a data set can be divided into training samples and test samples in a ratio of 8:2, where there are 80 groups of training samples and 20 groups of test samples.

[0142] S502: Input the training samples in the data set into the RVM model, and optimize the parameters of the RVM model using the PSO algorithm to obtain optimized parameters.

[0143] In this step, the parameters of the RVM model can be optimized using the PSO algorithm based on the training samples obtained at a preset ratio to obtain the optimized parameters. The specific implementation can include the following steps:

[0144] In step a, the particle swarm of the PSO algorithm is randomly initialized according to the preset number of particles to obtain an initial particle swarm. The position of each particle in the initial particle swarm is the kernel width parameter of the Gaussian kernel function in the RVM model.

[0145] For example, the preset number of particles is 20, which means the initial size of the particle swarm is 20. By randomly initializing the particle swarm of the PSO algorithm, the initial position of each particle in the particle swarm can be obtained.

[0146] In step b, the objective function of the RVM model is determined as the fitness function of each particle in the initial particle swarm, and the objective function is the mean square error function.

[0147] The mean square error function, as a fitness function, can quantify the difference between the predicted results and the actual values. The mean square error calculates the square of the difference between each predicted value and the corresponding actual value and takes the average value, so it can reflect the overall prediction accuracy of the model.

[0148] Step c: configure the kernel width parameter corresponding to each particle in the initialized particle swarm in the RVM model, perform prediction calculation on the training samples in the data set, and calculate the fitness value corresponding to each particle through the fitness function.

[0149] The fitness function evaluates the performance of each particle, typically based on the error between the predicted value and the actual value. By continuously updating the particle's position and velocity and adjusting the parameters based on its fitness value, the particle swarm algorithm can find the optimal parameter configuration, enabling the RVM model to perform more accurate predictions.

[0150] In step d, the position and velocity of each particle in the initial particle swarm are updated according to the fitness value corresponding to each particle, and the fitness value corresponding to each updated particle is recalculated. Steps c to d are repeated until the preset number of iterations is reached to obtain the target particle swarm.

[0151] Specifically, the velocity update calculation formula of particles in the particle swarm is:

[0152]

[0153] The position update calculation formula is:

[0154]

[0155] Among them, ω is the inertia weight parameter, which is used to balance the global search and local search capabilities. is the velocity vector of particle i in the dth dimension in the kth iteration, c1 is the cognitive coefficient, which reflects the ability of the particle to follow its own cognition, c2 is the social coefficient, which reflects the ability of the particle to follow the optimal position of the group; r1 and r2 are random numbers with values ​​in [0, 1]. is the individual best position of particle i in the d-th dimension in the k-th iteration, The global optimal position of particle i in the dth dimension at the kth iteration.

[0156] Optionally, existing methods for adjusting the inertia weight parameters for the PSO algorithm include linearly decreasing the inertia weight, adaptively adjusting the inertia weight, or randomizing the inertia weight.

[0157] In one alternative, since the performance of a particle swarm optimization algorithm is affected by inertia weight, a smaller inertia weight can improve local search capabilities and enhance solution accuracy, while a larger inertia weight can enhance global search capabilities and, to a certain extent, avoid being trapped in local optima. Therefore, when using the PSO algorithm to optimize the RVM model, the existing method for adjusting the inertia weight parameter in the PSO algorithm can be integrated with the local search algorithm to improve the accuracy of the settlement prediction model. Local search algorithms can include hill climbing, simulated annealing, or other optimization algorithms.

[0158] For example, the settlement prediction model is a data set constructed based on the historical status data and historical settlement of coal mines with folded geological structures that have been acquired in advance. While the PSO algorithm is used to optimize the RVM model for training, the simulated annealing algorithm is used to optimize the inertia weight of the PSO algorithm to obtain an algorithm model for predicting the settlement of coal mines.

[0159] In step e, the kernel width parameter corresponding to the position of the particle with the smallest fitness value in the target particle group is determined as the optimization parameter.

[0160] S503: Configure the optimized parameters in the RVM model to obtain an initial prediction model.

[0161] S504: Perform prediction calculation on the initial prediction model according to the test samples in the data set to obtain an output result of the initial prediction model.

[0162] In this step, the test samples in the data set can be used to perform prediction calculations on the initial prediction model after parameter optimization, and the output results of the initial prediction model.

[0163] S505: If the mean square error value of the output result of the initial prediction model is less than a preset mean square error threshold, the initial prediction model is determined as the settlement prediction model.

[0164] In this step, if the mean square error (MSE) of the initial prediction model's output is less than a preset MSE threshold, the initial prediction model can be determined as the settlement prediction model. This indicates that the initial prediction model performs well on the current dataset, with a small deviation between its predictions and the actual values, meeting the preset accuracy requirements.

[0165] S506: If the mean square error value of the output result of the initial prediction model is greater than or equal to the preset mean square error threshold, the number of iterations corresponding to the PSO algorithm is increased according to the preset rule to obtain a new number of iterations.

[0166] In this step, if the mean square error value of the output result of the initial prediction model is greater than or equal to the preset mean square error threshold, the performance of the model can be improved by increasing the number of iterations corresponding to the PSO algorithm.

[0167] Optionally, increasing the number of iterations gives the PSO algorithm more opportunities to search for the optimal solution in the parameter space. By increasing the number of iterations, the PSO algorithm can continue to update the positions and velocities of the particles and continuously explore better parameters to find a better kernel width parameter.

[0168] The specific rules for increasing the number of iterations can be defined based on the actual situation. A common strategy is to dynamically adjust the number of iterations based on previous experimental experience or adaptive methods. For example, if the initial number of iterations k is 100 and the number of iterations increases by 100, the second number of iterations k will be 200, and the third number of iterations k will be 300.

[0169] S507: Input the training samples in the data set into the RVM model, and optimize the parameters of the RVM model again using the PSO algorithm according to the new number of iterations to obtain new optimized parameters.

[0170] S508: Configure the new optimization parameters in the RVM model to obtain a new prediction model.

[0171] S509: Perform prediction calculation on the new prediction model according to the test samples in the data set to obtain the output result of the new prediction model.

[0172] S510: If the mean square error value of the output result of the new prediction model is less than a preset mean square error threshold, the new prediction model is determined as the settlement prediction model.

[0173] In an embodiment of the present application, the server can input the training samples obtained by dividing according to a preset ratio into the RVM model, optimize the parameters of the RVM model through the PSO algorithm to obtain the optimized parameters, obtain the initial prediction model based on the optimized parameters, input the test sample into the initial prediction model for prediction calculation, obtain the output result of the initial prediction model, and determine the initial prediction model as the settlement prediction model based on the mean square error of the output result being less than the pre-set mean square error threshold. On the contrary, if the mean square error value of the output result of the initial prediction model is greater than or equal to the pre-set mean square error threshold, the number of iterations corresponding to the PSO algorithm can be increased to obtain new optimized parameters and the corresponding prediction model, and then the new prediction model is predicted and calculated through the test sample until the mean square error value of the output result of the new prediction model is less than the pre-set mean square error threshold, and the new prediction model is determined as the settlement prediction model. In the above process, the parameters of the RVM model can be continuously adjusted by the PSO algorithm to obtain the settlement prediction model, thereby realizing the prediction of the settlement of coal mines with folded geological structures.

[0174] Figure 6 This is a schematic diagram of the structure of the third embodiment of the artificial intelligence-based coal mine geological structure intelligent analysis system provided in this application. Figure 6 On the basis of the above embodiment, the coal mine geological structure intelligent analysis system 10 based on artificial intelligence further includes a terminal device 14. During operation, the system may include:

[0175] S601: When the predicted settlement amount is greater than or equal to the settlement amount threshold, the monitoring device pushes an alarm message to the terminal device.

[0176] In this step, the monitoring device 12 can push an alarm message indicating that there is a safety risk in the coal mine to the terminal device 14 when the predicted settlement is greater than or equal to the settlement threshold according to the output result of the settlement prediction model.

[0177] Optionally, the terminal device 14 may include a smart phone and a large monitoring screen.

[0178] For example, the monitoring device 12 can obtain a predicted settlement of 3 mm after a preset period of time of 3 days based on the output results of the settlement prediction model. Since the predicted settlement of 3 mm is greater than the preset settlement threshold of 2 mm corresponding to the extremely dangerous safety condition of the coal mine, it is determined that there is an extremely serious safety hazard in the coal mine, and an alarm message can be sent to the smart phone of the coal mine manager.

[0179] S602: The terminal device receives the alarm information sent by the monitoring device.

[0180] In this step, the terminal device 14 can receive the alarm information sent by the monitoring device 12 to remind the management personnel to take corresponding measures.

[0181] For example, a smartphone can receive an alarm and alert management personnel by emitting a sound or vibrating. A monitoring screen can also receive the alarm and display it on the screen, letting management personnel know the predicted settlement amount.

[0182] In an embodiment of the present application, the artificial intelligence-based intelligent analysis system for coal mine geological structures also includes a terminal device that can receive alarm information sent by monitoring equipment and provide timely alarms and feedback to management personnel through various means to help management personnel take corresponding measures to ensure the safety and health of personnel.

[0183] Figure 7 This is a flow chart of the second embodiment of the intelligent analysis method for coal mine geological structure based on artificial intelligence provided by this application. Figure 7 , the method comprising:

[0184] S701: Build a data set.

[0185] In this step, the server 13 may construct a data set based on the historical status data of the coal mines with folded geological structures and the historical settlement of each coal mine uploaded by the sensing device 11 .

[0186] S702: Input the training samples in the data set into the RVM model, train the RVM model through the PSO algorithm, and obtain optimized parameters.

[0187] S703: Configure the optimized parameters into the RVM model to obtain a settlement prediction model.

[0188] S704: Perform prediction processing on the real-time status data of the coal mine sent by the sensing device, and output the predicted settlement result of the coal mine with a folded geological structure.

[0189] The artificial intelligence-based intelligent analysis method for coal mine geological structure provided in the embodiment of the present application is applied to the artificial intelligence-based intelligent analysis system for coal mine geological structure. During specific implementation, reference can be made to the working process of the artificial intelligence-based intelligent analysis system for coal mine geological structure in the above embodiment. Its implementation principles and beneficial effects are similar and will not be repeated here.

[0190] Figure 8 This is a schematic diagram of the system for intelligent analysis of coal mine geological structures based on artificial intelligence provided in the embodiment of this application. Figure 8The system 20 may include a data platform 21 and a service platform 22. The data platform 21 may include a sensing device 11 and a server 13. The service platform 22 may include a monitoring device 12 and a terminal device 14.

[0191] Among them, the sensing device 11 may include various types of intelligent coal mine geological sensors and real-time monitoring equipment to obtain real-time status data of the coal mine, remote sensing data, geophysical data, etc.

[0192] The server 13 can obtain real-time status data, remote sensing data and geophysical data uploaded by the sensing equipment and store the data.

[0193] Optionally, the sensing device 11 and the server 13 may select any one of the following methods for data transmission according to actual conditions.

[0194] Method 1: Direct connection. The sensor can be directly connected to the server through a network interface, such as Ethernet or wireless (Wireless-Fidelity, Wi-Fi).

[0195] Method 2: Gateway-mediated connection: Sensors can be connected to a local area network gateway, which is responsible for transmitting the sensor data to the server. The gateway can connect to the server via Ethernet, wireless, or cellular networks, and use TCP / IP, MQTT, or HTTP protocols for data transmission.

[0196] Method 3: Wireless communication. Sensors can use wireless communication technologies such as Bluetooth, Zigbee, and LoRa to transmit data to an intermediate device, which then uploads the data to the server. This method is suitable for situations where the distance between the sensor and the server is far or the transmission method is limited.

[0197] Method 4: Mobile network. Sensors can use mobile networks, such as 4G or 5G networks, to upload data directly to the server. This is suitable for situations where sensors need to work in a mobile environment.

[0198] This application does not limit the data transmission method between the sensing device 11 and the server 13.

[0199] The server 13 can use the PSO algorithm to optimize and train the RVM model based on the pre-acquired historical status data of the coal mine with folded geological structure and the data set constructed by historical settlement to obtain a settlement prediction model, and configure the settlement prediction model to the monitoring device 12.

[0200] Optionally, the server 13 can also use the PSO algorithm to optimize the RVM model based on the pre-acquired historical status data of coal mines with folded geological structures and a data set constructed by historical gas concentration hazard levels to obtain an algorithm model for predicting the gas concentration hazard level of the coal mine, and configure the algorithm model into the monitoring device 12.

[0201] The monitoring device 12 is used to push alarm information to the terminal device 14 when the predicted settlement amount is greater than or equal to the settlement amount threshold.

[0202] The terminal device 14 is used to receive the alarm information sent by the monitoring device.

[0203] The system principles provided in the embodiments of this application can be referred to the working process of the coal mine geological structure intelligent analysis system based on artificial intelligence in the above embodiments. Its implementation principles and beneficial effects are similar and will not be repeated here.

[0204] Accordingly, an embodiment of the present application provides a computer-readable storage medium, which stores computer execution instructions. When the computer execution instructions are executed by a processor, they are used to implement the artificial intelligence-based coal mine geological structure intelligent analysis system described in the above embodiment.

[0205] Accordingly, an embodiment of the present application may also provide a computer program product, including a computer program, which, when executed by a processor, can implement the artificial intelligence-based coal mine geological structure intelligent analysis system described in the above embodiment.

[0206] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0207] It should be understood that the present application is not limited to the exact structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. An artificial intelligence-based coal mine geological structure intelligent analysis system, characterized in that: Including sensing equipment and monitoring equipment; The sensing device includes a plurality of sensors, each of which is used to collect real-time status data of the coal mine, wherein the real-time status data includes coal seam mining depth, coal seam temperature, coal seam humidity, coal seam inclination, goaf pressure, coal mine vibration frequency, and water level; The monitoring equipment is used to: receiving real-time status data of the coal mine sent by the sensing device, determining a multidimensional geological characteristic vector corresponding to each moment in the real-time status data to obtain a plurality of geological characteristic vectors, and combining the plurality of geological characteristic vectors in a time sequence of the plurality of moments to obtain a geological characteristic matrix; Inputting the geological feature matrix into a pre-acquired settlement prediction model for prediction processing to obtain a predicted settlement after a preset time period, the settlement prediction model is a data set constructed based on pre-acquired historical status data and historical settlement of a coal mine with a folded geological structure, and an algorithm model for predicting the settlement of the coal mine is obtained by optimizing and training a relevance vector machine (RVM) model using a particle swarm optimization (PSO) algorithm integrated with a local search algorithm, wherein the local search algorithm is a hill climbing algorithm or a simulated annealing algorithm; The safety condition of the coal mine is determined based on the predicted settlement amount and a preset settlement threshold value, wherein the settlement threshold value is determined by analyzing historical accident data of coal mines with folded geological structures.

2. The system according to claim 1, wherein: The sensing devices include a depth sensor, a temperature sensor, a humidity sensor, a coal seam inclination sensor, a pressure sensor, a vibration sensor, and a water level sensor; The depth sensor is used to measure the mining depth of the coal seam; The temperature sensor is used to measure the temperature of the coal seam; The humidity sensor is used to measure the humidity of the coal seam; The coal seam inclination sensor is used to measure the coal seam inclination; The pressure sensor is used to measure the pressure of the goaf; The vibration sensor is used to measure the vibration frequency of the coal mine; The water level sensor is used to measure the water level height.

3. The system according to claim 1 or 2, characterized in that The system further includes a server configured to: Based on the historical status data of at least one coal mine with a folded geological structure and the historical settlement amount of each coal mine uploaded by the sensing device, the data set is constructed, wherein the data set includes multiple groups of samples, each group of samples includes a geological feature matrix obtained by processing the historical status data within a preset time period, and the settlement amount after a preset time period after the preset time period; According to the data set, the PSO algorithm integrated with the local search algorithm is used to optimize the RVM model for training to obtain the settlement prediction model; The settlement prediction model is configured in the monitoring equipment.

4. The system according to claim 3, characterized in that The server optimizes and trains the RVM model using a PSO algorithm integrated with a local search algorithm based on the data set to obtain the settlement prediction model, including: Dividing the samples in the data set into training samples and test samples according to a preset ratio; Inputting the training samples in the data set into the RVM model, optimizing the parameters of the RVM model by the PSO algorithm integrated with the local search algorithm to obtain optimized parameters; Configuring the optimization parameters in the RVM model to obtain an initial prediction model; Performing prediction calculation on the initial prediction model according to the test samples in the data set to obtain an output result of the initial prediction model; If the mean square error value of the output result of the initial prediction model is less than a preset mean square error threshold, the initial prediction model is determined as the settlement prediction model.

5. The system according to claim 4, characterized in that The server is also used to: If the mean square error value of the output result of the initial prediction model is greater than or equal to the preset mean square error threshold, the number of iterations corresponding to the PSO algorithm integrated with the local search algorithm is increased according to the preset rules to obtain a new number of iterations; Inputting the training samples in the data set into the RVM model, and optimizing the parameters of the RVM model again using the PSO algorithm integrated with the local search algorithm according to the new number of iterations to obtain new optimized parameters; Configuring the new optimization parameters in the RVM model to obtain a new prediction model; Performing prediction calculation on the new prediction model according to the test samples in the data set to obtain an output result of the new prediction model; If the mean square error value of the output result of the new prediction model is less than a preset mean square error threshold, the new prediction model is determined as the settlement prediction model.

6. The system according to claim 4, characterized in that The server inputs the training samples in the data set into the RVM model, and optimizes the parameters of the RVM model by the PSO algorithm integrated with the local search algorithm to obtain optimized parameters, including: Step a, randomly initializing the particle swarm of the PSO algorithm integrated with the local search algorithm according to a preset number of particles to obtain an initial particle swarm, wherein the position of each particle in the initial particle swarm is the kernel width parameter of the Gaussian kernel function in the RVM model; Step b, determining the objective function of the RVM model as the fitness function of each particle in the initial particle swarm, wherein the objective function is a mean square error function; Step c, configuring the kernel width parameter corresponding to each particle in the initial particle swarm in the RVM model, performing prediction calculation on the training samples in the data set, and calculating the fitness value corresponding to each particle using the fitness function; Step d, updating the position and velocity of each particle in the initial particle swarm according to the fitness value corresponding to each particle, and recalculating the fitness value corresponding to each updated particle, repeating steps c-d until a preset number of iterations is reached to obtain a target particle swarm; Step e: determining the kernel width parameter corresponding to the position of the particle with the smallest fitness value in the target particle group as the optimization parameter.

7. The system according to claim 1 or 2, characterized in that The system further includes: a terminal device; The monitoring device is further configured to push an alarm message to the terminal device when the predicted settlement amount is greater than or equal to the settlement amount threshold, wherein the alarm message is configured to indicate that there is a safety risk in the coal mine; The terminal device is used to receive the alarm information sent by the monitoring device.

Images