Analysis method and system for roadway floor heave mechanism

Abstract

The invention provides an analysis method and system for a roadway floor heave mechanism, and relates to the technical field of roadway floor heave mechanism analys.The method comprises the steps that firstly, a plurality of monitoring points are arranged in a roadway, and monitoring data, pressure, vibration, temperature, humidity and other data are collected in real time; and constructing a microscopic model and a macroscopic model of the roadway by applying a discrete element method and a finite element method based on monitoring data. State evaluation is carried out on historical monitoring data and corresponding contact force and total stress through an expert scoring method, and a risk prediction model based on a random forest algorithm is constructed. And predicting the state of each monitoring point at the current moment through the model so as to mark abnormal monitoring points. And analyzing the floor heave risk index at the abnormal state monitoring point, and judging whether to give an early warning or not. According to the method, multi-source data and model analysis are effectively integrated, a scientific basis is provided for roadway safety, early warning of the roadway floor heave risk can be achieved through the method, and potential safety hazards can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of roadway floor heave mechanism analysis, and specifically provides an analysis method and system for roadway floor heave mechanism. Background Art

[0002] Roadway floor heave is a common geological phenomenon in underground mines, tunnels, and other underground projects. It refers to the surface settlement or bottom uplift phenomenon of the roadway floor caused by factors such as pressure changes, soil structure damage, or hydrogeological condition changes. This phenomenon not only affects the safety and stability of underground projects but may also have an adverse impact on the surrounding environment and even trigger safety accidents.

[0003] In modern mine exploitation and underground engineering, the phenomenon of roadway floor heave is a widespread and serious technical problem. The occurrence of floor heave is usually closely related to the surrounding geological conditions, construction methods, and the passage of time. Traditional roadway floor heave analysis methods often rely on experience and simple physical models, resulting in insufficient prediction of the stress state of roadways in complex geological environments and difficulty in accurately identifying potential floor heave risks. Such deficiencies not only pose construction safety hazards but may also delay the project progress and increase costs.

[0004] The existing technologies have the following deficiencies:

[0005] The deficiencies of the existing technologies are mainly reflected in the insufficient timeliness of data collection and processing, limited accuracy in model establishment, and poor adaptability to complex geological conditions. Traditional monitoring methods often rely on a single sensor, resulting in incomplete data acquisition and inability to fully reflect the true state of the roadway. In addition, existing numerical models often fail to achieve ideal accuracy when dealing with complex geological environments, dynamic changes, and multi-factor coupling, thereby affecting the accuracy and timeliness of risk assessment. These deficiencies pose challenges to the monitoring and early warning systems for roadway floor heave in practical applications and urgently require improvement and innovation.

[0006] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure. Therefore, it may include information that does not constitute prior art known to those of ordinary skill in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide an analysis method and system for roadway floor heave mechanism to solve the problems raised in the above background art.

[0008] To achieve the above purpose, the present invention provides the following technical solutions:

[0009] An analysis method for roadway floor heave mechanism, the specific steps include:

[0010] Step 1: Arrange multiple monitoring points in the roadway to collect the monitoring data, pressure, vibration, temperature, and humidity data of the monitoring points at the current moment, and obtain the historical monitoring data of each monitoring point at previous historical moments. The monitoring data includes stress, displacement, and material parameters;

[0011] Step 2: Based on the monitoring data, construct a microscopic model and a macroscopic model of the roadway, and couple the microscopic model with the macroscopic model to obtain the contact force and total stress data of the particles in the roadway under the corresponding monitoring data;

[0012] Step 3: Use the expert scoring method to evaluate the state of the historical monitoring data of the monitoring points at previous historical moments and the corresponding contact force and total stress data, and assign corresponding state labels. Based on the random forest algorithm, construct a risk prediction model, and use the historical monitoring data and the corresponding contact force and total stress data as inputs, and the state labels as tags to train the risk prediction model. The state labels include normal state and abnormal state;

[0013] Step 4: Input the monitoring data, contact force, and total stress data of each monitoring point at the current moment into the risk prediction model to obtain the state labels of each monitoring point at the current moment;

[0014] Step 5: Analyze the monitoring points marked as abnormal states, calculate the floor heave risk index of the monitoring points at the current moment, and compare it with the preset risk threshold to determine whether to issue a warning.

[0015] Furthermore, the specific logic for collecting data at the roadway monitoring points is as follows:

[0016] Starting from the starting point of the roadway, set a monitoring point every 3 meters. At the monitoring points, arrange stress sensors, displacement sensors, pressure sensors, vibration sensors, temperature sensors, and humidity sensors to collect the stress, displacement, pressure, vibration, temperature, and humidity data at the monitoring points. The material parameters include normal stiffness, tangential stiffness, elastic modulus of the material, and Poisson's ratio.

[0017] Furthermore, the specific logic for constructing the microscopic model and macroscopic model of the roadway is as follows:

[0018] Use the discrete element method to construct a microscopic soil model. The expression for the contact force between particles in the model is as follows:

[0019] F ij =F n,ij +F t,ij =k n,ij ·δ n,ij +k t,ij ·δ t,ij

[0020] where Fij represents the contact force between particle i and particle j, F n,ij represents the normal force between particle i and particle j, F t,ij represents the tangential force between particle i and particle j, k n,ij is the normal stiffness between particle i and particle j, k t,ij is the tangential stiffness between particle i and particle j, δ n,ij is the normal displacement between particle i and particle j, δ t,ij is the tangential displacement between particle i and particle j, and i and j are the indices of the particles;

[0021] The normal displacement of the particles is determined by the amount of overlap at the contact point of the two particles, and the calculation formula is:

[0022] δ n,ij = R i + R j - d ij d ij < R i + R j

[0023] where, R i is the radius of particle i, R j is the radius of particle j, d ij is the distance between the centers of particle i and particle j. When d ij ≥ R i + R j the particles are not in contact, and the normal displacement δ n,ij = 0; The tangential displacement is the relative displacement within the contact surface and can be directly calculated and obtained in this discrete element method model. And in the discrete element method model, the tangential force needs to satisfy the Coulomb friction condition:

[0024] F t ≤ μ·F n

[0025] where, μ is the friction coefficient. If the tangential force F t exceeds the friction force limit, slip occurs and the tangential displacement needs to be readjusted;

[0026] The formula for obtaining the normal stiffness between particles is:

[0027]

[0028] where, E * is the equivalent elastic modulus, R * ij is the equivalent contact radius between particle i and particle j. The equivalent contact radius is dynamically calculated from the radii of particle i and particle j in the discrete element method model;

[0029]

[0030] Among them, E i is the elastic modulus of particle i, and E j is the elastic modulus of particle j, and v i is the Poisson's ratio of particle i, and v j is the Poisson's ratio of particle j;

[0031]

[0032] Among them, R i is the radius of the contacting particle i, and r j is the radius of the contacting particle J;

[0033] The tangential stiffness is calculated using an empirical ratio:

[0034] k t,ij = η·k n,ij

[0035] Among them, η is the proportionality coefficient;

[0036] The formula for obtaining the stress matrix by constructing a macroscopic roadway structure model using the finite element method is:

[0037] σ A = D·∈

[0038] Among them, σ A represents the macroscopic stress matrix, D is the elastic modulus matrix of the material, and ∈ is the strain matrix;

[0039] The formula for calculating the total stress in the macroscopic model is:

[0040]

[0041] Among them, σ macro represents the total stress in the macroscopic model, V represents the volume of the macroscopic model, F op represents the contact force between particles o and p in the microscopic model, n represents the total number of particles in the macroscopic model, o and p are the indices of the particles, and o, p ∈ [1, n], o ≠ p;

[0042] The formula for coupling the microscopic model and the macroscopic model and feeding back the contact force in the microscopic model to the macroscopic model is:

[0043]

[0044] Among them, σ macro ' represents the total stress after coupling the microscopic model, k n,ab is the normal stiffness of particles a and b, k t,ab is the tangential stiffness of particles a and b, δn,ab is the normal displacement of particle a and particle b, δ t,ab is the tangential displacement of particle a and particle b, N represents the total number of particles in the microscopic model, a and b are the indices of the particles, a, b ∈ [1, N], a ≠ b;

[0045] Feed the macroscopic stress back into the microscopic model, and the formula for correcting the contact force between particles is expressed as follows:

[0046] F ij ′ = k n,ij ·δ n,ij + k t,ij ·δ t,ij + σ macro ·A ij

[0047] Among them, F ij ′ represents the contact force between particle i and particle j after coupling the macroscopic model, A ij is the contact area between particle i and particle j;

[0048] For a spherical contact area, it is calculated through the contact radius, and the formula is:

[0049] A ij = π|R * ij ·δ n,ij |

[0050] Among them, R * ij is the equivalent contact radius of particle i and particle j, and the equivalent contact radius can be dynamically calculated from the radii of particle i and particle j in the discrete element method model.

[0051] Furthermore, the specific logic for constructing the risk prediction model is:

[0052] The historical monitoring data of the monitoring points at past historical moments, as well as the corresponding contact force and total stress data, are evaluated for their status using the expert scoring method, and are assigned tags for the corresponding normal or abnormal states. Among them, the normal state is represented by the risk value B = 0, and the abnormal state is represented by the risk value B = 1. The historical monitoring data of the monitoring points at past historical moments, as well as the corresponding contact force and total stress data, are normalized and then integrated into a sample data set. The sample data set is divided into a training set and a test set, with a division ratio of 7:3. A risk prediction model is constructed based on the random forest algorithm. Using the historical monitoring data, contact force, and total stress data in the training set as inputs, and the corresponding status labels as tags, the risk prediction model is trained. Then, the historical monitoring data, contact force, and total stress in the test set are input to evaluate the performance of the model. The evaluation indicators include accuracy, precision, and recall. If, in five iterations, all the change amplitudes of the evaluation indicators are less than 0.01, it is determined that the model training is completed. Further, the specific logic for obtaining the state predicted by the model at the current moment is as follows:

[0053] The monitoring data, contact force, and total stress data of each monitoring point at the current moment are normalized and then input into the trained risk prediction model to obtain the risk value predicted by the model. If the risk value B = 1, it is an abnormal state; if the risk value B = 0, it is a normal state, and the monitoring points in the abnormal state are marked.

[0054] Further, the specific logic for calculating the floor heave risk index at the abnormal state monitoring points at the current moment is as follows:

[0055] Obtain the stress, displacement, pressure, vibration, temperature, and humidity data at the monitoring points marked as abnormal states. The formula for obtaining the floor heave risk index is:

[0056]

[0057] where R is the floor heave risk index, P is the pressure, E is the stress, D is the displacement, V is the vibration amplitude, T is the temperature, is the ideal temperature value, H is the humidity, is the ideal humidity value, ω 1 、ω 2 、ω 3 、ω 4 、ω 5 are the weight coefficients of their respective terms, 0 < ω 1 < ω 5 < ω 4 < ω 2 < ω 3 < 1, and ω 1 + ω 2 + ω 3 + ω 4 + ω 5= 1;

[0058] Compare the floor heave risk index at the current moment with a preset risk threshold. If the floor heave risk index at the current moment is lower than the preset risk threshold, continue the monitoring. If the floor heave risk index at the current moment is higher than the preset risk threshold, there is an abnormality at the corresponding monitoring point, and a danger warning is issued.

[0059] The present invention further provides an analysis system for the mechanism of roadway floor heave. The analysis system for the mechanism of roadway floor heave is used to implement the above-mentioned analysis method for the mechanism of roadway floor heave, and includes:

[0060] A data acquisition module, which is used to arrange a plurality of monitoring points in the roadway to collect the monitoring data, pressure, vibration, temperature and humidity data of the monitoring points at the current moment, and obtain the historical monitoring data of each monitoring point at past historical moments. The monitoring data includes stress, displacement and material parameters;

[0061] A multi-scale model construction module, which is used to construct a microscopic model and a macroscopic model of the roadway based on the monitoring data, and couple the microscopic model with the macroscopic model to obtain the contact force and total stress data of the particles in the roadway under the corresponding monitoring data;

[0062] A risk prediction model construction module, which is used to adopt the expert scoring method to evaluate the state of the historical monitoring data of the monitoring points at past historical moments and the corresponding contact force and total stress data, and assign corresponding state marks. A risk prediction model is constructed based on the random forest algorithm, and the historical monitoring data and the corresponding contact force and total stress data are used as inputs, and the state marks are used as labels to train the risk prediction model. The state marks include normal state and abnormal state;

[0063] A real-time state prediction module, which is used to input the monitoring data, contact force and total stress data of each monitoring point at the current moment into the risk prediction model to obtain the state marks of each monitoring point at the current moment;

[0064] An abnormal state analysis module, which is used to analyze the monitoring points marked as abnormal states, calculate the floor heave risk index of the monitoring point at the current moment, and compare it with a preset risk threshold to determine whether to issue a warning.

[0065] In the above technical solution, the technical effects and advantages provided by the present invention:

[0066] By integrating multi-source data monitoring and advanced numerical simulation techniques, the present invention realizes real-time and comprehensive assessment of the risk of roadway floor heave. Its remarkable advantage lies in effectively improving the accuracy and timeliness of monitoring. By arranging multiple sensors to collect key data such as stress and displacement in real time, and combining the discrete element method and the finite element method to construct microscopic and macroscopic models, this method can not only deeply analyze the complex mechanical behavior inside the roadway, but also conduct risk prediction through the random forest algorithm, thus achieving accurate early warning. Combining the comprehensive analysis of data at monitoring points with abnormal conditions, the risk of floor heave can be identified in a timely manner. This systematic analysis method provides a scientific basis for roadway safety management, helps to identify potential risks in advance, reduce the accident rate, enhance the safety and stability of mine operations, and has high application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 is a schematic diagram of the overall method flow of the present invention;

[0068] Figure 2 is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] To make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to specific embodiments.

[0070] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those of ordinary skill in the field to which the present invention belongs. The "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this term cover the elements or objects listed after this term and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left", "right" are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0071] Embodiment:

[0072] Please refer to Figure 1 , the present invention provides a technical solution:

[0073] A method for analyzing the mechanism of roadway floor heave, the specific steps include:

[0074] Step 1: Arrange multiple monitoring points in the roadway to collect the monitoring data, pressure, vibration, temperature, and humidity data of the monitoring points at the current moment, and obtain the historical monitoring data of each monitoring point at past historical moments. The monitoring data includes stress, displacement, and material parameters;

[0075] In this embodiment, the specific logic for collecting data at the roadway monitoring points is as follows:

[0076] Starting from the starting point of the roadway, set a monitoring point every 3 meters. Arrange stress sensors, displacement sensors, pressure sensors, vibration sensors, temperature sensors, and humidity sensors at the monitoring points to collect the stress, displacement, pressure, vibration, temperature, and humidity data at the monitoring points. The material parameters include normal stiffness, tangential stiffness, elastic modulus, and Poisson's ratio of the material. The elastic modulus and Poisson's ratio of the material can be obtained through literature queries or material databases.

[0077] Setting monitoring points every 3 meters in the roadway and arranging various sensors can comprehensively and real-time obtain various environmental and mechanical data. The collection of such multi-dimensional data can effectively reflect the complex state inside the roadway, improving the comprehensiveness and accuracy of monitoring. Secondly, the combination of real-time data and historical monitoring data can better identify trends and change patterns, providing a scientific basis for the early warning of the risk of roadway floor heave. This step lays a solid foundation for the subsequent risk assessment and formulation of management measures.

[0078] Step 2: Based on the monitoring data, construct a microscopic model and a macroscopic model of the roadway, and couple the microscopic model with the macroscopic model to obtain the contact force and total stress data of the particles in the roadway corresponding to the monitoring data;

[0079] In this embodiment, the specific logic for constructing the microscopic model and macroscopic model of the roadway is as follows:

[0080] Use the discrete element method to construct a microscopic soil model. The expression for the contact force between particles in the model is as follows:

[0081] F ij =F n,ij +F t,ij =k n,ij ·δ n,ij +k t,ij ·δ t,ij

[0082] Among them, F ij represents the contact force between particle i and particle j, F n,ij represents the normal force between particle i and particle j, F t,ij represents the tangential force between particle i and particle j, k n,ij is the normal stiffness between particle i and particle j, k t,ijis the tangential stiffness of particle i and particle j, δ n,ij is the normal displacement of particle i and particle j, δ t,ij is the tangential displacement of particle i and particle j, where i and j are the indices of the particles;

[0083] The normal displacement of the particles is determined by the overlap amount at the contact point of the two particles, and the calculation formula is:

[0084] δ n,ij =R i +R j -d ij ,d ij <R i +R j

[0085] Among them, R i is the radius of particle i, R j is the radius of particle j, d ij is the distance between the centers of particle i and particle j. When d ij ≥R i +R j , there is no contact between the particles, and the normal displacement δ n,ij =0; The tangential displacement is the relative displacement within the contact surface and can be directly calculated and obtained in this discrete element method model. And in the discrete element method model, the tangential force needs to satisfy the Coulomb friction condition:

[0086] F t ≤μ·F n

[0087] Among them, μ is the friction coefficient. If the tangential force F t exceeds the friction force limit, slip occurs and the tangential displacement needs to be readjusted; The friction coefficient can be obtained from the recommended values in the literature or theoretical models. For example: for quartz sand, μ∈[0.3,0.6], for steel particles: μ∈[0.1,0.2], for soil particles μ∈[0.4,0.8]. If there is no experimental data and literature reference, it can be estimated according to the properties of the particle material. For example: for rough particles, the friction coefficient is relatively high, usually μ>0.4, for smooth particles, the friction coefficient is relatively low, usually μ≈[0.1,0.3];

[0088] The formula for obtaining the normal stiffness between particles is:

[0089]

[0090] Among them, E * is the equivalent elastic modulus, R * ij is the equivalent contact radius of particle i and particle j. The equivalent contact radius can be dynamically calculated from the radii of particle i and particle j in the discrete element method model;

[0091]

[0092] Among them, E i is the elastic modulus of particle i, and E j is the elastic modulus of particle j, and v i is the Poisson's ratio of particle i, and v j is the Poisson's ratio of particle j;

[0093]

[0094] Among them, R i is the radius of the contacting particle i, and R j is the radius of the contacting particle j;

[0095] The tangential stiffness is calculated using an empirical ratio:

[0096] k t,ij = η·k n,ij

[0097] Among them, η is the proportionality coefficient, and usually η = 1 / 3;

[0098] The macroscopic model uses the finite element method to describe the continuum mechanical behavior of the roadway. The macroscopic roadway structure model is constructed using the finite element method, and the formula for obtaining the stress matrix is:

[0099] σ A = D·∈

[0100] Among them, σ A represents the macroscopic stress matrix, D is the elastic modulus matrix of the material, and ∈ is the strain matrix;

[0101] The formula for calculating the total stress in the macroscopic model is:

[0102]

[0103] Among them, σ macro represents the total stress in the macroscopic model, V represents the volume of the macroscopic model, F op represents the contact force between particles o and p in the microscopic model, n represents the total number of particles in the macroscopic model, and o and p are the indices of the particles, and o, p ∈ [1, n], o ≠ p;

[0104] The microscopic model and the macroscopic model are coupled, and the formula for feeding back the contact force in the microscopic model to the macroscopic model is:

[0105]

[0106] Among them, σ macro ' represents the total stress after coupling the microscopic model, kn,ab is the normal stiffness of particles a and b, k t,ab is the tangential stiffness of particles a and b, δ n,ab is the normal displacement of particles a and b, δ t,ab is the tangential displacement of particles a and b, N represents the total number of particles in the microscopic model, a and b are the indices of the particles, a, b ∈ [1, N], a ≠ b;

[0107] The formula for feeding back the macroscopic stress into the microscopic model to correct the contact force between particles is as follows:

[0108] F ij ′ = k n,ij ·δ n,ij + k t,ij ·δ t,ij + σ macro ·A ij

[0109] where, F ij ′ represents the contact force between particles i and j after coupling the macroscopic model, A ij is the contact area between particles i and j;

[0110] For spherical contact areas, they are calculated through the contact radius, and the formula is:

[0111] A ij = π|R * ij ·δ n,ij |

[0112] where, R * ii is the equivalent contact radius of particles i and j, and the equivalent contact radius can be dynamically calculated from the radii of particles i and j in the discrete element method model.

[0113] Analyze the stress, displacement, and material parameters in historical monitoring data, and combine the discrete element method and the finite element method to construct microscopic and macroscopic models, which can effectively achieve an in-depth understanding of the mechanical behavior inside the roadway. The microscopic model details the contact forces between particles through the discrete element method, thus capturing subtle mechanical interactions, while the macroscopic model provides the stress distribution of the overall structure through the finite element method. This multi-level modeling approach not only improves the accuracy of the model but also more realistically reflects the response of the roadway under different conditions. Secondly, by coupling the microscopic model and the macroscopic model, dynamic feedback between the two is achieved, and the contact force information of microscopic particles can be effectively integrated into the macroscopic stress analysis. This coupling method allows the mechanical behaviors at different scales to influence each other, enhancing the comprehensiveness and reliability of the model, thus providing a more solid foundation for risk assessment. For example, changes in microscopic parameters such as normal stiffness and tangential stiffness can directly affect the calculation of macroscopic stress, and vice versa, making the model more adaptable to complex and changing geological conditions.

[0114] Step 3: Use the expert scoring method to evaluate the state of the historical monitoring data of the monitoring points and the corresponding total contact force stress data at past historical moments, and assign corresponding state markers. Based on the random forest algorithm, construct a risk prediction model, and use the historical monitoring data and the corresponding contact force and total stress data as inputs, and the state markers as labels to train the risk prediction model. The state markers include the normal state and the abnormal state;

[0115] In this embodiment, the specific logic for constructing the risk prediction model is as follows:

[0116] The historical monitoring data of the monitoring points at previous historical moments, as well as the corresponding contact force and total stress data, are evaluated for their status using the expert scoring method, and labels of normal status or abnormal status are assigned. Among them, the normal status is represented by a risk value B = 0, and the abnormal status is represented by a risk value B = 1. The historical monitoring data of the monitoring points at previous historical moments, as well as the corresponding contact force and total stress data, are normalized and then integrated into a sample data set. The sample data set is divided into a training set and a test set, with a division ratio of 7:3. A risk prediction model is constructed based on the random forest algorithm. Using the historical monitoring data, contact force, and total stress data in the training set as inputs, and the corresponding status labels as tags, the risk prediction model is trained. Then, the historical monitoring data, contact force, and total stress in the test set are input to evaluate the performance of the model. The evaluation indicators include accuracy, precision, and recall. If, in five iterations, all the change amplitudes of the evaluation indicators are less than 0.01, it is determined that the model training is completed. By using the expert scoring method to evaluate the status of the historical monitoring data of the monitoring points at previous historical moments and the corresponding contact force and total stress data, the experience and knowledge of domain experts can be effectively combined, and a label of normal status or abnormal status can be assigned to each monitoring point. This process ensures the objectivity and scientificity of the status label, making subsequent risk assessment and prediction more credible. Outliers are removed from the historical monitoring data and filled with the mean value, and Z-score is used for standardization processing, which can effectively improve the data quality. This preprocessing ensures the stability and consistency of the input data set, lays a solid foundation for subsequent model training, and reduces the impact of noise on the model performance. A risk prediction model is constructed based on the random forest algorithm and trained using the stress, displacement, contact force, and total stress data in the training set, which can fully explore the complex relationships between the data. As an ensemble learning method, the random forest has strong robustness and high prediction accuracy, is suitable for processing multi-dimensional data with non-linear characteristics, and can effectively improve the accuracy of risk prediction.

[0117] Step 4: Input the monitoring data, contact force, and total stress data of each monitoring point at the current moment into the risk prediction model to obtain the status labels of each monitoring point at the current moment;

[0118] In this embodiment, the specific logic for obtaining the status predicted by the model at the current moment is as follows:

[0119] The monitoring data, contact force, and total stress data of each monitoring point at the current moment are normalized together with the historical data, and then the monitoring data, contact force, and total stress data of each monitoring point at the current moment are input into the trained risk prediction model to obtain the risk value predicted by the model. If the risk value B = 1, it is an abnormal status; if the risk value B = 0, it is a normal status, and the abnormal status monitoring points are marked.

[0120] Step 5: Analyze the monitoring points marked as abnormal states, calculate the floor heave risk index of the monitoring point at the current moment, and compare it with the preset risk threshold to determine whether to issue a warning;

[0121] In this embodiment, the specific logic for calculating the floor heave risk index at the abnormal state monitoring point at the current moment is as follows:

[0122] Obtain the stress, displacement, pressure, vibration, temperature, and humidity data at the monitoring point marked as abnormal state. The formula for obtaining the floor heave risk index is:

[0123]

[0124] Where, R is the floor heave risk index, P is the pressure, E is the stress, D is the displacement, V is the vibration amplitude, T is the temperature, is the ideal temperature value, which can be set to 20 °C, H is the humidity, is the ideal humidity value, which can be set to 50%, ω 1 、ω 2 、ω 3 、ω 4 、ω 5 are the weight coefficients of their respective terms, 0 < ω 1 < ω 5 < ω 4 < ω 2 < ω 3 < 1, and ω 1 + ω 2 + ω 3 + ω 4 + ω 5 = 1; The method for obtaining the vibration amplitude at the current moment is: Obtain the vibration signal of the monitoring point in the previous time period before the current moment, and process it to obtain the vibration amplitude within this time period, and use this vibration amplitude as the vibration amplitude at the current moment;

[0125] Compare the floor heave risk index at the current moment with the preset risk threshold. If the floor heave risk index at the current moment is lower than the preset risk threshold, continue to monitor. If the floor heave risk index at the current moment is higher than the preset risk threshold, the corresponding monitoring point is abnormal and a danger warning is issued.

[0126] The logarithmic function is used to process the pressure P data because the change in pressure usually has non-linear characteristics. In the early stage of pressure, it has a proportional relationship with the floor heave risk index. However, when the pressure reaches a certain level, its impact on the increase in risk will weaken. Therefore, using the logarithmic function can effectively balance this non-linear relationship. Although pressure has an impact on the floor heave risk, its impact is relatively linear, and in some cases (such as above a certain threshold), its impact degree may weaken. Using the logarithmic function to process the change in pressure further reduces its weight, so its weight is the smallest. Stress reflects the internal force state generated within a material or structure due to external actions and is usually related to the strength and load-bearing capacity of the material. Displacement represents the degree of deformation of the structure due to the action of stress. Excessive displacement may mean damage or instability of the structure; the square of the stress (E 2)It is emphasized that the influence of stress on the risk index is non - linear. Higher stress will significantly affect the risk, which is consistent with the yield strength and fatigue limit theories in materials science. Excessive stress may lead to material failure, so squaring it helps to emphasize the risk in the case of high stress. Dividing the square of stress by displacement (D + 1) helps to consider the moderating effect of displacement on risk. The larger the displacement, it may mean that the structure has undergone a certain degree of deformation. As the displacement increases, the load - bearing capacity of the material usually decreases. Therefore, when calculating the risk, the greater the displacement, the relatively smaller the influence on stress. The ratio of stress to displacement reflects the stiffness of the material or structure. The greater the stiffness, the more stable it is, and the smaller the floor heave risk index, that is, it is inversely proportional to the floor heave risk index. The influence of humidity can be represented by a linear difference because the influence of humidity on the material is relatively simple. Higher humidity will increase the risk of floor heave. Humidity affects the floor heave risk, but usually does not increase the risk as sharply as other factors (such as vibration, stress), so its weight is only higher than that of pressure. The influence of temperature on the floor heave risk is usually non - linear. High - temperature environments may lead to material fatigue and degradation. Therefore, using the squared difference to reflect the influence of temperature deviation from the reference value can reasonably quantify this risk. The influence of temperature on the floor heave risk cannot be ignored, but its weight is higher than that of humidity because temperature changes will have a significant impact on the strength and stability of materials, especially under extreme temperature conditions. The influence of temperature on risk is non - linear, so a relatively high weight is set. The squared form of stress E reflects the significant influence of stress on the floor heave risk, especially in the case of high stress, where the risk will rise sharply. Therefore, its weight is only lower than that of vibration, reflecting the key role of stress in risk assessment. At the same time, in order to consider the influence of displacement D, it is normalized in the form of displacement plus 1 to ensure that the influence on risk will not be overly amplified when the displacement increases. The squared form of vibration V is used to show the linear relationship between the vibration level and the floor heave risk, that is, the greater the vibration intensity, the higher the risk. The squared relationship enhances the risk judgment in the case of high vibration. Vibration will directly affect the stability of the roadway structure, especially during the mining process. There is a strong correlation between the frequency and intensity of vibration and the floor heave risk. Therefore, a higher weight is set to reflect its direct influence on risk.

[0127] Please refer to Figure 2 , the present invention further provides an analysis system for the mechanism of roadway floor heave. The analysis system for the mechanism of roadway floor heave is used to implement the above - mentioned analysis method for the mechanism of roadway floor heave, and includes:

[0128] A data acquisition module, which is used to arrange a plurality of monitoring points in the roadway to collect the monitoring data, pressure, vibration, temperature and humidity data of the monitoring points at the current moment, and obtain the historical monitoring data of each monitoring point at past historical moments. The monitoring data includes stress, displacement and material parameters;

[0129] A multi-scale model construction module, which is used to construct a microscopic model and a macroscopic model of the roadway based on monitoring data, and couple the microscopic and macroscopic models to obtain the contact force and total stress data of the particles in the roadway under the corresponding monitoring data;

[0130] A risk prediction model construction module, which is used to adopt the expert scoring method to evaluate the state of the historical monitoring data of the monitoring points and the corresponding contact force and total stress data at past historical moments, and assign corresponding state marks. Based on the random forest algorithm, a risk prediction model is constructed, and the historical monitoring data and the corresponding contact force and total stress data are used as inputs, and the state marks are used as labels to train the risk prediction model. The state marks include normal state and abnormal state;

[0131] A real-time state prediction module, which is used to input the monitoring data, contact force and total stress data of each monitoring point at the current moment into the risk prediction model to obtain the state marks of each monitoring point at the current moment;

[0132] An abnormal state analysis module, which is used to analyze the monitoring points marked as abnormal states, calculate the floor heave risk index of the monitoring point at the current moment, and compare it with a preset risk threshold to determine whether to issue a warning.

[0133] All the above formulas are dimensionless and take their numerical values for calculation. The formula is obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0134] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by the combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0135] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units. They may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0136] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all of them should be covered by the protection scope of this application.

Claims

1. A method for analyzing the mechanism of tunnel floor heave, characterized in that: The specific steps include: Step 1: Arrange multiple monitoring points in the tunnel to collect monitoring data, pressure, vibration, temperature and humidity data of the monitoring points at the current moment, and obtain historical monitoring data of each monitoring point at previous historical moments, wherein the monitoring data includes stress, displacement and material parameters; Step 2: Construct a microscopic model and a macroscopic model of the tunnel based on the monitoring data, and couple the microscopic model with the macroscopic model to obtain the contact force and total stress data of the particles in the tunnel under the corresponding monitoring data; Step 3: Using the expert scoring method, the historical monitoring data of the monitoring points at previous historical moments and the corresponding contact force and total stress data are evaluated, and the corresponding state marks are assigned. A risk prediction model is constructed based on the random forest algorithm, and the historical monitoring data and the corresponding contact force and total stress data are used as inputs, and the state marks are used as labels to train the risk prediction model, and the state marks include normal state and abnormal state; Step 4: Input the monitoring data, contact force and total stress data of each monitoring point at the current moment into the risk prediction model to obtain the status mark of each monitoring point at the current moment; Step 5: Analyze the monitoring points marked as abnormal, calculate the bottom drum risk index of the monitoring point at the current moment, and compare it with the preset risk threshold to determine whether to issue an early warning.

2. The method for analyzing the floor heave mechanism of a roadway according to claim 1, characterized in that: The specific logic for collecting data at the tunnel monitoring points is: Starting from the starting point of the tunnel, a monitoring point is set up every 3 meters. Stress sensors, displacement sensors, pressure sensors, vibration sensors, temperature sensors and humidity sensors are arranged at the monitoring points to collect stress, displacement, pressure, vibration, temperature and humidity data at the monitoring points. The material parameters include normal stiffness, tangential stiffness, elastic modulus and Poisson's ratio of the material.

3. The method for analyzing the floor heave mechanism of a roadway according to claim 2, characterized in that: The specific logic for constructing the micro-model and macro-model of the tunnel is as follows: The microscopic soil model is constructed using the discrete element method. The contact force between particles in the model is expressed as follows: F ij =F n,ij +F t,ij =k n,ij ·d n,ij +k t,ij ·d t,ij Among them, F ij represents the contact force between particles i and j, F n,ij represents the normal force between particles i and j, F t,ij represents the tangential force between particles i and j, k n,ij is the normal stiffness of particles i and j, k t,ij is the tangential stiffness of particles i and j, δ n,ij is the normal displacement of particles i and j, δ t,ij is the tangential displacement of particles i and j, i and j are the indices of particles; The normal displacement of a particle is determined by the overlap between the contact points of the two particles, and the calculation formula is: δ n,ij =R i +R j -d ij ,d ij <R i +R j Among them, R i is the radius of particle i, R j is the radius of particle j, d ij is the distance between the centers of particles i and j, when d ij ≥R i +R j When there is no contact between particles, the normal displacement δ n,ij =0; the tangential displacement is the relative displacement within the contact surface, which can be directly calculated and obtained in the discrete element method model, and in the discrete element method model, the tangential force needs to satisfy the Coulomb friction condition: F t ≤μ·F n Where μ is the friction coefficient. If the tangential force F t If the friction limit is exceeded, slippage occurs and the tangential displacement needs to be readjusted; The formula for obtaining the normal stiffness between particles is: Among them, E * is the equivalent elastic modulus, R * ij is the equivalent contact radius between particles i and j. The equivalent contact radius is dynamically calculated from the radius of particles i and j in the discrete element method model. Among them, E i is the elastic modulus of particle i, E j is the elastic modulus of particle j, ν i is the Poisson's ratio of particle i, v j is the Poisson’s ratio of particle j; Among them, R i is the radius of contact particle i, R j is the radius of contact particle j; The tangential stiffness is calculated using the empirical ratio: k t,ij =η·k n,ij Where η is the proportionality coefficient; The finite element method is used to construct a macroscopic tunnel structure model, and the formula for obtaining the stress matrix is: s A =D·∈ Among them, σ A It is expressed as a macro stress matrix, D is the elastic modulus matrix of the material, and ∈ is the strain matrix; The total stress in the macromodel is calculated based on the formula: Among them, σ macro represents the total stress in the macro model, V represents the volume of the macro model, and F op represents the contact force between particles o and p in the microscopic model, n represents the total number of particles in the macroscopic model, o and p are the indexes of the particles, and o, p∈[1,n], o≠p; The micro-model and the macro-model are coupled, and the contact force in the micro-model is fed back to the macro-model based on the formula: Among them, σ macro ′ represents the total stress after coupling the microscopic model, k n,ab is the normal stiffness of particles a and b, k t,ab is the tangential stiffness of particles a and b, δ n,ab is the normal displacement of particles a and b, δ t,ab is the tangential displacement of particles a and b, N represents the total number of particles in the microscopic model, a and b are the indices of the particles, a, b∈[1,N], a≠b; Feeding back the macroscopic stress to the microscopic model, the formula for correcting the contact force between particles is expressed as follows: F ij ′=k n,ij ·d n,ij +k t,ij ·d t,ij +s macro ·A ij Among them, F ij ′ represents the contact force between particles i and j after coupling the macro model, A ij is the contact area between particles i and j; For a spherical contact area, the contact radius is calculated using the formula: A ij =π|R * ij ·d n,ij | Among them, |R * ij ·δ n,ij | represents the absolute value of the product.

4. The method for analyzing the tunnel floor heave mechanism according to claim 3, characterized in that: The specific logic for building the risk prediction model is: The expert scoring method is used to evaluate the status of the historical monitoring data of the monitoring points at previous historical moments and the corresponding contact force and total stress data, and assign corresponding normal state or abnormal state labels, where the normal state is represented by a risk value B=0, and the abnormal state is represented by a risk value B=1. The historical monitoring data of the monitoring points at previous historical moments and the corresponding contact force and total stress data are normalized and integrated into a sample data set, and the sample data set is divided into a training set and a test set with a division ratio of 7:

3. A risk prediction model is constructed based on the random forest algorithm. The historical monitoring data, contact force and total stress data in the training set are used as input, and the corresponding state marks are used as labels to train the risk prediction model. The historical monitoring data, contact force and total stress in the test set are then input to evaluate the performance of the model. The evaluation indicators include accuracy, precision and recall. If all changes in the evaluation indicators are less than 0.01 in five iterations, the model training is considered to be completed.

5. The method for analyzing the floor heave mechanism of a tunnel according to claim 4, characterized in that: The specific logic for obtaining the model's prediction of the current state is as follows: The monitoring data, contact force and total stress data of each monitoring point at the current moment are normalized and input into the trained risk prediction model to obtain the risk value predicted by the model. If the risk value B=1, it is an abnormal state. If the risk value B=0, it is a normal state, and the abnormal state monitoring point is marked.

6. The method for analyzing the floor heave mechanism of a tunnel according to claim 5, characterized in that: The specific logic for calculating the bottom drum risk index at the abnormal state monitoring point at the current moment is: The stress, displacement, pressure, vibration, temperature and humidity data at the monitoring points marked as abnormal states are obtained, and the formula for obtaining the bottom drum risk index is as follows: Among them, R is the bottom drum risk index, P is pressure, E is stress, D is displacement, V is vibration amplitude, T is temperature, is the ideal temperature, H is the humidity, is the ideal value of humidity, ω1, ω2, ω3, ω4, and ω5 are the weight coefficients of their respective items, 0<ω1<ω5<ω4<ω2<ω3<1, and ω1+ω2+ω3+ω4+ω5=1; Compare the current floor drum risk index with the preset risk threshold. If the current floor drum risk index is lower than the preset risk threshold, continue monitoring. If the current floor drum risk index is higher than the preset risk threshold, the corresponding monitoring point has an abnormality and a danger warning is issued.

7. An analysis system for the mechanism of tunnel floor heave, characterized in that: The tunnel floor heave mechanism analysis system is used to implement the tunnel floor heave mechanism analysis method according to any one of claims 1 to 6, comprising: A data acquisition module is used to arrange multiple monitoring points in the tunnel to collect monitoring data, pressure, vibration, temperature and humidity data of the monitoring points at the current moment, and obtain historical monitoring data of each monitoring point at previous historical moments, wherein the monitoring data includes stress, displacement and material parameters; The multi-scale model building module is used to build the micro-model and macro-model of the roadway based on the monitoring data, and couple the micro-model with the macro-model to obtain the contact force and total stress data of the particles in the roadway under the corresponding monitoring data; The risk prediction model construction module is used to use the expert scoring method to evaluate the status of the historical monitoring data of the monitoring points at previous historical moments and the corresponding contact force and total stress data, and assign corresponding status tags, build a risk prediction model based on the random forest algorithm, and use the historical monitoring data and the corresponding contact force and total stress data as inputs, and the status tags as labels to train the risk prediction model, wherein the status tags include normal status and abnormal status; The real-time state prediction module is used to input the monitoring data, contact force and total stress data of each monitoring point at the current moment into the risk prediction model to obtain the state mark of each monitoring point at the current moment; The abnormal state analysis module is used to analyze the monitoring point marked as abnormal state, calculate the bottom drum risk index of the monitoring point at the current moment, and compare it with the preset risk threshold to determine whether to issue an early warning.

Images