Multi-sensor fusion tunnel smooth blasting monitoring system

Abstract

The invention discloses a multi-sensor fusion tunnel smooth blasting monitoring system, and belongs to the technical field of artificial intelligence. The method comprises the following steps: preprocessing and integrating monitored data, constructing a tunnel digital model, and applying a static load and a dynamic load; obtaining a safety critical value of each parameter in the tunnel digital model to obtain a tunnel critical value digital model, extracting each parameter exceeding the safety critical value, generating a tunnel abnormal parameter digital model, and identifying an abnormal mode of smooth blasting by using a random forest algorithm; when the abnormal mode of smooth blasting is related to the interior of the tunnel, reconstructing a tunnel digital model; when the optimization suggestion is related to smooth blasting, new blasting parameters are obtained by using a support vector machine model in combination with expert suggestions; a tunnel-smooth blasting correlation model is constructed, correlation rules are identified, and the influence degree of blasting parameters on tunnel monitoring parameters is evaluated; and searching an optimal blasting parameter combination by applying particle swarm optimization to meet a set optimization target.

Description

Technical Field

[0001] The invention relates to the field of artificial intelligence technology, and in particular to a multi-sensor fusion tunnel smooth surface blasting monitoring system. Background Art

[0002] During tunnel construction, smooth blasting is a commonly used controlled blasting technology. By reasonably controlling blasting parameters, the tunnel wall can be neatly shaped and construction damage can be reduced. However, the complexity and uncertainty of smooth blasting often lead to safety hazards and quality problems in actual construction. Existing blasting monitoring methods mainly rely on a single sensor, which makes it difficult to obtain comprehensive information on the status of the tunnel during the blasting process, especially the inability to identify and correct abnormal blasting in a timely manner, which can easily lead to damage to the tunnel structure and affect construction efficiency and safety.

[0003] Existing blasting parameter tuning methods are mostly based on empirical formulas, lacking feedback tuning combined with real-time tunnel monitoring data, and unable to dynamically optimize specific construction environments and blasting abnormalities. Existing monitoring systems are difficult to achieve early recognition of blasting abnormal patterns, and are unable to adjust blasting parameters in a timely manner, increasing construction risks. In addition, the assessment of the impact of blasting on tunnel structure safety is mostly conducted after the fact, and preventive intervention is not possible. Existing technologies find it difficult to achieve an organic combination of tunnel status and blasting parameters through intelligent algorithms, and it is difficult to achieve efficient blasting parameter optimization, and fail to fully utilize artificial intelligence technology to intelligently adjust blasting parameters. Summary of the invention

[0004] The purpose of the present invention is to provide a multi-sensor fusion tunnel smooth surface blasting monitoring system to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A multi-sensor fusion tunnel smooth surface blasting monitoring system, comprising:

[0007] Tunnel digital model construction module: including: data acquisition unit and tunnel digital model construction unit; the data acquisition unit uses multiple sensors to obtain the stress, deformation and displacement of the tunnel wall, the vibration acceleration, frequency and duration in the tunnel, the gas composition and corresponding concentration in the tunnel, the temperature and humidity changes in the tunnel, and the background noise and changes; the tunnel digital model construction unit pre-processes and integrates the monitored data, uses the finite element model to build a tunnel digital model, applies static loads and dynamic loads, selects a solver according to the analysis target, and performs static and dynamic analysis;

[0008] Smooth surface blasting abnormal pattern acquisition module: including: critical value determination unit, abnormal parameter extraction unit and smooth surface blasting abnormal pattern recognition unit; wherein the critical value determination unit obtains the safety critical value of each parameter in the tunnel digital model to obtain the tunnel critical value digital model; the abnormal parameter extraction unit compares the tunnel critical value digital model with the tunnel digital model, extracts each parameter exceeding the safety critical value, and generates the tunnel abnormal parameter digital model; the smooth surface blasting abnormal pattern recognition unit extracts features from the tunnel abnormal parameter digital model, selects the random forest algorithm to analyze the features, uses historical data for training, and recognizes the abnormal pattern of smooth surface blasting;

[0009] Digital model and blasting parameter update module: including: internal repair suggestion unit and blasting parameter optimization unit; among them, when the abnormal pattern of smooth surface blasting is related to the interior of the tunnel, the internal repair suggestion unit will send the optimization suggestion to the staff to repair the interior of the tunnel, and rebuild the digital model of the tunnel after the tunnel is repaired; when the optimization suggestion is related to smooth surface blasting, the blasting parameter optimization unit obtains the blasting parameters of this smooth surface blasting, and uses the support vector machine model combined with expert suggestions to obtain new blasting parameters;

[0010] The optimal blasting parameter combination generation module includes: an association model construction unit, an association rule identification unit, an influence degree evaluation unit and an optimal parameter search unit; wherein, the association model construction unit associates the blasting parameters with the tunnel digital model to construct a tunnel-smooth surface blasting association model, the association rule identification unit uses correlation analysis to identify the association rules between each blasting parameter and the tunnel monitoring parameter, and the influence degree evaluation unit quantitatively evaluates the influence of the blasting parameters on the tunnel monitoring parameters through regression analysis; the optimal parameter search unit inputs the safety and efficiency indicators of the blasting operation, applies particle swarm optimization, and searches for the optimal blasting parameter combination based on the results of the tunnel-smooth surface blasting association model to meet the set optimization goals.

[0011] In the tunnel digital model construction module, stress sensors are used to measure the stress distribution of the tunnel wall, displacement sensors are used to monitor the displacement and deformation of the tunnel wall, vibration sensors are used to monitor the vibration acceleration, frequency and duration in the tunnel, gas sensors are used to monitor the gas composition and corresponding concentration in the tunnel, temperature and humidity sensors are used to monitor the temperature and humidity changes in the tunnel, and noise sensors are used to monitor background noise and changes; low-pass filters are applied to remove noise from the data, statistical methods are used to detect and eliminate outliers, data from different sensors are normalized, and all data are synchronized by timestamp.

[0012] In the tunnel digital model building module, the geometric model of the tunnel is created using ANSYS finite element analysis software, and the geometric parameters of the tunnel are input to form a three-dimensional model of the tunnel; the material properties of the tunnel and the physical and mechanical properties of the surrounding soil are input to define the material properties; the hexahedral grid is selected as the grid type for grid division; the contact surface between the tunnel wall and the soil is set as a fixed boundary to simulate the influence of soil pressure on the tunnel, and the displacement constraint conditions of the tunnel are set;

[0013] Apply static loads, including traffic loads, soil pressure and groundwater pressure in the tunnel; apply dynamic loads, combine the influence of dynamic loads of smooth blasting, input the blasting wave propagation model, and set the corresponding instantaneous pressure or vibration load; select the solver according to the analysis objectives, perform static or dynamic analysis, set solution parameters, monitor the calculation progress and convergence, and check the rationality of the calculation results during the solution process; generate stress distribution, displacement field and deformation diagrams of the tunnel wall, compare the model output with the monitoring data of the on-site sensors, and evaluate the sensitivity of the model to changes in different input parameters.

[0014] The choice of solver depends mainly on the type and goal of the analysis. Tunnel analysis involves static and dynamic problems, so the corresponding solver should be selected:

[0015] Static analysis is used to analyze the long-term response of a tunnel under constant loads, such as traffic loads, soil pressure, and groundwater pressure. In this case, a static solver is selected to calculate the displacement, stress, and deformation of the tunnel. Static solvers do not consider the effect of time on the results and are suitable for handling loads that are gradually applied or maintained constant. Among them, the linear static analysis solver is used for linear stress analysis, assuming that there is a linear relationship between the deformation and load of the material. It is suitable for small deformations and initial stability analysis of tunnels. Nonlinear static analysis solvers are used to handle situations with material nonlinearity, geometric nonlinearity, and contact nonlinearity. For example, when the tunnel material undergoes plastic deformation under ultimate stress, a nonlinear solver is required.

[0016] Dynamic analysis is used to simulate the effects of dynamic loads such as smooth blasting. Dynamic analysis takes into account time and inertia effects and evaluates the impact of blast wave propagation on the transient pressure and vibration of the tunnel. The transient dynamics solver is suitable for simulating the propagation of transient pressure waves generated by smooth blasting. It can capture the vibration response in the tunnel and the transient deformation and stress changes caused by the blast wave. The modal analysis solver is used to determine the natural frequencies and vibration modes of the tunnel structure to evaluate the dynamic response of the tunnel under blast loads. The frequency response solver is used to analyze the vibration response of the tunnel at specific frequencies, especially for studying the frequency-dependent effects of blast shock.

[0017] When performing finite element analysis, checking the rationality of the calculation results is an important step to ensure the accuracy and credibility of the model. The rationality check is performed through the following methods:

[0018] During the analysis, monitor the convergence of the solution. If the solution does not converge, that is, the solution keeps fluctuating or cannot reach a stable state, it means that the mesh is too coarse, the boundary conditions are not set properly, or the solution parameters are unreasonable. Therefore, check whether the displacement, stress or other key responses gradually stabilize with the increase in the number of iterations, and adjust the solution parameters, such as relaxation factor, time step, etc., to ensure convergence.

[0019] Compare the finite element analysis results with theoretical calculations or code standards, such as the stress or displacement distribution of linear elastic materials under known loads, to confirm whether the model is reasonable. If the results deviate too much, you may need to recheck the boundary conditions, load settings, and material properties.

[0020] Compare the stress distribution, displacement field and deformation diagrams obtained by finite element analysis with the monitoring data of sensors in the tunnel. For example, monitor the consistency of the stress and deformation of the tunnel wall with the model prediction value. If there is a large deviation, it is necessary to re-evaluate the material properties or the load application method.

[0021] Check whether there are any unreasonable phenomena such as abnormal stress concentration or displacement mutation in the model, especially near the contact surface and boundary conditions. If the local stress or displacement is significantly beyond expectations, it is necessary to re-evaluate the rationality of the contact conditions, meshing or boundary constraints.

[0022] Sensitivity analysis is used to evaluate the response of the model to different input parameters to ensure the adaptability of the model to the actual construction environment. The following are the steps of sensitivity analysis:

[0023] Select key input parameters, such as the elastic modulus of the material, charge density of smooth blasting, load, etc., and define the range of these parameters according to the actual working conditions and material uncertainties. For example, gradually adjust the elastic modulus, density and other parameters within the range of ±10%.

[0024] While keeping other parameters unchanged, adjust key parameters one by one to observe their effects on tunnel stress, displacement, deformation and other output results. For example, in the blast wave propagation model, change the propagation speed of the blast wave or the vibration load to observe the stress distribution or displacement changes in the tunnel.

[0025] The rate of change of the model output to the input parameters is calculated to quantify the sensitivity of the model to each parameter. The sensitivity indicators used include relative rate of change and elasticity coefficient.

[0026] The changes in each input parameter and the corresponding output changes are plotted into sensitivity curves to intuitively show which parameters have a greater impact on the model results. For example, if the change in the vibration frequency of the blast wave has a greater impact on the displacement of the tunnel wall, then this frequency will be a key optimization parameter.

[0027] Establish a blast wave propagation model, including establishing a physical model and establishing a mathematical model; determine the blasting materials used and their characteristics, and obtain the blasting parameters of this smooth blasting, including the uncoupling coefficient, the smooth eye spacing, the minimum resistance line, the blast hole proximity coefficient, the line charge density and the detonation interval time; wherein, the physical model is established by selecting a nonlinear elastic model, and a geometric model of the blasting source and the surrounding medium is created, and the surrounding medium includes the tunnel wall and the soil; the wave equation and the pressure decay model are used when establishing the mathematical model;

[0028] According to the propagation of pressure waves generated by the explosion, the wave equation is used to describe the wave characteristics:

[0029]

[0030] Among them, u is the wave displacement, c is the wave speed, and t is the time;

[0031] Using the instantaneous high-voltage pulse generated by the explosion of explosives, the following pressure decay model P(t) is used:

[0032]

[0033] Where P0 is the initial pressure at the time of explosion, α is the first empirical constant, n is the second empirical constant, and r is the distance from the explosion center;

[0034] Instantaneous pressure boundary conditions are applied at the blasting center to simulate the pressure impact at the moment of explosion, and the contact conditions between the tunnel wall and the soil are set to ensure the interaction during the propagation of the pressure wave. The grid density is set at the blasting source to capture the instantaneous pressure changes caused by the explosion. An implicit solver is selected to process dynamic loads. The time step and solution accuracy are set to capture the dynamic response at the moment of explosion. The distribution diagram and time history curve of the pressure wave after the blasting are generated to obtain the propagation of the pressure wave in the tunnel. The simulation results are compared with the actual monitoring data to verify the blasting wave propagation model.

[0035] In the smooth blasting abnormal pattern acquisition module, the safety critical value of each parameter is determined according to existing industry standards and research literature, and the determined safety critical values ​​are organized into a digital model to form a tunnel critical value digital model; the real-time monitoring data in the tunnel digital model is compared with the tunnel critical value digital model item by item, and the parameters exceeding the safety critical value are identified, and their specific values ​​and change trends are recorded; the various parameters exceeding the safety critical value are summarized to form a tunnel abnormal parameter digital model, and the parameters are classified according to their properties; the change trend of the abnormal parameters is analyzed to determine whether it is a single event or a continuous abnormality, and visualization tools are used to display the spatial distribution of abnormal parameters in the tunnel to identify high-risk areas; the abnormal parameters are compared with historical data to confirm whether they are truly abnormal.

[0036] In the smooth blasting abnormal pattern acquisition module, all parameters exceeding the safety critical value are extracted from the digital model of tunnel abnormal parameters to form a structured data set. The extracted abnormal parameters are annotated to distinguish between normal and abnormal states. Correlation analysis is used to extract the time series characteristics, change rate characteristics and frequency domain characteristics of the abnormal parameters.

[0037] The extracted features are divided into training set and test set. The training set data is used to train the random forest model. After the training is completed, the feature importance evaluation method provided by the random forest is used to identify the features that have the greatest impact on the abnormal pattern of smooth surface blasting. The test set is input into the trained random forest model to obtain the prediction result of each sample, identify the abnormal pattern of smooth surface blasting, form an abnormal pattern output list, and use accuracy, recall rate and F1-score to evaluate the performance of the model.

[0038] In the digital model and blasting parameter update module, based on the abnormal pattern of smooth blasting, combined with expert knowledge and historical data, a specific repair plan is formulated, and the optimization suggestions are automatically sent to the on-site staff through the system. The staff will carry out the repair work inside the tunnel according to the optimization suggestions to ensure that the tunnel structure is restored to a safe state. During the repair process, the parameters inside the tunnel are monitored through multiple sensors, and data is recorded to evaluate the repair effect.

[0039] The new data collected during the repair process are sorted and cleaned, and the digital model of the tunnel is regenerated according to the updated parameters. The reconstructed digital model of the tunnel is compared with the actual monitoring data, and the impact of different parameter changes on the model results is evaluated to ensure the robustness of the model.

[0040] In the digital model and blasting parameter update module, when the optimization suggestion is related to smooth surface blasting, the blasting parameters of this smooth surface blasting are obtained, and data related to historical blasting operations are collected, including: cases of successful and failed blasting, specific blasting parameters used in the cases, and corresponding tunnel monitoring parameters; data preprocessing is performed to extract the historical values ​​of each blasting parameter, changes in tunnel monitoring parameters after blasting, and environmental conditions of blasting operations, and divide the training set and test set;

[0041] Using the grid search method, the radial basis function is selected as the SVM kernel function, the hyperparameters are selected, and the SVM model is trained using the training set to learn the mapping relationship between different blasting parameters and tunnel monitoring parameters; the model performance is evaluated on the test set; the output of the support vector machine model is combined with expert suggestions for comprehensive analysis to obtain new blasting parameters.

[0042] In the optimal blasting parameter combination generation module, the blasting parameters and the corresponding tunnel monitoring parameters are integrated into a data set to ensure that each set of blasting parameters corresponds to the subsequent monitoring results; the Pearson correlation coefficient is used to calculate the correlation between each blasting parameter and the tunnel monitoring parameter, generate a correlation matrix, and identify positive and negative correlations; the association rule learning algorithm Apriori is used to mine the relationship between blasting parameters and tunnel monitoring parameters and identify potential association rules;

[0043] A linear regression model was selected for modeling. The tunnel monitoring parameters were used as dependent variables and the blasting parameters were used as independent variables. The data were divided into a training set and a test set. The regression model was trained using the training set and verified on the test set to evaluate the model's fit and prediction ability. The coefficients of the regression model were used to quantitatively evaluate the degree of influence of each blasting parameter on the tunnel monitoring parameters. The larger the absolute value of the coefficient, the greater the influence of the blasting parameter on the tunnel monitoring parameters.

[0044] In the optimal blasting parameter combination generation module, the safety index SI is constructed based on the tunnel monitoring parameters, and the efficiency index EI is constructed based on the blasting parameters. The objective function F is defined as:

[0045] F = w1*SI+w2*EI;

[0046] Wherein, w1 is the first weight coefficient, w2 is the second weight coefficient;

[0047] Set the number of particles to N, and each particle represents a set of blasting parameter combinations (x1, x2, ..., x m ), where m is the number of parameters of the explosion parameters; the initial position and velocity of each particle are randomly generated;

[0048] Calculate the fitness value F(x i), using the defined objective function, where i traverses from 1 to m respectively; save the best position p of each particle i and the global optimal position g; update the particle's velocity v using the following formula i and position x i :

[0049] v i (t+1)=w*v i (t)+c′1*r′1*(p i -x i (t))+c′2*r′2*(gx i (t));

[0050] x i (t+1)=x i (t)+v i (t+1);

[0051] Where w is the inertia weight, c′1 is the first acceleration constant, c′2 is the second acceleration constant, r′1 is the first random number, r′2 is the second random number, and the value range of r′1 and r′2 is [0, 1], v i (t+1) represents the speed at time t+1, x i (t+1) indicates the position at time t+1;

[0052] The maximum number of iterations and the convergence threshold are set. When the convergence condition reaches the convergence threshold or the maximum number of iterations is reached, the iteration is stopped. The global optimal position in the particle swarm is output, which is expressed as the optimal blasting parameter combination. The optimal parameters are applied to the actual blasting operation, and the various parameters of the tunnel are monitored to verify the optimization effect. According to the actual results, the model parameters and optimization objectives are further adjusted to perform model iteration.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] 1. The present invention combines the tunnel digital model with the association model of blasting parameters, uses artificial intelligence algorithms such as random forests and support vector machines to identify abnormal blasting patterns, and dynamically optimizes blasting parameters according to tunnel monitoring data to achieve real-time adjustment to ensure safe and efficient construction.

[0055] 2. The present invention compares the tunnel digital model with the safety critical value, extracts the parameters that exceed the safety critical value to generate an abnormal parameter model, and combines historical data for training and analysis. It can identify abnormal patterns of smooth blasting in advance, make optimization suggestions in time, and reduce the risk of tunnel structure damage.

[0056] 3. The present invention uses a particle swarm optimization algorithm to automatically search for the best blasting parameter combination based on the results of the tunnel-smooth surface blasting association model to ensure that the best blasting effect is achieved while meeting safety and efficiency indicators. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a system structure diagram of a multi-sensor fusion tunnel smooth surface blasting monitoring system of the present invention;

[0058] Figure 2 It is a schematic diagram of the steps for generating an optimal blasting parameter combination of a multi-sensor fusion tunnel smooth surface blasting monitoring system of the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution.

[0061] According to one embodiment of the present invention, Figure 1 As shown in the system structure diagram of a multi-sensor fusion tunnel smooth surface blasting monitoring system, a multi-sensor fusion tunnel smooth surface blasting monitoring system includes:

[0062] Tunnel digital model construction module: including: data acquisition unit and tunnel digital model construction unit; the data acquisition unit uses multiple sensors to obtain the stress, deformation and displacement of the tunnel wall, the vibration acceleration, frequency and duration in the tunnel, the gas composition and corresponding concentration in the tunnel, the temperature and humidity changes in the tunnel, and the background noise and changes; the tunnel digital model construction unit pre-processes and integrates the monitored data, uses the finite element model to build a tunnel digital model, applies static loads and dynamic loads, selects a solver according to the analysis target, and performs static and dynamic analysis;

[0063] Smooth surface blasting abnormal pattern acquisition module: including: critical value determination unit, abnormal parameter extraction unit and smooth surface blasting abnormal pattern recognition unit; wherein the critical value determination unit obtains the safety critical value of each parameter in the tunnel digital model to obtain the tunnel critical value digital model; the abnormal parameter extraction unit compares the tunnel critical value digital model with the tunnel digital model, extracts each parameter exceeding the safety critical value, and generates the tunnel abnormal parameter digital model; the smooth surface blasting abnormal pattern recognition unit extracts features from the tunnel abnormal parameter digital model, selects the random forest algorithm to analyze the features, uses historical data for training, and recognizes the abnormal pattern of smooth surface blasting;

[0064] Digital model and blasting parameter update module: including: internal repair suggestion unit and blasting parameter optimization unit; among them, when the abnormal pattern of smooth surface blasting is related to the interior of the tunnel, the internal repair suggestion unit will send the optimization suggestion to the staff to repair the interior of the tunnel, and rebuild the digital model of the tunnel after the tunnel is repaired; when the optimization suggestion is related to smooth surface blasting, the blasting parameter optimization unit obtains the blasting parameters of this smooth surface blasting, and uses the support vector machine model combined with expert suggestions to obtain new blasting parameters;

[0065] The optimal blasting parameter combination generation module includes: an association model construction unit, an association rule identification unit, an influence degree evaluation unit and an optimal parameter search unit; wherein, the association model construction unit associates the blasting parameters with the tunnel digital model to construct a tunnel-smooth surface blasting association model, the association rule identification unit uses correlation analysis to identify the association rules between each blasting parameter and the tunnel monitoring parameter, and the influence degree evaluation unit quantitatively evaluates the influence of the blasting parameters on the tunnel monitoring parameters through regression analysis; the optimal parameter search unit inputs the safety and efficiency indicators of the blasting operation, applies particle swarm optimization, and searches for the optimal blasting parameter combination based on the results of the tunnel-smooth surface blasting association model to meet the set optimization goals.

[0066] In the tunnel digital model construction module, stress sensors are used to measure the stress distribution of the tunnel wall, displacement sensors are used to monitor the displacement and deformation of the tunnel wall, vibration sensors are used to monitor the vibration acceleration, frequency and duration in the tunnel, gas sensors are used to monitor the gas composition and corresponding concentration in the tunnel, temperature and humidity sensors are used to monitor the temperature and humidity changes in the tunnel, and noise sensors are used to monitor background noise and changes; low-pass filters are applied to remove noise from the data, statistical methods are used to detect and eliminate outliers, data from different sensors are normalized, and all data are synchronized by timestamp.

[0067] In the tunnel digital model building module, the geometric model of the tunnel is created using ANSYS finite element analysis software, and the geometric parameters of the tunnel are input to form a three-dimensional model of the tunnel; the material properties of the tunnel and the physical and mechanical properties of the surrounding soil are input to define the material properties; the hexahedral grid is selected as the grid type for grid division; the contact surface between the tunnel wall and the soil is set as a fixed boundary to simulate the influence of soil pressure on the tunnel, and the displacement constraint conditions of the tunnel are set;

[0068] Apply static loads, including traffic loads, soil pressure and groundwater pressure in the tunnel; apply dynamic loads, combine the influence of dynamic loads of smooth blasting, input the blasting wave propagation model, and set the corresponding instantaneous pressure or vibration load; select the solver according to the analysis objectives, perform static or dynamic analysis, set solution parameters, monitor the calculation progress and convergence, and check the rationality of the calculation results during the solution process; generate stress distribution, displacement field and deformation diagrams of the tunnel wall, compare the model output with the monitoring data of the on-site sensors, and evaluate the sensitivity of the model to changes in different input parameters.

[0069] In this embodiment, the material properties of the tunnel wall, such as the elastic modulus, Poisson's ratio, density, etc. of concrete, and the physical and mechanical properties of the surrounding soil, such as the friction coefficient, cohesion, elastic modulus of soil, etc., are input to define the properties of each part of the material in the finite element model. The nonlinear elastic model is used to capture the dynamic response of the material during blasting. The contact surface between the tunnel wall and the soil is set as a fixed boundary to simulate the effect of soil pressure. Displacement constraints are applied to the boundaries of the tunnel structure to prevent unreasonable displacement changes. Static loads include traffic loads, soil pressure and groundwater pressure inside the tunnel. In terms of dynamic loads, instantaneous pressure or vibration loads generated by smooth blasting are applied to simulate the wave effect caused by the explosion.

[0070] Establish a blast wave propagation model, including establishing a physical model and establishing a mathematical model; determine the blasting materials used and their characteristics, and obtain the blasting parameters of this smooth blasting, including the uncoupling coefficient, the smooth eye spacing, the minimum resistance line, the blast hole proximity coefficient, the line charge density and the detonation interval time; wherein, the physical model is established by selecting a nonlinear elastic model to create a geometric model of the blasting source and the surrounding medium, the surrounding medium includes the tunnel wall and the soil; in this embodiment, a nonlinear elastic model is used to create a geometric model of the blasting source and its surrounding medium (tunnel wall and soil). Considering the mechanical properties of different materials, high-precision simulation of the blasting center is ensured by meshing.

[0071] The wave equation and pressure decay model were used to build the mathematical model;

[0072] According to the propagation of pressure waves generated by the explosion, the wave equation is used to describe the wave characteristics:

[0073]

[0074] Among them, u is the wave displacement, c is the wave speed, and t is the time;

[0075] Using the instantaneous high-voltage pulse generated by the explosion of explosives, the following pressure decay model P(t) is used:

[0076]

[0077] Where P0 is the initial pressure at the time of explosion, α is the first empirical constant, n is the second empirical constant, and r is the distance from the explosion center;

[0078] Instantaneous pressure boundary conditions are applied at the blasting center to simulate the pressure impact at the moment of explosion, and the contact conditions between the tunnel wall and the soil are set to ensure the interaction during the propagation of the pressure wave. The grid density is set at the blasting source to capture the instantaneous pressure changes caused by the explosion. An implicit solver is selected to process dynamic loads. The time step and solution accuracy are set to capture the dynamic response at the moment of explosion. The distribution diagram and time history curve of the pressure wave after the blasting are generated to obtain the propagation of the pressure wave in the tunnel. The simulation results are compared with the actual monitoring data to verify the blasting wave propagation model.

[0079] In the smooth blasting abnormal pattern acquisition module, the safety critical value of each parameter is determined according to existing industry standards and research literature, and the determined safety critical values ​​are organized into a digital model to form a tunnel critical value digital model; the real-time monitoring data in the tunnel digital model is compared with the tunnel critical value digital model item by item, and the parameters exceeding the safety critical value are identified, and their specific values ​​and change trends are recorded; the various parameters exceeding the safety critical value are summarized to form a tunnel abnormal parameter digital model, and the parameters are classified according to their properties; the change trend of the abnormal parameters is analyzed to determine whether it is a single event or a continuous abnormality, and visualization tools are used to display the spatial distribution of abnormal parameters in the tunnel to identify high-risk areas; the abnormal parameters are compared with historical data to confirm whether they are truly abnormal.

[0080] In this embodiment, the geometric parameters of a tunnel are as follows: tunnel diameter: 8 meters, tunnel length: 200 meters, tunnel wall thickness: 0.5 meters, tunnel wall material: concrete, elastic modulus is 30 GPa, Poisson's ratio is 0.2, density is 2400 kg / m 3 ; The blasting parameters are set as follows: initial pressure of blasting source: 10 MPa, distance from blasting point to tunnel wall: 2 meters, empirical constant: n = 2.5;

[0081] Through the pressure wave propagation model, the results show: Maximum stress: The maximum stress at the tunnel wall is 12MPa. Maximum displacement: The maximum displacement at the tunnel wall is 0.002 meters. Pressure wave attenuation: The pressure at 2 meters from the explosion center is 8MPa, and attenuates to 3MPa at 5 meters.

[0082] In the smooth blasting abnormal pattern acquisition module, all parameters exceeding the safety critical value are extracted from the digital model of tunnel abnormal parameters to form a structured data set. The extracted abnormal parameters are annotated to distinguish between normal and abnormal states. Correlation analysis is used to extract the time series characteristics, change rate characteristics and frequency domain characteristics of the abnormal parameters.

[0083] The extracted features are divided into a training set and a test set, and the training set data is used to train the random forest model. After the training is completed, the feature importance evaluation method provided by the random forest is used to identify the features that have the greatest impact on the abnormal pattern of the smooth surface blasting; the test set is input into the trained random forest model, the prediction results of each sample are obtained, the abnormal pattern of the smooth surface blasting is identified, and an abnormal pattern output list is formed. The performance of the model is evaluated using the accuracy, recall rate, and F1-score. In this embodiment, the extracted features are divided into a training set (70%) and a test set (30%).

[0084] In the digital model and blasting parameter update module, based on the abnormal pattern of smooth blasting, combined with expert knowledge and historical data, a specific repair plan is formulated, and the optimization suggestions are automatically sent to the on-site staff through the system. The staff will carry out the repair work inside the tunnel according to the optimization suggestions to ensure that the tunnel structure is restored to a safe state; during the repair process, the parameters inside the tunnel are monitored by multiple sensors, and data is recorded to evaluate the repair effect; in this embodiment, based on the abnormal pattern of smooth blasting, combined with expert knowledge and historical data, a specific repair plan is formulated. For example, according to the rate of change of stress and displacement, it is recommended to adjust the blasting charge and increase support measures. The optimization suggestions are automatically sent to the on-site staff through the system, and the on-site personnel carry out tunnel repair work according to the suggestions, such as reinforcing the tunnel wall, adjusting the blasting parameters, etc.

[0085] The new data collected during the repair process are sorted and cleaned, and the digital model of the tunnel is regenerated according to the updated parameters. The reconstructed digital model of the tunnel is compared with the actual monitoring data, and the impact of different parameter changes on the model results is evaluated to ensure the robustness of the model.

[0086] In the digital model and blasting parameter update module, when the optimization suggestion is related to smooth surface blasting, the blasting parameters of this smooth surface blasting are obtained, and data related to historical blasting operations are collected, including: cases of successful and failed blasting, specific blasting parameters used in the cases, and corresponding tunnel monitoring parameters; data preprocessing is performed to extract the historical values ​​of each blasting parameter, changes in tunnel monitoring parameters after blasting, and environmental conditions of blasting operations, and divide the training set and test set;

[0087] Using the grid search method, the radial basis function is selected as the SVM kernel function, the hyperparameters are selected, and the SVM model is trained using the training set to learn the mapping relationship between different blasting parameters and tunnel monitoring parameters; the model performance is evaluated on the test set; the output of the support vector machine model is combined with expert suggestions for comprehensive analysis to obtain new blasting parameters.

[0088] In the optimal blasting parameter combination generation module, the blasting parameters and the corresponding tunnel monitoring parameters are integrated into a data set to ensure that each set of blasting parameters corresponds to the subsequent monitoring results; the Pearson correlation coefficient is used to calculate the correlation between each blasting parameter and the tunnel monitoring parameter, generate a correlation matrix, and identify positive and negative correlations. In this embodiment, the correlation coefficient between the wire charge density and the tunnel wall stress is 0.82 (strong positive correlation), and the correlation coefficient between the detonation interval and the tunnel vibration acceleration is -0.68 (negative correlation); the association rule learning algorithm Apriori is used to mine the relationship between the blasting parameters and the tunnel monitoring parameters, and identify potential association rules; through association rule learning, it is found that when the wire charge density is greater than 0.7kg / m and the smooth eye spacing is less than 0.5m, the risk of exceeding the limit of the tunnel wall stress increases significantly (confidence level is 85%).

[0089] A linear regression model was selected for modeling. The tunnel monitoring parameters were used as dependent variables and the blasting parameters were used as independent variables. The data were divided into a training set and a test set. The regression model was trained using the training set and verified on the test set to evaluate the model's fit and prediction ability. The coefficients of the regression model were used to quantitatively evaluate the degree of influence of each blasting parameter on the tunnel monitoring parameters. The larger the absolute value of the coefficient, the greater the influence of the blasting parameter on the tunnel monitoring parameters.

[0090] The SVM model has good performance and can accurately predict the tunnel monitoring parameters after blasting. By combining the output of the SVM model with expert advice, the optimal blasting parameter scheme under different parameter combinations was identified. The results of linear regression analysis show that the wire charge density, detonation interval, and smooth eye spacing have significant effects on tunnel monitoring parameters. The wire charge density has the greatest positive effect on tunnel wall stress, and the detonation interval has a significant negative effect on vibration acceleration. The association rule mining results provide multiple potential parameter combination rules. For example, a combination of higher wire charge density and smaller smooth eye spacing may increase the risk of tunnel wall stress exceeding the limit.

[0091] According to another embodiment of the present invention, Figure 2 The schematic diagram of the optimal blasting parameter combination generation steps of a multi-sensor fusion tunnel smooth blasting monitoring system is shown in the figure. In the optimal blasting parameter combination generation module, the safety index SI is constructed based on the tunnel monitoring parameters, and the efficiency index EI is constructed based on the blasting parameters. The objective function F is defined as:

[0092] F=w1*SI+w2*EI.

[0093] Wherein, w1 is the first weight coefficient, w2 is the second weight coefficient;

[0094] Set the number of particles to N = 100, and each particle represents a set of blasting parameter combinations (x1, x2, ..., x m ), where m is the number of parameters of the explosion parameters; the initial position and velocity of each particle are randomly generated;

[0095] Calculate the fitness value F(x i ), using the defined objective function, where i traverses from 1 to m respectively; save the best position p of each particle i and the global optimal position g; update the particle's velocity v using the following formula i and position x i :

[0096] v i (t+1)=w*v i (t)+c′1*r′1*(p i -x i (t))+c′2*r′2*(gx i (t));

[0097] x i (t+1)=x i (t)+v i (t+1);

[0098] Where w is the inertia weight, set to 0.7, c′x is the first acceleration constant, set to 1.5, c′2 is the second acceleration constant, set to 1.5, r′1 is the first random number, r′2 is the second random number, and the value range of r′1 and r′2 is [0, 1], v i (t+1) represents the speed at time t+1, x i (t+1) indicates the position at time t+1;

[0099] The maximum number of iterations is set to 1000 and the convergence threshold is set to 0.001. When the convergence condition reaches the convergence threshold or the maximum number of iterations is reached, the iteration is stopped. The global optimal position in the particle swarm is output, which is expressed as the optimal blasting parameter combination. The optimal parameters are applied to the actual blasting operation, and the various parameters of the tunnel are monitored to verify the optimization effect. The model parameters and optimization objectives are further adjusted according to the actual results, and the model is iterated.

[0100] In this embodiment, some blasting parameters are as follows: charge: 0.5-1.5 kg / m, smooth eye spacing: 0.3-0.8 m, minimum resistance line: 0.8-1.5 m, line charge density: 0.4-1.0 kg / m, detonation interval: 0.001-0.005 s. Initial safety index (SI): tunnel wall stress: 11.2 MPa, tunnel displacement: 2.8 mm, vibration acceleration: 30 m / s 2 . Efficiency Index (EI): Blasting energy efficiency: increased by 15%.

[0101] Parameters before optimization: charge amount: 1.2kg / m, smooth eye spacing: 0.5m, minimum resistance line: 1.2m, line charge density: 0.85kg / m, detonation interval: 0.003s.

[0102] The monitoring results are as follows: tunnel wall stress: 11.2MPa, tunnel displacement: 2.8mm, vibration acceleration: 30m / s 2 .

[0103] Parameters after PSO optimization: charge: 1.0kg / m, smooth eye spacing: 0.45m, minimum resistance line: 1.0m, line charge density: 0.78kg / m, detonation interval: 0.0025s.

[0104] The monitoring results are as follows: Tunnel wall stress: 9.8MPa (reduced by 12.5%), tunnel displacement: 2.1mm (reduced by 25%), vibration acceleration: 25m / s 2 (A decrease of 16.7%).

[0105] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A multi-sensor fusion tunnel smooth surface blasting monitoring system, characterized in that: include: Tunnel digital model construction module: including: data acquisition unit and tunnel digital model construction unit; the data acquisition unit uses multiple sensors to obtain the stress, deformation and displacement of the tunnel wall, the vibration acceleration, frequency and duration in the tunnel, the gas composition and corresponding concentration in the tunnel, the temperature and humidity changes in the tunnel, and the background noise and changes; the tunnel digital model construction unit pre-processes and integrates the monitored data, uses the finite element model to build a tunnel digital model, applies static loads and dynamic loads, selects a solver according to the analysis target, and performs static and dynamic analysis; Smooth surface blasting abnormal pattern acquisition module: including: critical value determination unit, abnormal parameter extraction unit and smooth surface blasting abnormal pattern recognition unit; wherein the critical value determination unit obtains the safety critical value of each parameter in the tunnel digital model to obtain the tunnel critical value digital model; the abnormal parameter extraction unit compares the tunnel critical value digital model with the tunnel digital model, extracts each parameter exceeding the safety critical value, and generates the tunnel abnormal parameter digital model; the smooth surface blasting abnormal pattern recognition unit extracts features from the tunnel abnormal parameter digital model, selects the random forest algorithm to analyze the features, uses historical data for training, and recognizes the abnormal pattern of smooth surface blasting; Digital model and blasting parameter update module: including: internal repair suggestion unit and blasting parameter optimization unit; among them, when the abnormal pattern of smooth surface blasting is related to the interior of the tunnel, the internal repair suggestion unit will send the optimization suggestion to the staff to repair the interior of the tunnel, and rebuild the digital model of the tunnel after the tunnel is repaired; when the optimization suggestion is related to smooth surface blasting, the blasting parameter optimization unit obtains the blasting parameters of this smooth surface blasting, and uses the support vector machine model combined with expert suggestions to obtain new blasting parameters; The optimal blasting parameter combination generation module includes: an association model construction unit, an association rule identification unit, an influence degree evaluation unit and an optimal parameter search unit; wherein, the association model construction unit associates the blasting parameters with the tunnel digital model to construct a tunnel-smooth surface blasting association model, the association rule identification unit uses correlation analysis to identify the association rules between each blasting parameter and the tunnel monitoring parameter, and the influence degree evaluation unit quantitatively evaluates the influence of the blasting parameters on the tunnel monitoring parameters through regression analysis; the optimal parameter search unit inputs the safety and efficiency indicators of the blasting operation, applies particle swarm optimization, and searches for the optimal blasting parameter combination based on the results of the tunnel-smooth surface blasting association model to meet the set optimization goals.

2. The multi-sensor fusion tunnel smooth surface blasting monitoring system according to claim 1 is characterized by: In the tunnel digital model construction module, stress sensors are used to measure the stress distribution of the tunnel wall, displacement sensors are used to monitor the displacement and deformation of the tunnel wall, vibration sensors are used to monitor the vibration acceleration, frequency and duration in the tunnel, gas sensors are used to monitor the gas composition and corresponding concentration in the tunnel, temperature and humidity sensors are used to monitor the temperature and humidity changes in the tunnel, and noise sensors are used to monitor the background noise and changes; A low-pass filter is applied to remove noise from the data, outliers are detected and removed using statistical methods, data from different sensors are normalized, and all data are synchronized by timestamp.

3. The multi-sensor fusion tunnel smooth surface blasting monitoring system according to claim 2 is characterized by: In the tunnel digital model building module, the geometric model of the tunnel is created using ANSYS finite element analysis software, and the geometric parameters of the tunnel are input to form a three-dimensional model of the tunnel; Input the tunnel material properties and the physical and mechanical properties of the surrounding soil to define the material properties; select hexahedral mesh as the mesh type and perform meshing; set the contact surface between the tunnel wall and the soil as a fixed boundary to simulate the effect of soil pressure on the tunnel and set the displacement constraint conditions of the tunnel; Apply static loads, including traffic loads, soil pressure and groundwater pressure in the tunnel; apply dynamic loads, combine the influence of dynamic loads of smooth blasting, input the blasting wave propagation model, and set the corresponding instantaneous pressure or vibration load; select the solver according to the analysis objectives, perform static or dynamic analysis, set solution parameters, monitor the calculation progress and convergence, and check the rationality of the calculation results during the solution process; generate stress distribution, displacement field and deformation diagrams of the tunnel wall, compare the model output with the monitoring data of the on-site sensors, and evaluate the sensitivity of the model to changes in different input parameters.

4. The multi-sensor fusion tunnel smooth surface blasting monitoring system according to claim 3 is characterized by: Establish a blast wave propagation model, including establishing a physical model and establishing a mathematical model; determine the blasting materials used and their characteristics, and obtain the blasting parameters of this smooth blasting, including the uncoupling coefficient, the smooth eye spacing, the minimum resistance line, the blast hole proximity coefficient, the line charge density and the detonation interval time; wherein, the physical model is established by selecting a nonlinear elastic model, and a geometric model of the blasting source and the surrounding medium is created, and the surrounding medium includes the tunnel wall and the soil; the wave equation and the pressure decay model are used when establishing the mathematical model; According to the propagation of pressure waves generated by the explosion, the wave equation is used to describe the wave characteristics: Among them, u is the wave displacement, c is the wave speed, and t is the time; Using the instantaneous high-voltage pulse generated by the explosion of explosives, the following pressure decay model P(t) is used: Where P0 is the initial pressure at the time of explosion, α is the first empirical constant, n is the second empirical constant, and r is the distance from the explosion center; Instantaneous pressure boundary conditions are applied at the blasting center to simulate the pressure impact at the moment of explosion, and the contact conditions between the tunnel wall and the soil are set to ensure the interaction during the propagation of the pressure wave. The grid density is set at the blasting source to capture the instantaneous pressure changes caused by the explosion. An implicit solver is selected to process dynamic loads. The time step and solution accuracy are set to capture the dynamic response at the moment of explosion. The distribution diagram and time history curve of the pressure wave after the blasting are generated to obtain the propagation of the pressure wave in the tunnel. The simulation results are compared with the actual monitoring data to verify the blasting wave propagation model.

5. The multi-sensor fusion tunnel smooth surface blasting monitoring system according to claim 1 is characterized by: In the smooth blasting abnormal pattern acquisition module, the safety critical value of each parameter is determined according to existing industry standards and research literature, and the determined safety critical value is organized into a digital model to form a tunnel critical value digital model; the real-time monitoring data in the tunnel digital model is compared with the tunnel critical value digital model item by item, the parameters exceeding the safety critical value are identified, and their specific values ​​and change trends are recorded; Summarize the parameters that exceed the safety critical value to form a digital model of tunnel abnormal parameters, and classify them according to the nature of the parameters; analyze the changing trend of abnormal parameters to determine whether it is a single event or a continuous abnormality, use visualization tools to display the spatial distribution of abnormal parameters in the tunnel, and identify high-risk areas; Compare the abnormal parameters with historical data to confirm whether it is a true abnormality.

6. The multi-sensor fusion tunnel smooth surface blasting monitoring system according to claim 5 is characterized by: In the smooth blasting abnormal pattern acquisition module, all parameters exceeding the safety critical value are extracted from the digital model of tunnel abnormal parameters to form a structured data set. The extracted abnormal parameters are annotated to distinguish between normal and abnormal states. Correlation analysis is used to extract the time series characteristics, change rate characteristics and frequency domain characteristics of the abnormal parameters. The extracted features are divided into training set and test set. The training set data is used to train the random forest model. After the training is completed, the feature importance evaluation method provided by the random forest is used to identify the features that have the greatest impact on the abnormal pattern of smooth surface blasting. The test set is input into the trained random forest model to obtain the prediction result of each sample, identify the abnormal pattern of smooth surface blasting, form an abnormal pattern output list, and use accuracy, recall rate and F1-score to evaluate the performance of the model.

7. The multi-sensor fusion tunnel smooth surface blasting monitoring system according to claim 1 is characterized by: In the digital model and blasting parameter update module, a specific repair plan is formulated based on the abnormal pattern of smooth blasting, combined with expert knowledge and historical data. The optimization suggestions are automatically sent to the on-site staff through the system. The staff carry out the repair work inside the tunnel according to the optimization suggestions to ensure that the tunnel structure is restored to a safe state. During the repair process, multiple sensors are used to monitor the parameters inside the tunnel and record data to evaluate the repair effect; The new data collected during the repair process are sorted and cleaned, and the digital model of the tunnel is regenerated according to the updated parameters. The reconstructed digital model of the tunnel is compared with the actual monitoring data, and the impact of different parameter changes on the model results is evaluated to ensure the robustness of the model.

8. The multi-sensor fusion tunnel smooth surface blasting monitoring system according to claim 7 is characterized by: In the digital model and blasting parameter update module, when the optimization suggestion is related to smooth surface blasting, the blasting parameters of this smooth surface blasting are obtained, and data related to historical blasting operations are collected, including: cases of successful and failed blasting, specific blasting parameters used in the cases, and corresponding tunnel monitoring parameters; data preprocessing is performed to extract the historical values ​​of each blasting parameter, changes in tunnel monitoring parameters after blasting, and environmental conditions of blasting operations, and divide the training set and test set; Using the grid search method, the radial basis function is selected as the SVM kernel function, the hyperparameters are selected, and the SVM model is trained using the training set to learn the mapping relationship between different blasting parameters and tunnel monitoring parameters; the model performance is evaluated on the test set; the output of the support vector machine model is combined with expert suggestions for comprehensive analysis to obtain new blasting parameters.

9. The multi-sensor fusion tunnel smooth surface blasting monitoring system according to claim 1, characterized in that: In the optimal blasting parameter combination generation module, the blasting parameters and the corresponding tunnel monitoring parameters are integrated into a data set to ensure that each set of blasting parameters corresponds to the subsequent monitoring results; The Pearson correlation coefficient is used to calculate the correlation between each blasting parameter and the tunnel monitoring parameter, generate a correlation matrix, and identify positive and negative correlations. The association rule learning algorithm Apriori is used to mine the relationship between blasting parameters and tunnel monitoring parameters and identify potential association rules. A linear regression model was selected for modeling. The tunnel monitoring parameters were used as dependent variables and the blasting parameters were used as independent variables. The data were divided into a training set and a test set. The regression model was trained using the training set and verified on the test set to evaluate the model's fit and prediction ability. The coefficients of the regression model were used to quantitatively evaluate the degree of influence of each blasting parameter on the tunnel monitoring parameters. The larger the absolute value of the coefficient, the greater the influence of the blasting parameter on the tunnel monitoring parameters.

10. The multi-sensor fusion tunnel smooth surface blasting monitoring system according to claim 9, characterized in that: In the optimal blasting parameter combination generation module, the safety index SI is constructed based on the tunnel monitoring parameters, and the efficiency index EI is constructed based on the blasting parameters. The objective function F is defined as: F = w1*SI+w2*EI; Wherein, w1 is the first weight coefficient, w2 is the second weight coefficient; Set the number of particles to N, and each particle represents a set of blasting parameter combinations (x1, x2, ..., x m ), where m is the number of parameters of the explosion parameters; the initial position and velocity of each particle are randomly generated; Calculate the fitness value F(x i ), using the defined objective function, where i traverses from 1 to m respectively; save the best position p of each particle i and the global optimal position g; update the particle's velocity v using the following formula i and position x i : v i (t+1)=w*v i (t)+c′1*r′1*(p i -x i (t))+c′2*r′2*(g-x i (t)); x i (t+1)=x i (t)+v i (t+1); Where w is the inertia weight, c′1 is the first acceleration constant, c′2 is the second acceleration constant, r′1 is the first random number, r′2 is the second random number, and the value range of r′1 and r′2 is [0,1], v i (t+1) represents the speed at time t+1, x i (t+1) indicates the position at time t+1; The maximum number of iterations and the convergence threshold are set. When the convergence condition reaches the convergence threshold or the maximum number of iterations is reached, the iteration is stopped. The global optimal position in the particle swarm is output, which is expressed as the optimal blasting parameter combination. The optimal parameters are applied to the actual blasting operation, and the various parameters of the tunnel are monitored to verify the optimization effect. According to the actual results, the model parameters and optimization objectives are further adjusted to perform model iteration.

Images