Tunnel dynamic risk early warning method and system based on intelligent algorithm

Abstract

The invention discloses a tunnel dynamic risk early warning method and system based on an intelligent algorithm, and belongs to the technical field of tunnel safety early warning. Based on the external load data, the gas pressure, the soil pressure and the temperature data, calculating stress data of each part of the tunnel by using a structural mechanical model; calculating an effective gas diffusion coefficient based on the stress data and the temperature data; calculating a fluid dynamic pressure effect based on the stress data, the fluid density and the flow velocity data; and evaluating the safety risk of the tunnel, generating an early warning signal according to a preset risk threshold value, and triggering a risk early warning notification. According to the invention, the limitation of single data source or periodic detection in the traditional method is overcome. And the stress and the safety risk of the tunnel are accurately evaluated by combining a structural mechanical model, a gas diffusion model and fluid dynamic pressure effect calculation. Through a gradient elevator algorithm, the accuracy and reliability of risk prediction are improved by using historical and real-time data, and dynamic management of tunnel safety is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel safety early warning, and particularly relates to a tunnel dynamic risk early warning method and system based on intelligent algorithms. Background Art

[0002] As an important infrastructure, tunnels are widely used in fields such as transportation, mines, and urban construction, undertaking important transportation and passage functions. The safety of tunnels directly affects people's lives and property safety. Therefore, the stability assessment and risk early warning of tunnels have always been the focus in the engineering field and research area. However, traditional tunnel safety monitoring methods have limitations in many aspects and are difficult to meet the safety requirements during the construction and operation of modern tunnels.

[0003] Currently, tunnel safety monitoring technologies usually rely on the following methods:

[0004] Regular manual inspection: Many tunnels adopt the method of regular manual inspection, relying on technicians to evaluate the tunnel structure through visual inspection, manual inspection, or traditional instruments. Although these methods can detect obvious damages and potential hazards, their limitations lie in the long monitoring cycle, lagging response, and inability to achieve real-time dynamic monitoring.

[0005] Single data source analysis: Some traditional methods obtain certain environmental data or structural data inside the tunnel by installing monitoring devices such as stress sensors and temperature and humidity sensors, and then conduct stress analysis or risk assessment based on existing physical models. Although such methods can reflect the safety state of the tunnel to a certain extent, they often only rely on a single data source and cannot comprehensively consider various influencing factors.

[0006] Simplified risk assessment models: Currently, many tunnel safety assessment methods adopt simplified risk assessment models, such as models based on static stress calculation. These methods often ignore the impacts of various dynamic environmental factors, such as gas diffusion and fluid dynamic pressure, on the tunnel stability. Simple physical models often cannot accurately reflect the changes and stress distribution of the tunnel under complex environmental conditions, resulting in inaccurate assessment results.

[0007] Due to the fact that traditional monitoring methods are usually regular inspections or manual inspections, there is a problem of lagging information update and it is impossible to achieve real-time monitoring and rapid response. Especially when the tunnel faces sudden risks, manual inspections and periodic detections cannot discover problems in time, causing potential safety hazards.

[0008] Most of the existing technologies focus on single monitoring data, such as stress, temperature, or gas pressure, etc., and do not fully consider the comprehensive impacts of various factors on tunnel safety. For example, important environmental factors such as gas diffusion coefficient and fluid dynamic pressure effect are often not fully emphasized in existing assessment methods, which leads to incomplete and inaccurate assessment results.

[0009] Many existing tunnel safety assessment methods rely on static physical models for stress calculation, unable to handle the dynamic changes of tunnels under different working conditions and environmental conditions, and cannot effectively cope with the complex situations in tunnel operation.

[0010] Existing assessment methods usually rely on simplified assumptions and empirical models, unable to conduct comprehensive assessments through deep learning of complex data, resulting in low accuracy and reliability of model predictions. Especially when facing complex and variable tunnel environments, existing technologies often cannot give accurate safety risk assessment results. Summary of the Invention

[0011] To solve the above technical problems, a tunnel dynamic risk early warning method based on intelligent algorithms is proposed, including: real-time collecting soil pressure, gas pressure, fluid density, flow velocity data and temperature data in the tunnel through sensors to obtain soil humidity, soil density, soil type and soil temperature data; obtaining external load data on the tunnel through a load monitoring device; obtaining geometric dimension data of the tunnel; based on the external load data, gas pressure, soil pressure and temperature data, using a structural mechanics model to calculate stress data of each part of the tunnel; based on the stress data and temperature data, calculating the effective gas diffusion coefficient; based on the stress data, fluid density and flow velocity data, calculating the fluid dynamic pressure effect; inputting the effective gas diffusion coefficient and the fluid dynamic pressure effect into the intelligent algorithm to evaluate the safety risk of the tunnel, and generating an early warning signal according to a preset risk threshold to trigger a risk early warning notice.

[0012] As a preferred scheme of a tunnel dynamic risk early warning method based on intelligent algorithms according to the present invention, further including: before calculating the stress data of each part of the tunnel, based on the soil humidity, soil density, soil type and external load data, using a regression model or a machine learning method to calculate a preliminary value of the soil reaction coefficient; adjusting the preliminary value of the soil reaction coefficient according to the soil temperature data to obtain a corrected soil reaction coefficient; performing secondary correction on the corrected soil reaction coefficient according to historical load data, considering the long-term impact of historical loads on soil compaction, and determining the final soil reaction coefficient.

[0013] As a preferred scheme of a tunnel dynamic risk early warning method based on intelligent algorithms according to the present invention, the step of using a structural mechanics model to calculate the stress data of each part of the tunnel includes: obtaining external load, gas pressure, soil pressure, soil reaction coefficient and temperature data, and the cross-sectional area of the tunnel; using a structural mechanics model to calculate the overall stress of the tunnel according to the external load and gas pressure; correcting the overall stress based on the soil pressure and soil reaction coefficient to obtain the local stress; combining the local stress and the geometric dimension data to obtain the final stress value of each part of the tunnel.

[0014] As a preferred solution of a tunnel dynamic risk early warning method based on an intelligent algorithm according to the present invention, wherein: calculating the effective gas diffusion coefficient includes calculating a preliminary gas diffusion coefficient based on tunnel stress data and temperature data; correcting the gas diffusion coefficient according to the stress data to obtain a corrected diffusion coefficient; and further adjusting the gas diffusion coefficient using the temperature data to obtain a final effective gas diffusion coefficient.

[0015] As a preferred solution of a tunnel dynamic risk early warning method based on an intelligent algorithm according to the present invention, wherein: calculating the hydrodynamic pressure effect includes obtaining tunnel stress data, fluid density and flow velocity data, and inputting them into a fluid mechanics model for calculation; calculating the hydrodynamic pressure effect of the fluid on the tunnel wall based on the fluid density and flow velocity data; and correcting the hydrodynamic pressure effect according to the tunnel stress data to obtain a final corrected result.

[0016] As a preferred solution of a tunnel dynamic risk early warning method based on an intelligent algorithm according to the present invention, wherein: the intelligent algorithm includes selecting the gradient boosting machine algorithm, and modeling with the gas diffusion coefficient and the hydrodynamic pressure effect as input features; training the gradient boosting machine model using historical tunnel risk data including historical gas diffusion coefficients, hydrodynamic pressure effects and corresponding risk assessment values, and learning the relationship between the input features and the risk assessment values through the historical tunnel risk data; and using the trained gradient boosting machine model to predict risks based on the gas diffusion coefficient and the hydrodynamic pressure effect data to obtain the safety risk assessment value of the tunnel.

[0017] As a preferred solution of a tunnel dynamic risk early warning method based on an intelligent algorithm according to the present invention, wherein: evaluating the safety risk of the tunnel includes comparing the safety risk assessment value with a threshold value, and classifying the risk levels into low risk, medium risk and high risk; generating an early warning signal according to the risk level and starting safety response measures.

[0018] Another object of the present invention is to provide a tunnel dynamic risk early warning system based on an intelligent algorithm. The present invention solves the problems that the prior art fails to realize real-time dynamic monitoring and early warning of tunnel safety, especially in the face of various complex factors such as environmental changes, hydrodynamic pressure and gas diffusion, it is impossible to perform timely and accurate risk assessment; it is difficult for the prior art to comprehensively consider the combined effects of various environmental factors inside and outside the tunnel (such as soil reaction coefficient, gas diffusion coefficient, hydrodynamic pressure, etc.) on tunnel safety, and it is impossible to provide comprehensive input features for risk assessment, resulting in deviation of the assessment results; in the prior art, many risk assessment models cannot effectively utilize multi-dimensional data for deep learning and feature extraction, cannot improve the assessment accuracy and reliability, and often lack effective algorithm support.

[0019] As a preferred solution of a tunnel dynamic risk early warning system based on intelligent algorithms according to the present invention, it is characterized by including: a data acquisition module, configured to collect in real time soil pressure, gas pressure, fluid density, flow rate data and temperature data in the tunnel through sensors, obtain soil humidity, soil compactness, soil type and soil temperature data, obtain external load data borne by the tunnel through a load monitoring device, and obtain geometric dimension data of the tunnel; a stress analysis module, configured to calculate stress data of each part of the tunnel using a structural mechanics model based on the external load data, gas pressure, soil pressure and temperature data; a diffusion analysis module, configured to calculate an effective gas diffusion coefficient based on the stress data and temperature data; a dynamic pressure analysis module, configured to calculate a fluid dynamic pressure effect based on the stress data, fluid density and flow rate data; and a risk early warning module, configured to input the effective gas diffusion coefficient and the fluid dynamic pressure effect into an intelligent algorithm, evaluate the safety risk of the tunnel, generate a warning signal according to a preset risk threshold, and trigger a risk early warning notification.

[0020] A computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of a tunnel dynamic risk early warning method based on intelligent algorithms are implemented.

[0021] A computer-readable storage medium stores a computer program thereon. When the computer program is executed by a processor, the steps of a tunnel dynamic risk early warning method based on intelligent algorithms are implemented.

[0022] The beneficial effects of the present invention: By collecting environmental data inside and outside the tunnel in real time through sensors, the limitations of single data sources or periodic detections in traditional methods are avoided.

[0023] The stress of each part of the tunnel is calculated using a structural mechanics model. By considering various factors such as external loads, gas pressure, soil pressure and temperature, it is ensured that the calculation results of the tunnel stress can comprehensively reflect the actual load situation faced by the tunnel. This process not only considers the overall stress of the tunnel, but also further corrects the stress based on the soil reaction coefficient to obtain a more accurate local stress value.

[0024] After obtaining the stress data, the present invention calculates the effective gas diffusion coefficient through a gas diffusion model and corrects it in combination with temperature data. The calculation of the gas diffusion coefficient can accurately reflect the characteristics of gas propagation in the tunnel, ensuring that the diffusion situation of harmful gases in the tunnel is effectively evaluated. At the same time, the calculation of the fluid dynamic pressure effect can further analyze the influence of the fluid on the tunnel based on the stress data, fluid density and flow rate data. These calculation results together provide an important basis for the safety assessment of the tunnel.

[0025] Input these calculation results into the gradient boosting machine algorithm to further evaluate the safety risks of the tunnel. Through training the model, the intelligent algorithm can accurately predict the risks of the tunnel in a complex multi-factor environment. The model uses historical risk data and real-time data to learn the relationship between input features and risk assessment values, thereby improving the accuracy and reliability of risk assessment and reducing the possibility of misjudgment and missed judgment.

[0026] According to the evaluated safety risk values, the present invention classifies the risks of the tunnel and initiates corresponding warning responses according to different risk levels. By setting multiple risk levels, refined management of the risk status of the tunnel can be achieved. Brief Description of the Drawings

[0027] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0028] Figure 1 It is the overall flowchart of a method for dynamic risk warning of a tunnel based on an intelligent algorithm provided by an embodiment of the present invention. Detailed Embodiments

[0029] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0030] Example 1, referring to Figure 1 , which is the first embodiment of the present invention. This embodiment provides a method for dynamic risk warning of a tunnel based on an intelligent algorithm, including:

[0031] Step 101: Real-time collect the soil pressure, gas pressure, fluid density, flow rate data, and temperature data inside the tunnel through sensors to obtain the soil humidity, soil compactness, soil type, and soil temperature data. Obtain the external load data received by the tunnel through the load monitoring device, and obtain the geometric dimension data of the tunnel.

[0032] Step 102: Based on the external load data, gas pressure, soil pressure, and temperature data, use the structural mechanics model to calculate the stress data of each part of the tunnel.

[0033] Step 103: Calculate the effective gas diffusion coefficient based on the stress data and temperature data;

[0034] Step 104: Calculate the hydrodynamic pressure effect based on the stress data, fluid density, and flow velocity data;

[0035] Step 105: Input the effective gas diffusion coefficient and the hydrodynamic pressure effect into the intelligent algorithm, evaluate the safety risk of the tunnel, generate a warning signal according to the preset risk threshold, and trigger a risk warning notice.

[0036] In Step 102, based on the soil moisture, soil density, soil type, and external load data, use a regression model or machine learning method to calculate the initial value of the soil reaction coefficient;

[0037] Adjust the initial value of the soil reaction coefficient according to the soil temperature data to obtain the corrected soil reaction coefficient;

[0038] Perform a secondary correction on the corrected soil reaction coefficient according to the historical load data, considering the long-term impact of historical loads on soil compaction, and determine the final soil reaction coefficient.

[0039] In Step 102, the stress data of each part of the tunnel is used to obtain the external load, gas pressure, soil pressure, soil reaction coefficient, and temperature data, as well as the cross-sectional area of the tunnel;

[0040] Use a structural mechanics model to calculate the overall stress of the tunnel based on the external load and gas pressure;

[0041] Correct the overall stress based on the soil pressure and soil reaction coefficient to obtain the local stress;

[0042] Combine the local stress and geometric dimension data to obtain the final stress value of each part of the tunnel.

[0043] In Step 103, calculate the preliminary gas diffusion coefficient based on the tunnel stress data and temperature data;

[0044] Correct the gas diffusion coefficient according to the stress data to obtain the corrected diffusion coefficient;

[0045] Use the temperature data to further adjust the gas diffusion coefficient to obtain the final effective gas diffusion coefficient.

[0046] In Step 104, obtain the tunnel stress data, fluid density, and flow velocity data, and input them into the fluid mechanics model for calculation;

[0047] Calculate the hydrodynamic pressure effect of the fluid on the tunnel wall based on the fluid density and flow velocity data;

[0048] Correct the hydrodynamic pressure effect according to the tunnel stress data to obtain the final corrected result.

[0049] In step 105, a gradient boosting algorithm is selected to perform modeling using gas diffusion coefficient and fluid dynamic pressure effect as input features;

[0050] Using historical tunnel risk data including historical gas diffusion coefficients, fluid dynamic pressure effects and corresponding risk assessment values to train a gradient boosting machine model, the relationship between input features and risk assessment values is learned through the historical tunnel risk data;

[0051] The trained gradient boosting machine model is used to predict risks based on the gas diffusion coefficient and fluid dynamic pressure effect data to obtain the safety risk assessment value of the tunnel.

[0052] Compare the security risk assessment value with the threshold value and classify the risk level into low risk, medium risk and high risk;

[0053] Generate early warning signals based on risk levels and initiate security response measures.

[0054] In a preferred embodiment of the method of this embodiment, the stress of each part of the tunnel in step 102 is calculated by a stress model, and the expression is:

[0055]

[0056] Where, σ represents the stress in the tunnel; F ext Represents external load; F gas represents gas pressure; A represents the cross-sectional area of the tunnel; κ represents the soil reaction coefficient; P soil represents soil pressure; T represents temperature; T0 represents reference temperature.

[0057] It should be noted that in the prior art, the acquisition of tunnel stress mainly relies on two methods: sensor measurement and structural mechanics analysis. By installing strain gauges, fiber optic sensors and other equipment in the tunnel structure, the stress value is monitored in real time. These sensors can directly measure the changes in tunnel stress and convert the signals into stress data for engineers to evaluate the health status of the tunnel in real time. In addition, finite element analysis (FEA) is also commonly used to infer the stress state of the tunnel by simulating external loads, gas pressure, soil pressure and other factors. These methods are usually based on assumptions and rely on a lot of computing resources.

[0058] However, there are some deficiencies in the existing technologies. Firstly, although the sensor measurement method can provide real-time data, its installation and maintenance costs are high, and it cannot fully cover all parts of the tunnel, especially in some difficult-to-reach areas. In addition, the sensors may be interfered by the external environment, resulting in inaccurate data. Although the finite element analysis method can perform relatively accurate stress calculations, its calculation process is complex, relies on a large number of assumptions, and requires high computing resources, making it difficult to perform real-time analysis. Traditional theoretical calculation methods usually rely on simplified assumptions and cannot consider the effects of complex soil reactions, gas pressures, and temperature changes on tunnel stresses, resulting in inaccurate calculation results.

[0059] The formula of the method in this embodiment effectively compensates for the deficiencies in traditional methods by introducing various factors, especially the soil reaction coefficient and the temperature correction term. The external load and gas pressure in the formula are comprehensively calculated by the weighted summation method, and the stress is evenly distributed in combination with the cross-sectional area of the tunnel. The correction of soil pressure and temperature further adjusts the calculation result of the stress, enabling the stress calculation to more accurately reflect the changes in the actual environment. By introducing the soil reaction coefficient and the temperature correction term, the method in this embodiment can fully consider the changes in the tunnel surrounding environment, especially making more accurate corrections for changes in soil type, density, and humidity. The introduction of this formula not only avoids the simplified assumptions in traditional methods but also can reflect the stress changes of the tunnel under different environmental conditions in real time, providing more reliable stress calculation results.

[0060] Compared with the existing technologies, firstly, the method in this embodiment comprehensively considers more environmental factors, including the soil reaction coefficient and the temperature correction term, ensuring more comprehensive and accurate stress calculations. Secondly, this formula improves the real-time performance and adaptability of stress calculations and can update the stress assessment in real time according to environmental changes during the use of the tunnel. Traditional methods usually cannot achieve this. Finally, the method in this embodiment avoids the cumbersome calculation process and the high demand for computing resources. Compared with finite element analysis, the formula calculation is more simple and efficient and is applicable to the daily monitoring and risk assessment of tunnels. Therefore, the method in this embodiment not only improves the accuracy of stress calculations but also enhances the real-time performance and the adaptability of the system.

[0061] It should be noted that by collecting the soil humidity, soil density, soil type, soil temperature, external load data, and historical load data of the tunnel and combining with an appropriate calculation model, the soil reaction coefficient can be accurately calculated. In this process, first, sensors are used to collect the soil humidity, soil density, soil type, and soil temperature data in the tunnel in real time. The data of the external load on the tunnel is obtained through a load monitoring device. At the same time, the historical load data of the tunnel and the relevant physical properties of the soil are used to provide input parameters for the subsequent calculation of the soil reaction coefficient.

[0062] The preliminary calculation of the soil reaction coefficient adopts a regression model or machine learning method. The collected data of soil humidity, soil compaction, soil type, soil temperature and external load are used as input features, and the model is trained through regression analysis or machine learning algorithms (such as decision trees, random forests, etc.) to obtain the preliminary value of the soil reaction coefficient. Specifically, the calculation form of the regression model is as follows. By calculating the influence of soil humidity, soil compaction, soil type and external load on the soil reaction coefficient, the preliminary value is obtained.

[0063] After that, the soil temperature data is used to correct the preliminarily calculated soil reaction coefficient. The temperature correction takes into account the influence of soil temperature on soil compaction and reactivity. This correction process optimizes the preliminarily calculated coefficient using a temperature correction factor by adjusting the relationship between the preliminary soil reaction coefficient and temperature. The influence of temperature is quantified as a correction factor, which is related to the ratio between the deviation of the actual soil temperature and the reference temperature, so as to obtain the temperature-corrected soil reaction coefficient.

[0064] To ensure the accuracy of the final soil reaction coefficient, a secondary correction is also required according to the historical load data. The historical load data reflects the reaction changes of the soil under long-term load, especially the soil compaction changes during long-term use. By combining the historical load data with the corrected soil reaction coefficient and considering the long-term stress influence of the soil, the soil reaction coefficient is further adjusted. This step corrects the soil reaction coefficient by establishing the relationship between historical load and soil reaction, making it more accurately reflect the cumulative effect of long-term load.

[0065] Finally, the soil reaction coefficient obtained through the above steps, combined with the data of soil humidity, soil compaction, soil type, soil temperature, external load and historical load, can provide accurate parameters for tunnel stress calculation. This calculation process effectively considers the multi-dimensional factors of the soil environment, ensures the accuracy and reliability of the soil reaction coefficient, and provides an important basis for the safety risk assessment and dynamic risk warning of the tunnel.

[0066] In an alternative embodiment of the method of this embodiment, the following data is collected by installing sensors and measurement devices:

[0067] Soil humidity S soil : Obtained by a soil humidity sensor. This data describes the water content of the soil and usually affects soil compaction and reactivity.

[0068] Soil compaction D soil : Measured by a soil compaction sensor. Soil compaction affects soil porosity and stress response.

[0069] Soil type T soil: Determine the soil type through on-site investigation and laboratory tests, and convert the soil type into a numerical variable using the category encoding or OneHotEncoding method.

[0070] Soil temperature T soil : Obtained through a soil temperature sensor. Temperature has an important impact on the soil reaction coefficient because it affects the compaction and structural stability of the soil.

[0071] External load F ext : Obtained through a load monitoring device. External loads (such as traffic loads, construction equipment loads, etc.) affect the stress of the tunnel structure, which in turn affects the soil reaction.

[0072] Historical load data F historical : Obtain the historical loads experienced by the tunnel through long-term monitoring records, with the unit of N. This data reflects the long-term impact of load accumulation on the soil reaction.

[0073] Based on the collected soil moisture, soil density, soil type, and external load data, use a regression model or machine learning method to calculate the preliminary soil reaction coefficient. Assuming a linear regression model is used, its formula is:

[0074] κ initial =α0 + α1S soil +α2D soil +α3T soil +α4F ext

[0075] Where, κ initial is the preliminary soil reaction coefficient; S soil is the soil moisture; D soil is the soil density; T soil is the soil type, converted into a numerical form; F ext is the external load; α0, α1, α2, α3, α4 are regression coefficients obtained through data fitting.

[0076] Fit the regression coefficients through methods such as the least squares method or the gradient descent method to obtain the preliminary calculated value of the soil reaction coefficient.

[0077] If a machine learning method is used, such as support vector machines, decision trees, or random forests, learn the relationship between soil characteristics and the reaction coefficient through the training set data to obtain the preliminary calculation results.

[0078] The soil reaction coefficient needs to be corrected according to the soil temperature data because temperature directly affects the compaction and reactivity of the soil. The correction formula is as follows:

[0079]

[0080] Among them, κ adjusted is the soil reaction coefficient after temperature correction; T soil is the soil temperature; T0 is the reference temperature; ω is the temperature correction coefficient, which is determined according to the influence of different soil types and temperature changes.

[0081] Through this correction formula, the preliminary soil reaction coefficient can be adjusted according to the temperature change of the soil, reflecting the influence of temperature on the soil reaction.

[0082] Over time, the loads on the tunnel accumulate continuously, and the influence of historical loads on the soil reaction coefficient should be taken into account. Therefore, it is necessary to perform a secondary correction on the soil reaction coefficient through historical load data. The correction formula is as follows:

[0083]

[0084] Among them, κ final is the final soil reaction coefficient; F historical is the historical load data; F max is the design maximum load; γ is the historical load correction coefficient, reflecting the long-term influence of the load on the soil compaction.

[0085] This correction process can accurately consider the cumulative effect of long-term loads, so as to obtain a soil reaction coefficient that more conforms to the actual situation.

[0086] It should be noted that to obtain the temperature correction coefficient of the soil reaction coefficient through experiments, temperature change tests need to be carried out on different soil samples. First, the soil samples are processed at different temperatures through experiments, and their physical properties such as compressive strength, shear strength, compressive capacity, density, and porosity at different temperatures are measured. By measuring the changes in the soil under temperature change conditions, the influence of temperature on the soil can be evaluated. After the experimental data is processed by regression analysis, the relationship between the soil reaction coefficient and temperature is obtained. Through regression model calculation, the temperature correction coefficient can be obtained, which is used to adjust the preliminarily calculated soil reaction coefficient to reflect the influence of soil temperature on its reaction characteristics. The specific method is to set the standard temperature value as a reference, quantify the difference between the actual temperature and the reference temperature as a correction factor, and the finally obtained correction coefficient can accurately reflect the influence of temperature change on the soil reaction coefficient, ensuring the accuracy of the soil reaction coefficient calculation.

[0087] The acquisition of the historical load correction factor involves the influence of long-term loads on the soil reaction coefficient. To this end, it is necessary to simulate the influence of long-term loads on the tunnel through experiments, conduct compaction experiments and settlement measurements under long-term loads. Through the experiments, it is possible to record the changes in soil density, porosity and settlement under continuous loads. At the same time, by measuring the soil reaction coefficient at different time nodes, the dynamic changes in soil response under loads can be obtained. Based on these data, a regression analysis method is used to establish the relationship between the load and the soil reaction coefficient, and finally the historical load correction factor is obtained. This coefficient reflects the long-term influence of load accumulation on the compaction and reactivity of the soil. By correlating the historical load data with the soil reaction coefficient, the soil reaction coefficient is finally corrected so that it can more accurately reflect the reaction changes of the soil under long-term loads, providing more reliable calculation results of the soil reaction coefficient.

[0088] Furthermore, the external load is a key parameter affecting the tunnel stress calculation, which specifically includes multiple sources, such as traffic load, construction equipment load, groundwater pressure, seismic load and soil pressure, etc. Each load source has different degrees of influence on the tunnel stress, so it is necessary to quantify the specific values of each load source one by one.

[0089] First of all, the traffic load model is used to quantify the influence of traffic on the tunnel. The traffic load is usually generated by the vehicles passing through the tunnel. According to the road type, traffic density and vehicle types passing through the tunnel, a standard vehicle load model can be used to estimate the load. The quantification of the traffic load usually uses the standard vehicle load as the basis and is calculated through specific axle loads, vehicle weights and wheel loads. According to the design requirements of the tunnel and the number of passing vehicles, the influence of the traffic load on the tunnel structure is calculated.

[0090] During the tunnel construction process, the action of construction equipment and heavy machinery will also generate external loads. By evaluating the mechanical equipment used during the construction process, the pressure exerted on the tunnel structure by the self-weight and working conditions of these equipment can be measured. The quantification of the construction equipment load is calculated by recording the weights and operation frequencies of the equipment actually used in the tunnel construction. For example, when using equipment such as cranes, excavators and transport vehicles, certain loads will be exerted. This load can be estimated according to the self-weight of the equipment and the operation time period and combined with the construction stage as the load input.

[0091] Groundwater pressure is another external load that affects tunnel structures, especially in areas with a high water table. The calculation of groundwater pressure usually depends on factors such as the head height, soil permeability, and tunnel burial depth. By measuring the head height and the permeability of the soil layer, and combining with the depth of the tunnel, the pressure exerted by groundwater on the tunnel is calculated using the head pressure formula. These calculation results provide a basis for the quantification of groundwater loads, and usually calculate the external load of water on the tunnel through the water pressure formula.

[0092] The influence of seismic loads is particularly significant in seismically active areas. The calculation of seismic loads is based on local seismic intensities, soil characteristics, and tunnel design characteristics. Through seismic intensity data, soil response characteristics, and the structural form of the tunnel, a seismic load coefficient can be used to estimate the impact of earthquakes on the tunnel structure. This coefficient is generally calculated using the standard values provided in seismic engineering design codes, and combined with the actual situation of the tunnel to obtain a quantified value of the seismic load.

[0093] Finally, soil pressure is also one of the external loads on the tunnel. Soil pressure mainly comes from the tunnel burial depth and the compactness of the soil layer. Soil pressure is usually calculated through soil density, soil depth, and gravitational acceleration. According to the type of soil, the porosity of the soil, and its stress distribution characteristics, combined with the geometric parameters of the tunnel, the external load exerted by the soil on the tunnel is calculated. The calculation of soil pressure usually adopts a soil mechanics model, considering the physical properties of the soil and its long-term impact on the tunnel.

[0094] After calculating each load, these load values are superimposed according to their weights to obtain the total external load, which is used as the input for tunnel stress calculation. The final external load is obtained by weighted summation of various loads to ensure more accurate and comprehensive stress calculation of the tunnel under different environmental and load conditions.

[0095] In another preferred embodiment of the method of this embodiment, the diffusion rate of gas in the tunnel is affected by tunnel stress and temperature. The stress of the tunnel will change the gas diffusion path through its influence on the structure. Therefore, the effective diffusion coefficient of the gas can be calculated through the following formula, and the expression is:

[0096]

[0097] where D eff represents the effective gas diffusion coefficient; D0 represents the base diffusion coefficient; μ represents the influence coefficient of stress on diffusion; σ represents the tunnel stress; σ yield represents the material yield stress; λ represents the influence coefficient of temperature on the diffusion coefficient; T represents the temperature in the tunnel; T0 represents the reference temperature.

[0098] It should be noted that, first of all, the basic diffusion coefficient D0 is the default diffusion ability of a gas under specific environmental conditions, which usually depends on the properties of the gas and the physical characteristics of the medium. The basic diffusion coefficient can be measured by conducting gas diffusion experiments under standard experimental conditions. For example, in the laboratory, by setting the initial concentration of the gas and measuring the concentration changes at different time points, the diffusion rate of the gas in a certain medium can be obtained. By comparing the actual experimental data with the theoretical model, the diffusion ability of the gas in the medium can be deduced. In practical applications, the basic diffusion coefficient D0 can also refer to existing literature data and standard values, especially for common gases (such as air, nitrogen, carbon dioxide, etc.) and common soil types.

[0099] Secondly, the influence coefficient μ of stress on diffusion reflects the influence of the tunnel structure stress on the gas diffusion process. When the tunnel is subjected to external loads, the structure may undergo slight deformation, thereby changing the gas diffusion path. The influence of stress on gas diffusion is usually obtained by comparing the gas diffusion rates under different stress levels. During the experiment, different stress levels can be applied (for example, by loading plates, compression tests, etc.), and the gas diffusion rate in the soil or tunnel can be measured. Through regression analysis, the relationship between stress change and gas diffusion rate can be obtained, and then the value of μ can be determined. For different types of soil and different gases, μ may vary, so it should be determined according to the actual situation of the tunnel.

[0100] Finally, the influence coefficient λ of temperature on the diffusion coefficient mainly considers the influence of temperature on the movement speed and diffusion rate of gas molecules. As the temperature increases, the movement of gas molecules speeds up, resulting in an increase in the diffusion rate. The influence of temperature on the diffusion coefficient can be quantified through experiments. In the experiment, the gas diffusion rate can be measured under different temperature conditions, usually by changing the environmental temperature and recording the gas concentration changes within a specific time period. Through regression analysis, the relationship between temperature change and gas diffusion rate can be obtained. Then, using the linear or non-linear relationship between temperature data and gas diffusion rate, the value of λ can be calculated. The influence of temperature on the diffusion coefficient is usually closely related to the physical properties of the gas and the environmental conditions of the tunnel, so it should be adjusted according to the temperature changes in practical applications.

[0101] In another preferred embodiment of the method of this embodiment, the dynamic pressure effect of the fluid is jointly determined by the density of the fluid, the flow velocity, and the stress of the tunnel. The movement of the fluid not only generates dynamic pressure but may also affect the stability of the tunnel. Calculating this effect helps to comprehensively evaluate the safety of the tunnel. The expression is:

[0102]

[0103] where, σ fluidIndicates the influence of hydrodynamic pressure on tunnel stress; Indicates fluid density; v fluid Indicates fluid velocity; β represents the influence coefficient of hydrodynamic pressure on tunnel stress; σ represents the stress of the tunnel; σ yield Indicates the yield stress (the pressure resistance limit of the tunnel material).

[0104] It should be noted that the influence coefficient β of hydrodynamic pressure on tunnel stress reflects the specific influence of the fluid motion on tunnel stress. First, the interaction between the fluid and the tunnel structure under different fluid conditions is simulated through experiments. The experiments usually include applying pressure to the tunnel model with a fluid at a certain velocity (such as water flow, air flow, etc.) in a laboratory environment. By setting sensors on the tunnel model, the change in the surface stress of the tunnel under the action of the fluid is measured. Based on these experimental data, the specific influence of hydrodynamic pressure on tunnel stress can be deduced, and the value of β can be further obtained through regression analysis.

[0105] Secondly, based on the principles of fluid mechanics, the calculation of hydrodynamic pressure effect can be quantified by flow velocity and fluid density. For tunnels with irregular shapes or different fluid conditions, numerical simulations (such as CFD simulations) can be used to analyze fluid flow and stress distribution, and then the specific value of β can be obtained. The interaction between the fluid and the tunnel is affected by factors such as fluid velocity, fluid density, and tunnel surface roughness, and these factors need to be considered through experiments or numerical models.

[0106] Then, consider the influence of different fluid conditions and tunnel geometric characteristics on β. For example, at higher flow velocities, the pressure of hydrodynamic pressure on the tunnel wall will be greater, and thus stronger stress will be generated on the tunnel. In practical applications, the type of fluid (such as water, gas or other liquids) and flow velocity may vary, so the β obtained through experiments and numerical analysis needs to be corrected according to the actual situation.

[0107] By integrating experimental data, numerical simulations, and fluid mechanics models, the influence coefficient β of hydrodynamic pressure on tunnel stress can finally be obtained.

[0108] It should be noted that the prior art usually evaluates the safety of a tunnel through a single parameter (such as stress or fluid pressure). Traditional methods mostly rely on static stress analysis or only consider the influence of a single factor (such as hydrodynamic pressure effect or gas diffusion characteristics) on the tunnel structure. However, these methods often fail to comprehensively consider multiple dynamic factors, especially the interaction of multiple external loads that a tunnel may be subjected to in a complex environment. For example, traditional methods for calculating the gas diffusion coefficient usually do not consider the influence of stress on diffusion, and the calculation of the hydrodynamic pressure effect lacks a comprehensive consideration of stress and temperature changes. Therefore, the defect of the prior art is that it cannot effectively combine multiple factors such as stress, gas diffusion, and hydrodynamic pressure for dynamic risk assessment, resulting in insufficient accuracy of the assessment results and difficulty in coping with complex and changing environmental conditions.

[0109] The method of this embodiment provides a more comprehensive and dynamic tunnel risk warning method by comprehensively considering multiple influencing factors such as tunnel stress, gas diffusion coefficient, and hydrodynamic pressure effect, and using the gradient boosting algorithm for risk prediction. By taking the gas diffusion coefficient and the hydrodynamic pressure effect as input features and training through machine learning algorithms, the system can perform real-time risk assessment based on multiple factors. This method no longer relies on a single parameter, but predicts the safety risk of the tunnel according to multi-dimensional input data, and can timely reflect the potential risks of the tunnel under different environmental and load conditions.

[0110] It should be noted that the gradient boosting algorithm is an ensemble learning method with the ability to process large-scale high-dimensional data, and can comprehensively use multiple input features for efficient prediction in a complex environment. In the present invention, the gradient boosting algorithm is mainly used to predict the safety risk of the tunnel, based on two core inputs: the gas diffusion coefficient and the hydrodynamic pressure effect, and these two inputs are respectively from the calculation results of Formula 2 and Formula 3. The gradient boosting algorithm can effectively learn how gas diffusion and hydrodynamic pressure jointly act on the risk assessment of the tunnel, and then make a judgment on the safety situation of the tunnel under different conditions.

[0111] First, the gradient boosting algorithm divides the data through a tree structure and continuously adjusts the error in each iteration to gradually optimize the accuracy of the model. Compared with traditional linear regression models or simple statistical models, the gradient boosting algorithm can automatically capture the implicit non-linear relationships and interaction effects in the data. This is crucial for tunnel dynamic risk assessment because the risk of a tunnel is not only affected by a single factor, but also by the complex interactions between multiple factors. For example, there may be an interaction between the gas diffusion coefficient and the hydrodynamic pressure effect, and traditional linear models may not be able to effectively capture these complex relationships, while the gradient boosting algorithm can automatically learn these non-linear dependence relationships through multiple iterations of optimization.

[0112] During the prediction process, the gradient boosting machine can model the complex relationship between the gas diffusion coefficient and the hydrodynamic pressure effect, solving the problem that traditional methods cannot capture multi-dimensional interaction effects. Specifically, the gradient boosting machine used in the present invention not only takes the gas diffusion coefficient and the hydrodynamic pressure effect as input features, but also combines historical tunnel risk data for training, enabling the model to adjust the prediction results in a timely manner in a dynamically changing environment.

[0113] For the results of the gas diffusion coefficient and the hydrodynamic pressure effect, traditional models often only regard these factors as independent input parameters for prediction, without considering their interaction. Through the gradient boosting machine, it is possible to learn from historical data how these parameters interact in different environments, and then make a more accurate prediction of the tunnel safety risk. This improvement ensures that the prediction results not only depend on the single change of gas diffusion and hydrodynamic pressure, but comprehensively consider the combined impact of these two factors and their relationship on tunnel safety.

[0114] In addition, the iterative optimization process of the gradient boosting machine can gradually adjust the error, making the final model more accurate. Compared with traditional static models, the gradient boosting machine continuously improves the model accuracy through multiple iterations, enabling it to dynamically adjust and provide efficient risk assessment in complex and changing real-world environments.

[0115] Furthermore, in the prior art, the risk assessment of tunnels usually depends on the influence of a single factor or uses simple statistical methods to process complex input data. Traditional regression analysis models often ignore the interaction between different factors, which makes it impossible to accurately capture the complex non-linear relationship between multiple factors when dealing with tunnel safety risk assessment. For example, the gas diffusion coefficient and the hydrodynamic pressure effect are often processed separately, without comprehensively considering their combined impact on tunnel safety. In addition, existing methods have poor processing ability for high-dimensional data and cannot effectively extract meaningful information from a large amount of complex data.

[0116] To overcome these problems, the gradient boosting machine selected in this embodiment is a machine learning method that gradually optimizes the prediction results by integrating multiple decision trees, with strong non-linear modeling ability and good high-dimensional data processing ability. In the present invention, the gradient boosting machine model takes the gas diffusion coefficient and the hydrodynamic pressure effect as features and combines historical tunnel risk data for training, and can comprehensively consider the influence of multiple factors on tunnel safety. This method can capture the complex relationship between different input features, avoiding the deficiency of relying only on a single factor in traditional methods, and can more comprehensively and accurately evaluate the tunnel safety risk.

[0117] The advantage of the gradient boosting machine model is that it can automatically adjust the weight of each feature during the training process and identify the factors that have the greatest impact on tunnel safety risks. Through multiple iterations of optimization, the gradient boosting machine can extract implicit non-linear relationships from a large amount of input data, avoiding the problem of over-simplification. In addition, by gradually reducing the error, the gradient boosting machine continuously optimizes the prediction accuracy of the model. Compared with traditional statistical methods, its prediction ability is stronger and it can adapt to a dynamically changing environment.

[0118] Compared with traditional methods, the present invention can perform risk assessment more flexibly and efficiently through the gradient boosting machine model. During the operation of the tunnel, as new data is continuously collected, the gradient boosting machine can perform online learning and incremental updates to adjust the prediction results in real time. This feature enables the model to adapt to changes in the tunnel environment and maintain high prediction accuracy in the long term. By considering the interaction and dynamic changes of multiple factors, the gradient boosting machine model provides a more scientific, accurate and real-time tunnel safety risk assessment method, which can significantly improve the safety management level of the tunnel.

[0119] In a preferred embodiment of the present invention, the gradient boosting machine algorithm (GBM) is used for dynamic risk prediction of the tunnel. Specifically, the present invention models two key factors, namely the effective gas diffusion coefficient and the hydrodynamic pressure effect of the tunnel, and combines the gradient boosting machine algorithm for training and prediction, so as to provide an accurate tunnel safety risk assessment. The whole process is described in detail below.

[0120] First of all, the input features include the calculated effective gas diffusion coefficient and hydrodynamic pressure effect respectively. The effective gas diffusion coefficient reflects the rate of gas diffusion in the tunnel, and the hydrodynamic pressure effect reflects the dynamic pressure of the fluid on the tunnel structure. These two factors are very key input features in tunnel risk assessment and can accurately reflect the safety status of the tunnel.

[0121] To improve the prediction ability of the model, the input features are optimized. The effective gas diffusion coefficient and the hydrodynamic pressure effect are first subjected to non-linear transformation. For example, the effective gas diffusion coefficient is logarithmically transformed, while the hydrodynamic pressure effect is subjected to power transformation. These transformations help to capture more complex non-linear relationships between the features and the tunnel safety risks, enhancing the adaptability of the model to changes in the tunnel environment. The logarithmic transformation of the gas diffusion coefficient and the power transformation of the hydrodynamic pressure effect make these features.

[0122] Gas diffusion coefficient D eff reflects the diffusion rate of gas in the tunnel. In practical applications, the change range of the gas diffusion coefficient may be very large, and its relationship with the tunnel safety risk is non-linear. Therefore, logarithmic transformation of it helps to enhance the fitting ability of the model.

[0123] The logarithmic transformation formula is:

[0124] D′ eff = log(D eff + 1)

[0125] where D′ eff is the effective gas diffusion coefficient after logarithmic transformation, D eff is the original effective gas diffusion coefficient, and +1 is to avoid the problem that taking the logarithm cannot be calculated when D eff is zero.

[0126] The hydrodynamic pressure effect σ fluid represents the dynamic pressure of the fluid on the tunnel wall. Usually, the flow velocity and density have a greater impact on it, and the relationship between these factors may be non-linear. Therefore, using power transformation to adjust it can better capture the impacts under different flow velocities and densities.

[0127] The power transformation formula is:

[0128]

[0129] where σ fluid is the hydrodynamic pressure effect after power transformation, σ fluid is the original hydrodynamic pressure effect, and α is the best power exponent obtained through cross-validation or other optimization methods. This exponent α controls the sensitivity of the power transformation and is usually adjusted according to the data during the training process to ensure that the model can effectively fit complex non-linear relationships.

[0130] After non-linear transformation, the two input features of the gas diffusion coefficient and the hydrodynamic pressure effect will be passed to the gradient boosting algorithm as the inputs of the model. Through feature transformation, the model can more effectively learn the complex non-linear relationships between these features and the tunnel safety risk, and improve the prediction accuracy.

[0131] Input features:

[0132]

[0133] These transformed features will be used as inputs and passed into the training data so that the gradient boosting machine can accurately predict the tunnel safety risk assessment value by learning the patterns in the data.

[0134] Next, the training process based on the gradient boosting algorithm begins. During the training stage, the goal of the model is to minimize the error between the predicted safety risk assessment value and the true value. During training, the loss function used is the weighted mean squared error, and its calculation formula is:

[0135]

[0136] where N is the number of samples, and y pred,i is the predicted risk assessment value of the i-th sample, and y true,i is the true risk assessment value of the i-th sample, and w i is the weight of the sample. In the present invention, higher weights are set for high-risk events, so that the model pays more attention to high-risk samples.

[0137] To prevent the model from overfitting, multiple regularization techniques are adopted during the training process. First, by restricting the depth of each tree to avoid excessive growth of the tree, this effectively prevents the model from overfitting to the training set. Second, by ensuring that each node has enough samples for splitting through the minimum sample splitting parameter, this further reduces the risk of overfitting. At the same time, to ensure the stability of the model, L2 regularization is used to limit the influence of each feature, thereby avoiding the model's over-reliance on certain features.

[0138] In each round of training, the gradient boosting machine algorithm updates the prediction result of the model by calculating the residuals. In each iteration, the algorithm adjusts the weights of each tree according to the prediction error of the previous round, so as to improve the fitting ability of the model to the target variable. Each new decision tree helps the previous tree correct errors by fitting the residuals, and finally realizes the accurate prediction of the target value.

[0139] After the training is completed, the trained model is used to predict the risk of new tunnel data. For the new tunnel data, the input features include the gas diffusion coefficient and hydrodynamic pressure effect data collected in real time. The gradient boosting machine algorithm will predict the safety risk assessment value of the tunnel by bringing these input features into the trained model.

[0140] Embodiment 2 is the second embodiment of the present invention, which is different from the previous embodiment in that:

[0141] If the said function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that makes a contribution to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. And the aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0142] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definable list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0143] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection portion having one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, a computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then storing it in a computer memory.

[0144] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0145] Embodiment 3, the third embodiment of the present invention, provides a tunnel dynamic risk warning system based on an intelligent algorithm, including.

[0146] A data acquisition module, configured to collect in real time soil pressure, gas pressure, fluid density, flow rate data, and temperature data in the tunnel through sensors, obtain soil humidity, soil compactness, soil type, and soil temperature data, obtain external load data on the tunnel through a load monitoring device, and obtain geometric dimension data of the tunnel;

[0147] A stress analysis module, which is used to calculate the stress data of each part of the tunnel using a structural mechanics model based on external load data, gas pressure, soil pressure, and temperature data;

[0148] A diffusion analysis module, which is used to calculate the effective gas diffusion coefficient based on the stress data and temperature data;

[0149] A dynamic pressure analysis module, which is used to calculate the fluid dynamic pressure effect based on the stress data, fluid density, and flow velocity data;

[0150] A risk warning module, which is used to input the effective gas diffusion coefficient and the fluid dynamic pressure effect into an intelligent algorithm, evaluate the safety risk of the tunnel, generate a warning signal according to a preset risk threshold, and trigger a risk warning notification.

[0151] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A tunnel dynamic risk early warning method based on intelligent algorithms, characterized in that: including Collecting soil pressure, gas pressure, fluid density, flow rate data and temperature data in the tunnel in real time through sensors to obtain soil humidity, soil compaction, soil type and soil temperature data, obtaining external load data on the tunnel through a load monitoring device, and obtaining geometric dimension data of the tunnel; Calculating stress data of each part of the tunnel using a structural mechanics model based on external load data, gas pressure, soil pressure and temperature data; Calculating the effective gas diffusion coefficient based on stress data and temperature data; Calculating the fluid dynamic pressure effect based on stress data, fluid density and flow rate data; Inputting the effective gas diffusion coefficient and the fluid dynamic pressure effect into an intelligent algorithm to evaluate the safety risk of the tunnel, generating a warning signal according to a preset risk threshold, and triggering a risk warning notification.

2. The method for dynamically warning tunnel risks based on an intelligent algorithm according to claim 1, wherein: Before calculating the stress data of each part of the tunnel, it also includes Calculating a preliminary value of the soil reaction coefficient using a regression model or machine learning method based on soil humidity, soil compaction, soil type and external load data; Adjusting the preliminary value of the soil reaction coefficient according to the soil temperature data to obtain a corrected soil reaction coefficient; Performing a secondary correction on the corrected soil reaction coefficient according to historical load data, considering the long-term influence of historical loads on soil compaction, and determining the final soil reaction coefficient.

3. The method for dynamic risk early warning of tunnels based on intelligent algorithms according to claim 2, characterized in that: The calculating the stress data of each part of the tunnel using the structural mechanics model includes Obtaining external load, gas pressure, soil pressure, soil reaction coefficient and temperature data, as well as the cross-sectional area of the tunnel; Using a structural mechanics model to calculate the overall stress of the tunnel according to the external load and gas pressure; Correcting the overall stress based on the soil pressure and the soil reaction coefficient to obtain the local stress; Combining the local stress and the geometric dimension data to obtain the final stress value of each part of the tunnel.

4. The dynamic risk early warning method for tunnels based on intelligent algorithms according to claim 3, wherein: The calculating the effective gas diffusion coefficient includes Calculating a preliminary gas diffusion coefficient based on the tunnel stress data and temperature data; Correcting the gas diffusion coefficient according to the stress data to obtain a corrected diffusion coefficient; Further adjusting the gas diffusion coefficient using the temperature data to obtain the final effective gas diffusion coefficient.

5. The dynamic risk early warning method for tunnels based on intelligent algorithms according to claim 4, characterized in that: The calculating the fluid dynamic pressure effect includes Obtaining tunnel stress data, fluid density and flow rate data, and inputting them into a fluid mechanics model for calculation; Calculating the dynamic pressure effect of the fluid on the tunnel wall based on the fluid density and flow rate data; Correcting the dynamic pressure effect according to the tunnel stress data to obtain the final corrected result.

6. The dynamic risk early warning method for tunnels based on intelligent algorithms according to claim 5, characterized in that: The intelligent algorithm includes Selecting the gradient boosting machine algorithm and modeling with the gas diffusion coefficient and the fluid dynamic pressure effect as input features; Training the gradient boosting machine model using historical tunnel risk data including historical gas diffusion coefficient, fluid dynamic pressure effect and corresponding risk assessment values, and learning the relationship between input features and risk assessment values through the historical tunnel risk data; Using the trained gradient boosting machine model to perform risk prediction based on the gas diffusion coefficient and fluid dynamic pressure effect data to obtain the safety risk assessment value of the tunnel.

7. The method for dynamically warning of tunnel risks based on intelligent algorithms according to claim 6, wherein: The evaluating the safety risk of the tunnel includes Comparing the safety risk assessment value with the threshold to classify the risk level as low risk, medium risk and high risk; Generate warning signals according to the risk level and initiate safety response measures.

8. A tunnel dynamic risk early warning system based on an intelligent algorithm, which applies an intelligent algorithm-based tunnel dynamic risk early warning method as described in any one of claims 1 to 7, characterized in that, Including: A data acquisition module for real-time collecting soil pressure, gas pressure, fluid density, flow velocity data and temperature data inside the tunnel through sensors, obtaining soil humidity, soil compactness, soil type and soil temperature data, obtaining external load data on the tunnel through a load monitoring device, and obtaining geometric dimension data of the tunnel; A stress analysis module for calculating stress data of each part of the tunnel using a structural mechanics model based on external load data, gas pressure, soil pressure and temperature data; A diffusion analysis module for calculating the effective gas diffusion coefficient based on stress data and temperature data; A dynamic pressure analysis module for calculating the fluid dynamic pressure effect based on stress data, fluid density and flow velocity data; A risk warning module for inputting the effective gas diffusion coefficient and the fluid dynamic pressure effect into an intelligent algorithm, evaluating the safety risk of the tunnel, generating warning signals according to a preset risk threshold, and triggering a risk warning notification.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of a method for dynamic risk warning of a tunnel based on an intelligent algorithm as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of a method for dynamic risk warning of a tunnel based on an intelligent algorithm as described in any one of claims 1 to 7.

Images