A tbm jamming prediction and early warning method based on threshold calibration and dynamic probability calculation

Abstract

The present application relates to the technical field of tunnel excavation construction safety, and proposes a TBM jam prediction and early warning method based on threshold calibration and dynamic probability calculation, which comprises the following steps: first, a moment balance criterion is established at the cutter head end, a thrust-resistance balance criterion is established at the shield body end, and the collapse amount parameter is taken as an unknown variable, and the critical collapse height is obtained by step-by-step integration and analytical solution. Then, a multiplication deviation factor is introduced to correct the mechanical term, the deviation distribution is updated online through the Bayesian method, and finally the failure probability is statistically calculated by Monte Carlo simulation, and compared with the predetermined intolerable failure probability threshold to realize real-time risk early warning. The present application can quantify the jam risk, dynamically correct the prediction results, ensure that the risk identification accuracy rate reaches more than 80%, and provide timely and reasonable intervention decision basis for construction management.

Description

Technical Field

[0001] This invention belongs to the field of tunnel excavation construction safety technology, specifically involving a TBM card machine prediction and early warning method based on threshold calibration and dynamic probability calculation. Background Technology

[0002] Tunnel boring machines (TBMs) are widely used in urban subways, long-distance tunnels, water conservancy and hydropower projects, and mining. However, under complex geological conditions, TBMs often face the risk of "jamming" during their advancement. This occurs when the machine is blocked or even completely stopped due to factors such as landslides, surrounding rock deformation, excessive cutterhead resistance, or a sharp increase in shield friction. Jamming not only causes significant delays in the construction cycle but can also lead to equipment damage, accelerated cutter wear, shield attitude deviation, and even construction safety accidents, resulting in severe economic losses and social impact.

[0003] Existing TBM jamming risk warning technologies mainly fall into two categories: one is based on empirical thresholds, which monitors indicators such as thrust, cutterhead torque, and penetration depth and compares them with manually set thresholds, issuing a warning when an indicator exceeds a certain empirical value; the other is based on numerical simulation, which performs offline simulation of the tunneling process to assess the stress conditions under different geological and thrust parameters. Both methods have significant shortcomings: the empirical threshold method relies on the experience of construction personnel and cannot reflect the randomness and uncertainty of working conditions, resulting in overly conservative or delayed warnings and frequent false alarms and missed alarms; while the numerical simulation method can reflect the mechanical mechanism, it usually requires a large amount of manual parameter input, making it difficult to run in real time on the construction site, and the simulation results do not consider the statistical distribution of parameters, cannot provide failure probabilities, and lack quantitative risk standards.

[0004] Furthermore, existing studies often assume that geological and equipment parameters are deterministic values, failing to adequately consider the randomness of surrounding rock properties, contact friction, in-situ stress, and thrust and torque measurement errors. This leads to systematic biases in the prediction results. Parameter calibration often employs single regression or manual correction, making continuous updates difficult as construction progresses. When geological conditions change or equipment performance fluctuates, the accuracy of model predictions decreases significantly, failing to meet the safety and real-time requirements of long-distance tunneling projects.

[0005] In summary, existing technologies lack a probabilistic risk prediction method that can both consider parameter uncertainties and be updated in real time with on-site data; at the same time, there is a lack of a clear "intolerable failure probability threshold" to guide construction intervention decisions, resulting in a lack of quantitative standards for risk management. Summary of the Invention

[0006] This invention aims to provide a TBM jamming prediction and early warning method based on threshold calibration and dynamic probability calculation, which solves the problems of lack of dynamic correction of jamming criteria, large deviation between prediction results and actual working conditions, and inability to quantify failure probability in the prior art. It realizes real-time quantitative assessment of jamming risk and achieves an identification accuracy rate of over 80%.

[0007] The technical solution of the present invention is as follows:

[0008] A predictive early warning method for TBM (Traffic Machine) cassette players based on threshold calibration and dynamic probability calculation includes a pre-stage and a real-time stage, specifically including:

[0009] (1) Preliminary stage:

[0010] Establish a jamming criterion equation, which includes the cutterhead torque balance equation and the shield thrust-resistance balance equation. The collapse amount is used as an unknown criterion variable, which is the collapse height or the equivalent height of the collapsed body volume.

[0011] A multiplication deviation factor is introduced to correct the cutterhead torque balance equation and the shield thrust-resistance balance equation, and the surrounding rock physical parameters, contact parameters and equipment parameters are modeled as random variables;

[0012] Historical data is obtained, and N sets of parameter samples are generated using Monte Carlo sampling. For each set of samples, the modified card criterion equation is solved, and the threshold of intolerable failure probability is statistically obtained.

[0013] (2) Real-time phase:

[0014] Real-time data collection at the construction site; solving the jamming criterion equation under the current working conditions; and statistically obtaining the real-time failure probability.

[0015] The real-time failure probability is compared with the intolerable failure probability threshold. If the real-time failure probability is greater than or equal to the intolerable failure probability threshold, an early warning is triggered and an intervention plan is output.

[0016] Furthermore, the multiplication deviation factor is used to correct the resistance-side mechanical terms in the cutterhead torque balance equation and the shield thrust-resistance balance equation. The resistance-side mechanical terms include at least the cutterhead front friction term, the cutterhead side friction term, the rock breaking cutting term, and the shield pressure-friction term. The multiplication deviation factor is set as a random variable that follows a normal distribution.

[0017] Furthermore, the modified card criterion equation is specifically as follows:

[0018] ,

[0019] ,

[0020] in, The driving torque of the cutter head, , , For correction factor, H ′ represents the collapse height. , , For torque, This refers to the coefficient of friction between subsequent equipment and the track. For the friction of the shield body, W The total weight of the machine. For correction factor, The collapse pressure is related to the collapse height.

[0021] Furthermore, the historical data consists of sample data from no fewer than 100 instances of "collapse causing machine jamming," and includes surrounding rock physical parameters, contact parameters, equipment parameters, and corresponding collapse data. Before adopting Monte Carlo sampling, observation residuals are constructed based on historical data, and the multiplicative bias factor is updated using a normal-normal conjugate form to obtain the posterior distribution of the bias factor. The measurement noise variance is estimated through the sample variance of the initial observation residual samples.

[0022] Furthermore, the process for determining the intolerable failure probability threshold is as follows:

[0023] Based on the Monte Carlo sampling solution of the modified jamming criterion equation, the failure probability under different collapse amounts is obtained. The 80th quantile of the failure probability distribution is calculated, and this quantile is the intolerable failure probability threshold. The criterion for determining the failure probability is the critical collapse amount that causes the jamming.

[0024] Furthermore, the construction site data includes data on propulsion thrust, cutterhead torque, penetration depth, displacement, and collapse. Before solving the jamming criterion equation under the current working conditions in the real-time stage, the real-time collected construction site data is converted into observation residuals. The Bayesian update process of the multiplication deviation factor in the pre-stage is repeated to continuously update the distribution of surrounding rock physical properties, contact parameters, and equipment parameters, as well as the posterior distribution of the deviation factor.

[0025] Furthermore, the a priori distribution of the surrounding rock physical properties includes: unit weight, internal friction angle, cohesion, geostress level, fault dip angle or width; the equipment parameters include at least: maximum thrust, conventional propulsion thrust, cutterhead torque, cutterhead rotation speed, and propulsion speed.

[0026] Furthermore, the intervention schemes include: slowing down or pausing the advance, increasing the support / grouting pressure, adjusting the cutterhead torque and speed, modifying / replacing the cutter, pre-reinforcing or pre-grouting, and adjusting the advance posture or correction strategy.

[0027] Compared with the prior art, the present invention has the following advantages:

[0028] This invention establishes a jamming criterion equation that includes cutterhead torque balance and shield thrust-resistance balance, and introduces a multiplicative deviation factor to correct the resistance-side mechanical terms. It then uses Bayesian updates of the deviation factor based on historical sample data to reduce parameter uncertainty. Monte Carlo or Latin hypercube sampling is used to solve the criterion equation and statistically determine the failure probability. Simultaneously, an intolerable failure probability threshold is determined using the 80th percentile of at least 100 jamming samples. During the construction phase, data such as propulsion thrust and cutterhead torque are collected in real time, and the parameter distribution and failure probability are continuously updated and compared with the threshold. When the threshold is exceeded, an intervention plan is output. This achieves both quantitative assessment and dynamic correction of jamming risk, effectively improving the risk identification accuracy to over 80%, and providing timely and reasonable intervention decision-making basis for construction management. It solves the problems of existing technologies relying on experience thresholds, which are prone to false alarms and omissions, and numerical simulations that are difficult to perform in real time and cannot quantify failure probabilities, significantly ensuring the safety and efficiency of tunnel excavation. Attached Figure Description

[0029] The accompanying drawings illustrate various embodiments generally by way of example rather than limitation, and are used, together with the specification and claims, to explain embodiments of the invention. Where appropriate, the same reference numerals are used in all drawings to refer to the same or similar parts. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the apparatus or method.

[0030] Figure 1 is a flowchart of the proposed method in an embodiment of the present invention;

[0031] Figure 2 is a schematic diagram of the pre-stage aspect in an example of the present invention;

[0032] Figure 3 is a schematic diagram of the real-time stage in an example of the present invention. Detailed Implementation

[0033] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0034] like Figure 1-Figure 3 As shown, this invention provides a method for predicting and intervening in TBM card machine risks based on probabilistic modeling and Bayesian updating, including a pre-stage and a real-time stage:

[0035] 1. Pre-stage (Offline calibration stage)

[0036] Collect no fewer than 100 historical data points of "collapse volume causing machine jamming", model the input parameters such as surrounding rock properties, friction coefficient, ground stress, equipment thrust and torque as a priori distribution, and establish a jamming criterion equation that includes cutterhead torque balance and shield thrust-resistance balance.

[0037] First, consider the frictional torque on the front of the cutterhead. Treat the front of the cutterhead as a radius. The circular surface, using polar coordinates Describe the contact area, with surface elements as Its tangential friction strength is given by the Coulomb criterion: The torque exerted by this infinitesimal element on the center of the cutter head is:

[0038] ,

[0039] in, r , i These are radial and angular coordinates in polar coordinates. t For tangential friction strength, dA Let the area be a microelement. This refers to the cohesion between the cutterhead and the soil / rock mass. The coefficient of friction between the cutterhead and the soil / rock mass. Normal stress;

[0040] Considering the cutter head opening ratio Integrating the front friction torque, we get:

[0041] ,

[0042] in, This refers to the cohesion between the cutterhead and the soil / rock mass. The coefficient of friction between the cutterhead and the soil / rock mass. For normal stress, R Where is the radius of the cutter head.

[0043] right The integral yields:

[0044] ,

[0045] make Simplified to:

[0046] ,

[0047] in, D The diameter is the cutter head.

[0048] The average normal stress at the front is calculated using passive earth pressure:

[0049] ,

[0050] ,

[0051] This is the passive earth pressure coefficient. The weight of the broken parts For the cohesive force of the fragmented body, q To loosen the load, For passive earth pressure, The height of the collapse. This represents the loosening load factor.

[0052] Next, consider the frictional torque on the side of the cutter head, treating the side of the cutter head as a circular annular sidewall, and the contact height. Circumferential angle domain The frictional resistance of the micro-element is Its lever arm is approximately taken as The integral of the lateral moment is:

[0053] ,

[0054] in, The height of the cutter head side. This represents lateral earth pressure.

[0055] The distribution of lateral earth pressure is as follows:

[0056] ,

[0057] right Integral result:

[0058] ,

[0059] thereby:

[0060] ,

[0061] in, This is the reduction factor.

[0062] The rock-breaking cutting torque is obtained by empirical integration:

[0063] ,

[0064] in, It represents the uniaxial compressive strength of the fractured rock mass.

[0065] Substituting the above terms into the torque balance:

[0066] ,

[0067] After sorting, we can get information about The quadratic equation:

[0068] ,

[0069] The coefficients are:

[0070] ,

[0071] ,

[0072] ,

[0073] in, This is the driving torque of the cutter head.

[0074] The critical collapse height is obtained by solving:

[0075] ,

[0076] At the shield end, the thrust-drag balance equation is:

[0077] ,

[0078] in, , , , For the friction of the shield body, W This refers to the total weight of the machine.

[0079] The critical collapse height of the shield body was determined as follows:

[0080] ,

[0081] in, The length of the shield body. The diameter of the shield body section. This is the coefficient of friction between the subsequent equipment and the track.

[0082] Introducing a multiplication deviation factor for the resistance side mechanical terms It follows a normal distribution and is used to correct the difference between theoretical and actual values. This is achieved by calculating the observed residuals. Perform a Bayesian update on the bias factor distribution to obtain a convergent posterior distribution.

[0083] To account for the differences between the model and actual measurements, this invention introduces a multiplication deviation factor before the resistance term. :

[0084] ,

[0085] ,

[0086] in, Observation residuals are constructed using monitoring data:

[0087] ,

[0088] in, These are measured values. For predicted values, t This is a time parameter.

[0089] Suppose the likelihood Update using normal-normal conjugate:

[0090] ,

[0091] ,

[0092] in, , For the prior mean and variance, , For the posterior mean and variance n For the number of observations, This represents the t-th observation. This represents the variance of a single observation.

[0093] Monte Carlo sampling was performed based on the updated parameter distribution and deviation factor distribution to solve the card criterion equation and calculate the failure probability. And determine the threshold for the probability of intolerable failure under an 80% accuracy condition. This serves as a standard for risk assessment during the construction phase.

[0094] After obtaining the updated parameter distribution, it is generated through Monte Carlo sampling. A set of parameter samples, each set obtained by solving the above equation. Determine if it has failed and calculate the failure probability:

[0095] ,

[0096] in, N For Monte Carlo sampling number, This is the sample failure indicator function.

[0097] In the preliminary stage, at least 100 samples of machine jamming caused by collapse were collected, the empirical distribution was calculated, and the 80th quantile was taken as the threshold for the probability of intolerable failure.

[0098] ,

[0099] in, Quantities.

[0100] 2. Real-time phase (construction and operation phase)

[0101] During construction, real-time monitoring data such as propulsion thrust, cutterhead torque, penetration depth, and displacement are collected, and the new observations are converted into observation residuals. A rolling Bayesian update is performed on the posterior distribution of the bias factor to ensure that the model parameters continue to converge.

[0102] Monte Carlo sampling is performed based on the latest posterior distribution within each construction time window to solve the jamming criterion equation under the current working condition, directly obtaining the failure probability. and compared with the threshold obtained in the pre-stage. Compare.

[0103] when When the machine jams, the system triggers an early warning and automatically generates an intervention plan, including measures such as slowing down or pausing the advance, increasing the support or grouting pressure, adjusting the cutterhead torque and speed, advancing the grouting, changing the cutter or correcting the deviation, so as to take proactive control before the machine jams.

[0104] Intervention measures include, but are not limited to, slowing down the advance, increasing support or grouting pressure, adjusting cutterhead torque and speed, replacing or modifying cutters in advance, reinforcing the strata ahead of time, and adjusting the advance posture or correction strategy. If the prediction results for multiple consecutive time windows indicate that the risk remains at a high level, the system will recommend suspending tunneling until engineering measures are taken to reduce the risk.

[0105] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A TBM (Traffic Machine Controller) predictive early warning method based on threshold calibration and dynamic probability calculation, characterized in that, It includes a pre-phase and a real-time phase, specifically including: (1) Preliminary stage: Establish a jamming criterion equation, which includes the cutterhead torque balance equation and the shield thrust-resistance balance equation. The collapse amount is used as an unknown criterion variable, which is the collapse height or the equivalent height of the collapsed body volume. The torque balance equation is: ， in, The driving torque of the cutter head, H ′ represents the collapse height. For torque; The thrust-drag balance equation is as follows: ， in, For the friction of the shield body, W The total weight of the machine. This refers to the coefficient of friction between subsequent equipment and the track. Let be the coefficient of friction of the cutter head. H ′ represents the collapse height; A multiplication deviation factor is introduced to correct the cutterhead torque balance equation and the shield thrust-resistance balance equation, and the surrounding rock physical parameters, contact parameters and equipment parameters are modeled as random variables; The modified card criterion equation is as follows: ， in, The driving torque of the cutter head, For correction factor, H ′ represents the collapse height. For torque, This refers to the coefficient of friction between subsequent equipment and the track. For the friction of the shield body, W The total weight of the machine. For correction factor, The collapse pressure related to the collapse height, The coefficient of friction of the cutter head; Historical data is obtained, and N sets of parameter samples are generated using Monte Carlo sampling. For each set of samples, the modified card criterion equation is solved, and the threshold of intolerable failure probability is statistically obtained. The historical data consists of sample data from no fewer than 100 instances of collapse leading to machine jamming, and includes surrounding rock physical parameters, contact parameters, equipment parameters, and corresponding collapse data. Before adopting Monte Carlo sampling, observation residuals are constructed based on historical data, and the multiplicative bias factor is updated using a normal-normal conjugate form in a Bayesian manner to obtain the posterior distribution of the bias factor. The measurement noise variance is estimated by the sample variance of the initial observation residual samples. (2) Real-time phase: Real-time data collection at the construction site; solving the jamming criterion equation under the current working conditions; and statistically obtaining the real-time failure probability. The construction site data includes propulsion thrust, cutterhead torque, penetration depth, displacement and collapse data; before solving the jamming criterion equation under the current working condition in the real-time stage, the real-time collected construction site data is converted into observation residuals, and the Bayesian update process of the multiplication deviation factor in the pre-stage is repeated to continuously update the distribution of surrounding rock physical properties, contact parameters and equipment parameters and the posterior distribution of deviation factors. The real-time failure probability is compared with the intolerable failure probability threshold. If the real-time failure probability is greater than or equal to the intolerable failure probability threshold, an early warning is triggered and an intervention plan is output.

2. The TBM card machine prediction and early warning method based on threshold calibration and dynamic probability calculation according to claim 1, characterized in that, The multiplication deviation factor is used to correct the resistance-side mechanical terms in the cutterhead torque balance equation and the shield thrust-resistance balance equation. The resistance-side mechanical terms include at least the cutterhead front friction term, the cutterhead side friction term, the rock breaking cutting term, and the shield pressure-friction term. The multiplication deviation factor is set as a random variable that follows a normal distribution.

3. The TBM card machine prediction and early warning method based on threshold calibration and dynamic probability calculation according to claim 1, characterized in that, The process for determining the intolerable failure probability threshold is as follows: The failure probability under different collapse amounts is obtained by solving the modified jamming criterion equation based on Monte Carlo sampling. The 80th quantile of the failure probability distribution is calculated to represent the threshold of intolerable failure probability. The criterion for determining the failure probability is the critical collapse amount that causes the jamming.

4. The TBM card machine prediction and early warning method based on threshold calibration and dynamic probability calculation according to claim 1, characterized in that, The a priori distribution of the surrounding rock physical properties includes: unit weight, internal friction angle, cohesion, geostress level, fault dip angle or width; the equipment parameters include: maximum thrust, conventional propulsion thrust, cutterhead torque, cutterhead rotation speed and propulsion speed.

5. The TBM card machine prediction and early warning method based on threshold calibration and dynamic probability calculation according to claim 1, characterized in that, The intervention schemes include: slowing down or pausing the advance, increasing the support / grouting pressure, adjusting the cutterhead torque and speed, modifying / replacing the cutter, pre-reinforcing or advanced grouting, and adjusting the advance posture or correction strategy.

Images