CN-122022067-A - Grouting control method based on stratum stability evaluation

CN122022067ACN 122022067 ACN122022067 ACN 122022067ACN-122022067-A

Abstract

The application discloses a grouting control method based on stratum stability assessment, which relates to the technical field of shield construction grouting, and comprises the following steps of S1, obtaining multisource information in stratum and shield construction process, S2, training a time sequence prediction model by taking simulation data as a supplementary training set, S3, predicting stratum response according to static parameter data and dynamic detection data through the trained time sequence prediction model, calculating stability index S and risk probability P based on predicted values, S4, constructing a grouting strategy rule base, generating a grouting strategy, and grouting according to the grouting strategy, constructing an assessment model of the stability index and the risk probability, realizing stratum stability quantification and predictability, providing scientific basis for construction decisions, and simultaneously establishing a multistage grouting strategy rule base, and being capable of rapidly selecting a corresponding grouting scheme under different risk grades, thereby effectively preventing stratum instability or excessive disturbance.

Inventors

LI YONGPO
ZHOU RENJUN
HUANG SHOUHUI
ZHANG LIYUAN
LI YU
YU FU
WANG YOU
HU MEIQI
WANG YUNFENG
ZHANG LI
Hou Lide
ZHOU HUABO
WANG JINLONG
YIN SEN
DENG JUN
LIU JUN

Assignees

中国交通建设股份有限公司轨道交通分公司
中交隧道工程局有限公司
中交第二公路工程局有限公司
中交第一航务工程局有限公司
中南大学

Dates

Publication Date: 20260512
Application Date: 20260310

Claims (9)

1. A grouting control method based on stratum stability evaluation is characterized by comprising the following steps: s1, acquiring multi-source information in stratum and shield construction processes, wherein the multi-source information comprises static parameter data, dynamic detection data and simulation data; S2, training a time sequence prediction model by taking the simulation data as a supplementary training set; s3, predicting stratum response by a trained time sequence prediction model according to the static parameter data and the dynamic detection data, and calculating a stability index S and a risk probability P based on a predicted value; S4, constructing a grouting strategy rule base, generating a grouting strategy according to the grouting strategy rule base, the stability index S and the risk probability P, and grouting according to the grouting strategy.
2. The grouting control method based on stratum stability assessment according to claim 1, wherein the static parameter data in the step S1 comprise stratum types, soil compactness, permeability coefficients, groundwater levels, stratum elastic modulus, undisturbed stress and shield tunneling machine structural parameters, and the dynamic detection data comprise soil pressure, displacement, sedimentation, pore water pressure, seepage change, stratum vibration, shield tunneling machine propelling force, torque and cutter head rotating speed.
3. The grouting control method based on stratum stability assessment according to claim 2, wherein the simulation data are data samples based on stratum-grouting coupling numerical models, simulated different geological conditions and extreme construction conditions, including vertical displacement differences, chamber convergence, effective stress changes, pore pressure changes, plastic region proportions and grouting diffusion ranges, and are used for expanding a training data set.
4. The grouting control method based on stratum stability assessment according to claim 3, wherein in step S3, the time sequence prediction model adopts a mixed LSTM architecture of static-dynamic characteristic splicing, static parameter data is firstly encoded into a low-dimensional characteristic vector through a full-connection layer and then copied in a time dimension, the low-dimensional characteristic vector is spliced with dynamic monitoring data of each time step and is input into the LSTM layer together, a training set 70%, a verification set 15% and a test set 15% are divided, a mean square error is used as a loss function, an Adam optimizer is used for optimization, and finally a stability index S and a risk probability P are output.
5. The grouting control method based on stratum stability evaluation of claim 4, wherein when calculating stability index S, several key response values are normalized, and normalized instability of each index is defined Wherein, the In order to predict the response quantity, For the purpose of reference to the normal value, For designing allowable limits or empirical thresholds, when When >1, the index exceeds the allowable value, and the instability risk exists; Defining a stability index S: Wherein, the The weight of each index is determined empirically and meets the constraint 。
6. The method of claim 5, wherein the risk probability P is calculated based on the stability index S by the following formula: Wherein P is risk probability, the value interval is 0-1, k is the coefficient of the steepness degree of the control curve; is a central value of risk sensitivity.
7. The grouting control method according to claim 6, wherein the grouting strategy rule base comprises four-level rules, The first-level rule is that the stability grade of the current working condition is determined according to the stability index S and the risk probability P, and the specific form is as follows: r1 is that if S < S1 and P < P1, the grade is stable; r2 is that if S1 is less than or equal to S2 or P1 is less than or equal to P2, the grade is light early warning; r3, if S2 is less than or equal to S3 or P2 is less than or equal to P3, the grade is serious early warning; R4 is unstable if S is greater than or equal to S3 or P is greater than or equal to P3. S1, S2 and S3 are stability index thresholds, P1, P2 and P3 are risk probability thresholds, and are determined empirically; Considering that certain extremely monitored values may need to be processed preferentially, setting a secondary correction rule, and correcting the result of a primary rule, wherein the specific form is as follows: r5, if the sedimentation increment in unit time exceeds a set limit value, the grade is increased by at least one grade; r6, if the soil pressure fluctuates severely in a short time, the grade is increased to be higher than the serious early warning; r7, if the pore water pressure is close to or exceeds the seepage damage control value, the grade is adjusted to be unstable; The three-level rule is to generate a grouting strategy according to the level, and the concrete form is as follows: r8, if the grade is stable, only maintaining a monitoring strategy, and not starting grouting; r9, if the grade is mild early warning, generating a preventive grouting strategy, and selecting a mild reinforcement mode, such as controlling a grouting area or moderately improving grouting frequency; r10, if the grade is serious early warning, generating an enhanced grouting strategy, expanding a grouting range, enhancing a reinforcing effect and tracking a grouting process in real time; R11, if the grade is unstable, generating an emergency grouting strategy, and adopting a rapid and strong grouting mode to preferentially control deformation and seepage risks; the fourth-level rule is based on data feedback in the grouting process, and adjusts the grouting strategy, and the concrete form is as follows: R12, if grouting pressure is continuously increased and flow is obviously reduced in the grouting process, judging that the soil body is gradually compact, and properly reducing grouting strength or finishing grouting in advance; r13, if the grouting pressure is lower but the flow is abnormally increased, judging that the slurry is too fast to diffuse or a leakage channel exists, and adjusting a grouting mode or increasing grouting points; and R14, if the recalculated stability index S after grouting is insufficient in amplitude reduction or the risk probability P is not reduced to a safe range, triggering a supplementary grouting strategy and regenerating a grouting scheme.
8. The method for controlling grouting based on formation stability evaluation according to claim 7, further comprising step S5, S5, after grouting is completed, stratum monitoring data including displacement, soil pressure, pore water pressure, seepage, vibration response and the like are continuously collected, and are compared and analyzed with the state before grouting, and the stability index S and the risk probability P are recalculated to form a quantized grouting effect evaluation result; if the stability index S is obviously improved and the risk probability P is reduced to a safe range, the grouting effect is judged to be effective, and if the stability is not improved enough or the key index still has an overrun trend, secondary strategy analysis and adjustment are carried out to generate a reinforced grouting scheme, grouting operation is carried out again, and the continuous maintenance of the stratum stability is realized.
9. The method for controlling grouting based on formation stability evaluation according to claim 8, further comprising step S6, S6, comparing the actual grouting effect with the predicted effect to form closed-loop data feedback, providing reference for the grouting strategy of the next stage, updating the parameters or weight of the time sequence prediction model, and optimizing the prediction capacity of the time sequence prediction model.

Description

Grouting control method based on stratum stability evaluation Technical Field The invention relates to the technical field of shield construction grouting, in particular to a grouting control method based on stratum stability evaluation. Background The shield construction is a main construction method for underground engineering construction of the subway, the tunnel and the like in the current city. In the shield tunneling process, the stability of the excavated surface and surrounding stratum is ensured, and the method is a core challenge of ensuring construction safety, controlling earth surface subsidence and protecting surrounding building (construction) structures. The synchronous grouting is used as a key working procedure of shield tunneling, and is used for timely filling an annular gap (shield tail gap) formed after shield tail separation so as to support a segment, stabilize a stratum and prevent water and soil loss and excessive settlement. However, existing synchronous grouting control techniques have significant limitations in practice. First, grouting decisions rely excessively on human experience, lacking in quantitative assessment of formation status. In the current construction, grouting parameters are preset according to geological investigation reports and experience of engineers, and are roughly adjusted according to limited monitoring point data in the construction. Because of the lack of fusion analysis and intelligent prediction of multi-source real-time information, grouting operation is often in a passive response state rather than an active control state, and accurate intervention cannot be performed before risk occurs. Most of the existing grouting systems adopt open-loop or simple closed-loop control, and the set value is usually a fixed value or is adjusted in stages according to the number of tunneling rings. The grouting is insufficient, the slurry cannot effectively fill gaps, and the surface subsidence exceeding, segment staggering and even water leakage are caused, or the grouting is excessive or the pressure is excessive, so that the segment is unevenly loaded, floats upwards or deforms, or the slurry splits the stratum, wastes materials and possibly causes unnecessary disturbance to the surrounding environment. Disclosure of Invention The grouting control method based on stratum stability evaluation is provided for improving the safety of shield tunneling construction, reducing construction risk and improving grouting control precision and construction efficiency. The application provides a grouting control method based on stratum stability evaluation, which adopts the following technical scheme: A grouting control method based on formation stability evaluation comprises the following steps: s1, acquiring multi-source information in stratum and shield construction processes, wherein the multi-source information comprises static parameter data, dynamic detection data and simulation data; S2, training a time sequence prediction model by taking the simulation data as a supplementary training set; s3, predicting stratum response by a trained time sequence prediction model according to the static parameter data and the dynamic detection data, and calculating a stability index S and a risk probability P based on a predicted value; S4, constructing a grouting strategy rule base, generating a grouting strategy according to the grouting strategy rule base, the stability index S and the risk probability P, and grouting according to the grouting strategy. Optionally, the static parameter data in the step S1 comprises stratum category, soil compactness, permeability coefficient, groundwater level, stratum elastic modulus, undisturbed stress and shield tunneling machine structural parameters, and the dynamic detection data comprises soil pressure, displacement, sedimentation, pore water pressure, seepage change, stratum vibration, shield tunneling machine propelling force, torque and cutter head rotating speed. Optionally, the simulation data is based on a stratum-grouting coupling numerical model, and simulated data samples including different geological conditions and extreme construction conditions, such as vertical displacement difference, chamber convergence, effective stress change, hole pressure change, plastic region proportion, grouting diffusion range and the like are used for expanding the training data set. Optionally, in step S3, the time sequence prediction model adopts a hybrid LSTM architecture of static-dynamic feature concatenation, static parameter data is first encoded into a low-dimensional feature vector through a full-connection layer, then copied in a time dimension, and spliced with dynamic monitoring data of each time step, and then input into the LSTM layer together, a training set 70%, a verification set 15% and a test set 15% are divided, and an Adam optimizer is used for optimization by using a mean square error as a loss function, and finally a stabilit