Dynamic alpha gray scale model optimization method for pipe sinking project settlement prediction

Abstract

The invention relates to a dynamic alpha gray scale model optimization method for immersed tube project settlement prediction. The method comprises the steps of S1, immersed tube tunnel multi-source data preprocessing; s2, layered determination of a dynamic development coefficient alpha; and S3, constructing a dynamic alpha gray scale model and predicting settlement. According to the method, dynamic optimization of the core parameter alpha of the gray model is focused, the multi-source data-dynamic alpha-gray prediction integrated mathematical method adaptive to the immersed tube engineering characteristics is constructed, the limitation of a traditional fixed alpha gray model is broken through, and high-precision prediction of immersed tube engineering settlement is achieved.

Description

Technical Field

[0001] This invention belongs to the field of mathematical modeling and settlement prediction technology for immersed tunnel engineering, specifically involving a dynamic α-grayscale model optimization method for settlement prediction in immersed tunnel engineering. Background Technology

[0002] As the core structural form of large-scale cross-river and sea transportation projects, the settlement control of immersed tunnels directly determines the accuracy of tunnel segment connection, the waterproofing effect of GINA waterstops, and long-term operational safety, making it a core management objective in engineering construction. The settlement phenomenon in immersed tunnel projects stems from the coupled effects of three dimensions: geology, construction, and time. Geologically, the natural heterogeneity of marine environments, such as ultra-soft soil and mixed strata, leads to spatial differences in the bearing capacity of the foundation. Construction-wise, the gradual application of dynamic loads, including tunnel segment placement, foundation treatment, and siltation accumulation, causes continuous changes in the soil stress field. Temporally, the soil consolidation effect and historical stress accumulation result in a long-term evolutionary trend in settlement, with significant differences in settlement rates at different construction stages (placement, consolidation, and post-construction).

[0003] Settlement prediction is a prerequisite for settlement control in immersed tunnel engineering, and its accuracy directly affects the success or failure of the project. If the predicted value is too conservative, it will lead to an excessive pre-lifting amount, increasing construction costs and difficulties. If the predicted value is too small, it will cause excessive differential settlement between pipe sections (usually required to be ≤5mm), leading to risks such as waterstop leakage and uneven structural stress. In severe cases, it may require shutdown for maintenance or even rework, resulting in huge economic losses and project delays. Therefore, developing a high-precision mathematical method for settlement prediction that is suitable for the characteristics of immersed tunnel engineering is of decisive significance for ensuring project quality and safety.

[0004] However, the core challenge in predicting settlement in immersed tunnel projects lies in their dynamic evolution characteristics and complex coupling effects. Existing grayscale models have significant shortcomings in the application of immersed tunnel projects and are unable to meet the accuracy requirements. (1) The existing grayscale model parameter α is fixed and its adaptability is insufficient. In the traditional GM(1,1) grayscale model, the development coefficient α (the core parameter reflecting the settlement rate) is usually fixed by fitting once. However, the settlement process of immersed tunnel project has significant stages: in the early stage of construction, the settlement rate is fast due to the influence of dynamic load (the absolute value of α is large); in the later stage of construction, the settlement rate slows down (the absolute value of α decreases); in the post-construction stage, the settlement tends to be stable (the absolute value of α approaches 0). Fixing α cannot characterize the dynamic rate change of the immersed tunnel settlement, resulting in a large deviation between the predicted results and the actual settlement trend.

[0005] (2) The physical meaning of α is disconnected from the characteristics of immersed tunnel engineering. Existing technology does not clearly define the relationship between α and the key influencing factors of immersed tunnel engineering. α is only used as a mathematical fitting parameter and is not adapted to the geological conditions of immersed tunnel engineering (such as the thickness of ultra-soft soil and compression modulus), construction parameters (such as the self-weight of the tunnel section and the pushing speed), and environmental factors (such as the backfill thickness and seawater load). As a result, the value of α lacks engineering basis and the model has poor generalization ability.

[0006] (3) Low data utilization and lack of systematic optimization of α. The immersed tunnel project has accumulated a large amount of settlement measurement data, geological survey data and construction record data, but the existing grayscale model has not established a correlation mechanism between multi-source data and dynamic α. It only relies on a small amount of settlement time series data for fitting and does not make full use of the data specific to the immersed tunnel project to optimize parameters, resulting in a lack of support for its dynamic adjustment and limited prediction accuracy.

[0007] (4) The model is not adapted to the special scenarios of immersed tunnel projects. Immersed tunnel projects face special situations such as fluctuations in marine environmental siltation loads, complex geological stratification, and clear construction stages. The existing grayscale model has not designed adjustment rules for α for these scenarios. For example, when the siltation thickness increases, α is not adjusted accordingly to reflect the change in settlement rate, which further aggravates the prediction bias. Summary of the Invention

[0008] The purpose of this invention is to address the shortcomings and deficiencies of existing technologies by providing a dynamic α grayscale model optimization method for predicting settlement in immersed tunnel projects. This method focuses on the dynamic optimization of the core parameter α of the grayscale model, and constructs an integrated mathematical method of multi-source data-dynamic α-grayscale prediction that is adapted to the characteristics of immersed tunnel projects. This method breaks through the limitations of the traditional fixed α grayscale model and achieves high-precision prediction of settlement in immersed tunnel projects.

[0009] The technical problem solved by this invention is achieved through the following technical solution: A dynamic α-grayscale model optimization method for settlement prediction in immersed tunnel engineering, the method comprising the following steps: S1. Preprocessing of multi-source data for immersed tunnel: Focusing on the characteristics of immersed tunnel projects, three types of core data are collected, including settlement time series data, influencing factor data and stage division data; outliers in settlement time series data are removed and accumulated to generate sequences, and influencing factor data are normalized. S2. Determination of the dynamic development coefficient α in a stratified manner: Based on the stage characteristics and influencing factors of the immersed tunnel project, it is dynamically determined. α This includes initial stage determination, factor correction, and data fitting calibration, through the establishment of... α The corrected model has completed calibration; S3. Construction of Dynamic α-Gray Model and Settlement Prediction: Based on Dynamic Determinism αA grayscale model of the immersed tunnel project was constructed, and the predicted settlement value of the immersed tunnel project was obtained by cumulative subtraction restoration of the model over time. The model was updated in real time for different construction stages. α Values ​​enable dynamic prediction throughout the entire lifecycle.

[0010] Moreover, S1 specifically refers to: (1) Data collection: Focusing on the characteristics of immersed tunnel projects, three types of core data were collected: 1) Settlement time series data: Settlement data of the constructed pipe sections were collected using an automated settlement meter. Data was collected every 3 days during the construction period and monthly during the later stages, forming a non-negative discrete settlement sequence. ; In the formula, For the first n The settlement measured in each monitoring session is in mm. 2) Influencing Factor Data: Collect parameters of influencing factors specific to the immersed tunnel project, including geological parameters: cohesion of each stratum. c internal friction angle Compression modulus Es, formation thickness h Construction parameters: Pipe section self-weight q Temporary loads q_temp Top thrust F Launch speed v Environmental parameters: Siltation thickness h_silt siltation rate v_silt seawater density γ ; 3) Stage division data: Clearly define the immersed tunnel construction stages, including the foundation treatment period, the tunnel section placement period, the consolidation and stabilization period, and the post-construction operation period, as well as the time nodes for each stage; (2) Data preprocessing: 1) Outlier removal: Outliers in the settlement time series data are removed using the 3σ criterion; 2) Cumulative Generation: The preprocessed sedimentation sequence is cumulatively generated once. ; In the formula, ; 3) Normalization of influencing factor data: Geological, construction, and environmental parameters are normalized using the following formula: ; in: x ’ For the first i Normalized values ​​of influencing factors x i For the first i The original values ​​of the influencing factors. xmax and x min These are the maximum and minimum values ​​for this type of factor, respectively.

[0011] Moreover, S2 specifically refers to: (1) Preliminary determination of stages: Based on the stages of immersed tunnel construction, the stages are preliminarily determined. α The benchmark range, combined with industry engineering practice and settlement mechanism analysis, clarifies α Phase adaptation rules: 1) Foundation treatment period: Significant soil disturbance and rapid settlement. α The benchmark range is ; 2) Pipe section laying period: Dynamic loads are applied, and the settlement rate is relatively fast. α The benchmark range is ; 3) Consolidation stabilization period: The load stabilizes, and the settlement rate slows down. α The benchmark range is ; 4) Post-construction operation period: Settlement tends to stabilize. α The benchmark range is ; (2) Factor correction: Based on the normalized influencing factor data, establish α The modified model quantifies the influence weight of each factor, and the specific formula is as follows: ; In the formula, For each construction stage of the immersed tunnel project, based on the physical mechanism of settlement and industry engineering practice, the initial reference value of the gray-scale model development coefficient α is pre-defined, which is the initial anchor point for dynamically determining α. For the first i The weights of the influencing factors were obtained using the analytic hierarchy process (AHP). For the first i Normalized values ​​of influencing factors; (3) Data fitting and calibration: Based on the existing settlement time series data of the immersed tunnel project, the least squares method was used to calibrate the corrected data. α Perform final calibration to ensure α The formula closely matches the actual settlement trend: ; In the formula, For based on α The calculated settlement prediction value is obtained by cumulative subtraction of the grayscale model time response sequence.

[0012] Moreover, S3 specifically refers to: (1) Model construction: based on dynamic determination αA dedicated GM(1,1) grayscale model for immersed tunnel engineering was constructed, and the core parameter grayscale sequence was solved as follows: ; In the formula, B and Y These are the data matrix and the constant term matrix, respectively. (2) Time response sequence: The time response sequence of the model is: ; In the formula, For the first k+1 The cumulative predicted value; (3) Settlement prediction: The settlement prediction value for the immersed tunnel project is obtained by cumulative subtraction and restoration of the time response sequence; and updated in real time for different construction stages. α Values ​​enable dynamic prediction throughout the entire lifecycle.

[0013] The advantages and beneficial effects of this invention are as follows: 1. This invention proposes for the first time a method for determining the dynamic development coefficient α specific to immersed tunnel engineering, breaking through the limitations of the fixed α in the traditional grayscale model, accurately depicting the dynamic rate change of immersed tunnel settlement, and providing a clear basis for protecting the core parameter α; 2. Highly adaptable: It deeply integrates the geological characteristics, dynamic construction, marine environment and other exclusive scenarios of immersed tunnel projects. The dynamic adjustment rules of α are highly consistent with the settlement mechanism of immersed tunnels, which solves the pain point of poor adaptability of general grayscale models. 3. High prediction accuracy: Through three-level optimization of initial stage determination, factor correction and data calibration, the average relative error of the gray model prediction is ≤5%, which is far better than the traditional fixed α model and meets the stringent settlement control requirements of immersed tunnel engineering. 4. High engineering practicality: The mathematical methods and steps are clear and the formulas are well-defined. It can be directly integrated into immersed tunnel engineering prediction software without relying on other models. It can independently achieve settlement prediction, providing accurate data support for pre-lifting setting and construction control, and has extremely high promotion value. Attached Figure Description

[0014] Figure 1 This is a flowchart of the present invention.

[0015] Figure 2 This is a comparison chart of the predicted and measured values ​​of the settlement of the immersed tunnel section during the consolidation stabilization period according to the present invention. Figure 3 This is a comparison chart of predicted and measured values ​​at different observation points of the immersed tunnel segment during the immersion period of this invention. Detailed Implementation

[0016] The present invention will be further described in detail below through specific embodiments. The following embodiments are merely descriptive and not limiting, and should not be used to limit the scope of protection of the present invention.

[0017] Taking the settlement prediction of the E16 standard pipe section (165m straight type) of Contract Section S09 of the Shenzhen-Zhongshan Bridge as an example, the implementation process of the method of the present invention is explained in detail: Step 1: Multi-source data collection and preprocessing (1) Data collection: 1) Settlement data collection: Geological parameters (overly soft soil c=8kPa, =3°, Es=2.5MPa, thickness 4.2m), construction parameters (pipe section self-weight q=25kPa, jacking force F=1100kN, pushing speed v=0.6mm / s), environmental parameters (siltation thickness h_silt=0.25m, siltation rate v_silt=0.04m / month); 2) Data preprocessing: 1. Remove outliers: Data without outliers is retained directly; 2. Accumulation generation ; 3. Normalization: Normalized values ​​of geological parameters Construction parameters Environmental parameters .

[0018] Step 2: Determining Dynamic α (1) Initial stage α Pipe section sinking period α 基准 =-0.12, consolidation stabilization period α 基准 =-0.06; (2) Factor correction α : , , Corrected calculation: 1) Settling period: α 修正1 = -0.12 × (1 + 0.4×0.75 + 0.35×0.68 + 0.25×0.52) = -0.20016; 2) Consolidation stabilization period: α 修正2 = -0.06 ×(1 + 0.4×0.75 + 0.35×0.68 + 0.25×0.52)= -0.10008; (3) Data fitting calibration α The final determination of the settling period was made through least squares calibration. Consolidation stabilization period In this embodiment, α in each of the aforementioned construction stages 基准 The range is only used for the initial α 基准 The value is anchored and does not affect the final dynamic obtained after subsequent factor correction and fitting calibration. The value constitutes a mandatory constraint.

[0019] Step 3: Model Building and Prediction (1) Solution of grayscale parameters This embodiment B and Y The values ​​of the matrix strictly follow the standard modeling rules of the GM(1,1) model in grey system theory, matching the technical solutions of S1 and S3. All values ​​are derived from the measured data in this embodiment. The specific value selection process is as follows: The core grey differential equation of the GM(1,1) model is: ; in: u The gray action quantity of the GM(1,1) model reflects the background value of the settlement sequence; is the original measured value of the immersed tube settlement during the kth monitoring, a non-negative discrete sequence; The nearest neighbor mean of the generated sequence is calculated using the formula: [Formula omitted for brevity]. ; in: is the cumulative value generated by the first accumulation of the kth term in the original settlement sequence, and is the cumulative sum of the measured settlement values ​​from the previous k periods, calculated according to the S1 rule. k The range of values ​​is k =2,3,...., n ; Will k =2,3,...., n The simultaneous solution of the grey differential equations is transformed into the standard matrix form for least squares solution as follows: ; Where: data matrix B constant term matrix Y The construction formula is a standard practice in the grey system theory industry, specifically: ; ; Following the data collection rules in step S1, during the construction period, an automated settlement meter was used to collect settlement data for three consecutive periods during the E16 pipe section's immersion phase, once every three days. After verification using the 3σ criterion, the settlement data in this embodiment showed no anomalies and was all directly retained, resulting in a compliant non-negative discrete original settlement sequence. ; According to the calculation formula in step S1: ; Substitute the data from this embodiment: ; According to the legal rules of the GM(1,1) model, the formula for calculating the nearest neighbor mean is: ; Input the data in this embodiment ; Therefore, we can conclude that: 1) Settling period: , Solving for the given information, we can obtain the following results: , ; 2) Consolidation and Stabilization Period: Renewal B Matrix and Y Vectors, solved to obtain , ; (2) Time response sequence and prediction: 1) Settling period: Restore the predicted value .

[0020] 2) Consolidation stabilization period: Restore the predicted value .

[0021] (3) Full-cycle prediction results: Model prediction accuracy evaluation and parameter iterative optimization (1) Accuracy evaluation: using the average relative error δ Relevance r As an evaluation metric, the accuracy of the model's predictions is verified. ; ; (2) Iterative optimization: As the immersed tunnel construction progresses, new settlement measurement data are continuously collected, and steps 2-3 are repeated to dynamically update the data. α The values ​​and model parameters ensure that the prediction accuracy continues to improve as data accumulates.

[0022] Figure 2 This graph compares the predicted and measured settlement values ​​of the immersed tunnel sections during the consolidation and stabilization period. The horizontal axis uses a monthly time unit, matching the settlement monitoring frequency of once a month during the later stages of construction as specified in S1. Figure 3 This is a comparison chart of predicted and measured settlement values ​​at different observation points during the tunnel section installation period. The horizontal axis uses days as the time unit, matching the settlement monitoring frequency specified in S1 during the construction period. The chart shows that the prediction results of the grey theory model are generally consistent with the observed data in terms of trend, with a relatively small error. δ= 0.35%, correlation r= 0.96, which meets the engineering accuracy requirements.

[0023] Although embodiments and drawings of the present invention have been disclosed for illustrative purposes, those skilled in the art will understand that various substitutions, variations and modifications are possible without departing from the spirit and scope of the present invention and the appended claims. Therefore, the scope of the present invention is not limited to the contents disclosed in the embodiments and drawings.

Claims

1. A dynamic α-grayscale model optimization method for settlement prediction in immersed tunnel engineering, characterized in that: The steps of the method are as follows: S1. Preprocessing of multi-source data for immersed tunnel: Focusing on the characteristics of immersed tunnel projects, three types of core data are collected, including settlement time series data, influencing factor data and stage division data; outliers in settlement time series data are removed and accumulated to generate sequences, and influencing factor data are normalized. S2. Determination of the dynamic development coefficient α in a stratified manner: Based on the stage characteristics and influencing factors of the immersed tunnel project, it is dynamically determined. α This includes initial stage determination, factor correction, and data fitting calibration, through the establishment of... α The corrected model has completed calibration; S3. Construction of Dynamic α-Gray Model and Settlement Prediction: Based on Dynamic Determinism α A grayscale model of the immersed tunnel project was constructed, and the predicted settlement value of the immersed tunnel project was obtained by cumulative subtraction restoration of the model over time. The model was updated in real time for different construction stages. α Values ​​enable dynamic prediction throughout the entire lifecycle.

2. The dynamic α-grayscale model optimization method for settlement prediction of immersed tunnel projects according to claim 1, characterized in that: Specifically, S1 is: (1) Data collection: Focusing on the characteristics of immersed tunnel projects, three types of core data were collected: 1) Settlement time series data: Settlement data of the constructed pipe sections were collected using an automated settlement meter. Data was collected every 3 days during the construction period and monthly during the later stages, forming a non-negative discrete settlement sequence. ； In the formula, For the first n The settlement measured in each monitoring session is in mm. 2) Influencing Factor Data: Collect parameters of influencing factors specific to the immersed tunnel project, including geological parameters: cohesion of each stratum. c internal friction angle Compression modulus Es, formation thickness h Construction parameters: Pipe section self-weight q Temporary loads q_temp Top thrust F Launch speed v Environmental parameters: Siltation thickness h_silt siltation rate v_silt seawater density γ ; 3) Stage division data: Clearly define the immersed tunnel construction stages, including the foundation treatment period, the tunnel section placement period, the consolidation and stabilization period, and the post-construction operation period, as well as the time nodes for each stage; (2) Data preprocessing: 1) Outlier removal: Outliers in the settlement time series data are removed using the 3σ criterion; 2) Cumulative Generation: The preprocessed sedimentation sequence is cumulatively generated once. ； In the formula, ; 3) Normalization of influencing factor data: Geological, construction, and environmental parameters are normalized using the following formula: ； in: x ’ For the first i Normalized values ​​of influencing factors x i For the first i The original values ​​of the influencing factors. x max and x min These are the maximum and minimum values ​​for this type of factor, respectively.

3. The dynamic α-grayscale model optimization method for settlement prediction of immersed tunnel projects according to claim 1, characterized in that: Specifically, S2 is: (1) Preliminary determination of stages: Based on the stages of immersed tunnel construction, the stages are preliminarily determined. α The benchmark range, combined with industry engineering practice and settlement mechanism analysis, clarifies α Phase adaptation rules: 1) Foundation treatment period: Significant soil disturbance and rapid settlement. α The benchmark range is ; 2) Pipe section laying period: Dynamic loads are applied, and the settlement rate is relatively fast. α The benchmark range is ; 3) Consolidation stabilization period: The load stabilizes, and the settlement rate slows down. α The benchmark range is ; 4) Post-construction operation period: Settlement tends to stabilize. α The benchmark range is ; (2) Factor correction: Based on the normalized influencing factor data, establish α The modified model quantifies the influence weight of each factor, and the specific formula is as follows: ； In the formula, For each construction stage of the immersed tunnel project, based on the physical mechanism of settlement and industry engineering practice, the initial reference value of the gray-scale model development coefficient α is pre-defined, which is the initial anchor point for dynamically determining α. For the first i The weights of the influencing factors were obtained using the analytic hierarchy process (AHP). For the first i Normalized values ​​of influencing factors; (3) Data fitting and calibration: Based on the existing settlement time series data of the immersed tunnel project, the least squares method was used to calibrate the corrected data. α Perform final calibration to ensure α The formula closely matches the actual settlement trend: ； In the formula, For based on α The calculated settlement prediction value is obtained by cumulative subtraction of the grayscale model time response sequence.

4. The dynamic α-grayscale model optimization method for settlement prediction of immersed tunnel projects according to claim 1, characterized in that: Specifically, S3 is: (1) Model construction: based on dynamic determination α A dedicated GM(1,1) grayscale model for immersed tunnel engineering was constructed, and the core parameter grayscale sequence was solved as follows: ； In the formula, B and Y These are the data matrix and the constant term matrix, respectively. (2) Time response sequence: The time response sequence of the model is: ； In the formula, For the first k+1 The cumulative predicted value; (3) Settlement prediction: The settlement prediction value of the immersed tunnel project is obtained by cumulatively subtracting and restoring the time response sequence. Updated in real time for different construction stages α Values ​​enable dynamic prediction throughout the entire lifecycle.

Images