Bridge swivel construction state risk grading early warning method and system

Abstract

This invention discloses a method and system for risk classification and early warning of bridge rotation construction status. The method includes: determining the structural state parameters of the bridge rotation and acquiring monitoring data during the bridge rotation process; deriving the posterior distribution of the structural state parameters by combining Bayesian inference and Laplace prior information, and introducing mixed variables to derive the posterior mixed distribution of the structural state parameters; constructing the logarithmic posterior distribution of the mixed variables and deriving its gradient and Hessian matrix; using the Laplace approximation method to derive the low-dimensional approximate posterior of the mixed variables and derive the approximate posterior distribution of the structural state parameters; sampling approximate posterior samples, calculating the probability of exceeding limits for the structural state parameters, and performing graded early warning and handling. This invention provides early warning of bridge rotation construction status risks based on the probability of exceeding limits for structural state parameters, realizing a shift from traditional monitoring index threshold alarms to refined early warning driven by mechanism parameters. It boasts high computational efficiency, high-efficiency early warning, and high accuracy.

Description

Technical Field

[0001] This invention relates to the field of safety control technology for bridge rotation construction, and in particular to a method and system for risk classification and early warning of bridge rotation construction status. Background Technology

[0002] With the rapid development of transportation infrastructure and the increasingly sophisticated transportation network in my country, the number of newly built bridges crossing existing railway lines is constantly increasing. The bridge rotation method involves constructing the bridge beams at a position nearly parallel to the railway line, then rotating them to the designed alignment of the completed bridge. After the closure section is constructed, the system transformation is complete. This method offers advantages such as minimal interference with existing railway lines, high safety, and good economic efficiency. For large-span bridges, the weight of the bridge beams during rotation often reaches tens of thousands of tons. The rotation time is short, and the rotation system is complex. The safety and stability requirements during the rotation process are extremely high. Monitoring and control of the rotation construction can promptly grasp the stress state, deformation, and stability of the bridge structure, allowing for early detection and correction of construction deviations.

[0003] Currently, early warning systems for bridge rotation construction status largely rely on empirical thresholds. Information about the rotation process is obtained through monitoring indicators such as traction force, torque, angular velocity, and support reaction force, combined with existing standards to assess the safety, stability, and accuracy of the rotation construction. However, existing methods do not consider structural state parameters such as friction coefficient, support stiffness, equivalent stiffness, and eccentricity parameters. This leads to risks that cannot be explicitly characterized, such as sudden increases in rotation resistance, imbalances in reaction force distribution, trajectory deviation, abnormal structural response, and even difficulties in positioning. Furthermore, they fail to consider uncertainties such as monitoring noise, model errors, and parameter variations, making it difficult to achieve a probabilistic representation and graded early warning of construction status risks driven by the inherent mechanisms of bridge rotation. This results in untimely risk warnings and mitigation. Therefore, there is an urgent need to propose a graded early warning method and system for bridge rotation construction status risks based on monitoring data to improve the timeliness of risk warnings and mitigation. Summary of the Invention

[0004] To overcome the shortcomings of the existing technology, the present invention provides a method for inverting the construction state parameters of bridge rotation and for risk classification and early warning, so as to improve the timeliness of risk warning and risk management.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: Firstly, a risk classification and early warning method for bridge rotation construction status includes the following steps: Step 1: Determine the structural state parameters of the bridge rotation and set up a sensor monitoring network to acquire monitoring data during the rotation construction process; Step 2: Assuming the structural state parameters follow a Laplace prior, monitor the data and structural state parameters. The linear Gaussian observation relationship is satisfied, and the posterior distribution of the structural state parameters is derived by combining Bayesian inference. Mixed variables are introduced and it is assumed that the mixed variables follow a product-type exponential distribution, so that the Laplace prior distribution of the structural state parameters can be expressed as a Gaussian-scale mixture form. Then the posterior mixture distribution of the structural state parameters is derived. Step 3: Based on the prior and posterior distributions of the mixed variables, construct their log-posterior distribution, and derive the corresponding gradient and Hessian matrix accordingly. Step 4: Using the Laplace approximation method, maximize the log-posterior distribution of the mixed variables to solve for the maximum a posteriori point of the mixed variables. Based on the r-dimensional positive components of the maximum a posteriori point, the mixed variables are divided into r-dimensional components and the remaining components. Based on the r-dimensional components and the Hessian matrix at the maximum a posteriori point, construct the r-dimensional approximate marginal a posteriori mixture distribution of the mixed variables. Then, combined with the prior distribution of the remaining components, derive the dimension-reduced approximate a posteriori mixture distribution of the overall mixed variables. Replace the corresponding posterior distribution with this distribution to obtain the approximate posterior distribution of the structural state parameters. Sample the r-dimensional components and the remaining components separately to obtain the approximate posterior samples of the structural state parameters. Step 5: Based on the principle of probabilistic risk control, establish a graded early warning discrimination criterion based on the probability of exceeding limits; calculate the corresponding probability of exceeding limits based on the approximate posterior sample of the structural state parameters at the current moment and the engineering threshold, and determine the early warning status of the structural state parameters at the current moment according to the graded early warning discrimination criterion.

[0006] It should be noted that, in step one, the structural state parameters include at least one of the following: friction characteristic parameters, stiffness parameters, rotational parameters, contact state parameters, geometric eccentricity parameters, and environmental coupling parameters.

[0007] It should be noted that in step two, the observation noise in the linear Gaussian observation relationship follows a zero-mean Gaussian distribution.

[0008] It should be noted that, in step two, the posterior mixture distribution of the structural state parameters is derived as follows: The corresponding posterior distribution is derived from the prior distribution and likelihood function of the structural state parameters under mixed variable conditions. The corresponding posterior distribution is derived from the prior distribution and likelihood function of the mixed variables; Establish the joint posterior distribution of structural state parameters and mixed variables based on the two posterior distributions; The posterior mixed distribution is obtained by integrating the mixture variables in the posterior joint distribution.

[0009] It should be noted that in step three, the first derivative of the log-posterior distribution of the mixed variables is used as the gradient of the mixed variables, and the second derivative of the log-posterior distribution of the mixed variables is used as the Hessian matrix of the mixed variables.

[0010] It should be noted that in step four, the r-dimensional approximate marginal posterior mixture distribution of the mixture variable is multiplied by the prior distribution of the remaining components of the mixture variable to obtain the dimension-reduced approximate posterior mixture distribution of the mixture variable as a whole.

[0011] It should be noted that in step five, based on the principle of probabilistic risk control, three levels of early warning thresholds are set, and four early warning states are established accordingly as the criteria for determining early warning based on the probability of exceeding limits: Normal state: ; Level 1 Warning: ; Level 2 warning: ; Alarm status: ; In the formula: α1 represents the probability of structural state parameters exceeding the engineering threshold range; α2, α3 represent the graded warning thresholds. α1, α2, and α3 can be referenced to the two-tailed probabilities corresponding to ±3σ, ±2σ, and ±1σ under the standard normal distribution, i.e., α1≈0.003, α2≈0.05, and α3≈0.32, where σ is the standard deviation of the standard normal distribution.

[0012] It should be noted that in step five, the r-dimensional component and the remaining components in the mixed variable are sampled separately to obtain approximate posterior samples of the structural state parameters. Based on the approximate posterior samples and the engineering threshold, the probability of exceeding the limit of the structural state parameters is calculated using the following formula: ; In the formula: This is the engineering threshold; The probability of structural state parameters exceeding the engineering threshold range; This is an indicator function that takes the value 1 when the condition inside the parentheses is true, and 0 otherwise. Let N be the nth approximate posterior sample, and N be the number of samples.

[0013] It should be noted that in step five, based on the warning status of the structural state parameters, the following measures can be taken: Under normal conditions, continue the rotation construction; Level 1 warning: Conduct reliability verification of monitoring data from warning monitoring points, confirm whether the monitoring data is abnormal and record abnormal monitoring points; continue the rotation construction; A Level II warning requires a special investigation of the structural status parameters that are under warning, and the rotation construction will continue after adjusting the corresponding rotation construction parameters. Alarm status detected. Stop rotation construction and conduct a comprehensive risk assessment.

[0014] Secondly, a bridge rotation construction status risk classification and early warning system includes: The data acquisition module is used to collect monitoring data; The data processing module is used to calculate the probability of exceeding the limits of structural state parameters and determine the corresponding early warning status; The interactive module is used to select structural status parameters and display the warning status of structural status parameters; The data storage module is used to store monitoring data, probability of exceeding limits, and early warning status. This is to achieve the aforementioned method for risk classification and early warning of bridge rotation construction status.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) This invention establishes a Bayesian inversion framework for bridge rotation structure state parameters based on Laplace prior. It achieves posterior estimation of structural state parameters through collaborative updating of multi-source monitoring data and prior information. It suppresses non-critical parameter disturbances and strengthens the identification of critical parameters through the sparsity and heavy-tailedness of Laplace prior, thereby improving the physical interpretability of parameter inversion results. (2) This invention proposes a fast posterior inference method based on maximum a posteriori estimation and Laplace approximation. It can quickly locate the high probability region of the posterior distribution of mixed variables by using the maximum a posteriori point. At the maximum a posteriori point, the Laplace approximation is constructed using Hessian curvature information. Combined with the selected activation components, it can achieve effective dimensionality reduction calculation, significantly reduce computational overhead, improve computational efficiency, achieve efficient early warning, and have high early warning accuracy. (3) This invention proposes a bridge rotation construction state risk early warning method with the structural state parameter exceeding the limit probability as the core. By constructing the posterior probability distribution of the state parameter and calculating the exceeding probability of it exceeding the engineering threshold, and combining the graded threshold, the normal state, first-level early warning, second-level early warning and alarm state are dynamically determined. Based on this, the rotation construction parameters are controlled and dealt with, realizing the transformation from monitoring index alarm to mechanism parameter driven refined early warning. Attached Figure Description

[0016] Figure 1 This is a flowchart of the bridge rotation construction status risk classification and early warning method of the present invention.

[0017] Figure 2 This is a flowchart of the sampling process for the approximate posterior distribution based on maximum a posteriori estimation and Laplace approximation in this invention.

[0018] Figure 3This is a structural schematic diagram of the bridge rotation construction status risk classification and early warning system of the present invention; wherein, 100-data acquisition module, 200-data processing module, 300-interaction module, 400-data storage module. Detailed Implementation

[0019] The technical solutions of the present invention will now be clearly and completely described in conjunction with the accompanying drawings.

[0020] For Implementation Method 1, please refer to [link / reference]. Figure 1-2 ; like Figure 1 , Figure 1 This is a flowchart of the bridge rotation construction status risk classification and early warning method of the present invention.

[0021] like Figure 2 , Figure 2 This is a flowchart of the posterior mixture density sampling process based on maximum a posteriori estimation and Laplace approximation of the present invention.

[0022] This embodiment provides a method for risk classification and early warning of bridge rotation construction status, including the following steps: Step 1: Determine the structural state parameters of the bridge rotation. Deploy a sensor monitoring network to acquire monitoring data during the rotation construction process. .

[0023] The structural state parameters include at least one of the following: friction characteristic parameters, stiffness parameters, rotational parameters, contact state parameters, geometric eccentricity parameters, and environmental coupling parameters.

[0024] The monitoring indicators of the sensor monitoring network include external environment monitoring indicators, support system monitoring indicators, traction system monitoring indicators, rotation attitude monitoring indicators, and structural stress monitoring indicators. Specifically, these may include: external environment monitoring indicators such as wind speed, wind direction, temperature, and humidity, with wind speed being the main indicator affecting rotation stability; support system monitoring indicators such as the gap between the support feet and the slide, support foot reaction force, turntable stress, eccentricity, and frictional resistance; traction system monitoring indicators mainly consisting of traction force, traction length, and traction speed, supplemented by traction torque, friction torque, and inertial torque; rotation attitude monitoring indicators mainly consisting of rotation tilt angle, rotation angle, and rotation angular velocity, supplemented by changes in vibration acceleration at the pier top and beam end, and the relative elevation difference at the cantilever end; and structural stress monitoring indicators mainly consisting of stress at the cantilever root section, supplemented by stress at the pier bottom section.

[0025] The specific selection of structural state parameters and monitoring indicators can be adjusted according to the actual project type, bridge scale, construction conditions, and monitoring conditions. For small bridge rotation construction, key indicators can be selected for monitoring, while for large bridge rotation construction, the types of monitoring indicators and the number of measuring points can be increased. Multiple measuring points can be arranged for different monitoring indicators according to project requirements, and the monitoring values ​​collected at each measuring point for each indicator are arranged in a preset order to form a monitoring data vector. The structural state parameters are arranged in a preset order to form a structural state parameter vector. , and The expression for vector is omitted in both cases.

[0026] Step 2: Assume structural state parameters Obey the Laplace prior and monitor the data. With structural state parameters The structural state parameters are derived by satisfying the linear Gaussian observation relationship and combining it with Bayesian inference. posterior distribution Further introduce mixed variables And assume mixed variables It follows a product-type exponential distribution, making the structural state parameters The Laplace prior distribution can be expressed as a Gaussian-scale mixture, and then the structural state parameters can be derived. The posterior mixed distribution.

[0027] It should be noted that in step two, we assume the monitoring data... With structural state parameters To satisfy the linearized Gaussian observation relation, the following linear Gaussian observation model can be established as shown in equation (2.1): (2.1); In the formula: Let i be the component index of the structural state parameter, and d be the number of structural state parameters. For monitoring data; Here is the observation matrix; ε is the observation noise, which follows a zero-mean Gaussian distribution. To observe the noise covariance matrix; Therefore, the corresponding likelihood function It can be determined according to formula (2.2): (2.2); In the formula: For monitoring data The likelihood function; This is the inverse of the observation noise covariance matrix.

[0028] It should be noted that in step two, when acquiring monitoring data... Then, combining Bayesian inference and prior information, the structural state parameters can be derived according to equation (2.3). posterior distribution for: (2.3); In the formula: Structural state parameters The prior distribution; This is the normalization constant.

[0029] In the actual bridge rotation construction process, most structural state parameters change relatively little. To characterize the sparsity of structural state parameters, the structural state parameters are... Setting it as a product-type Laplace prior, it can be defined as follows according to equation (2.4): (2.4); In the formula: The rate parameter is the Laplace prior.

[0030] It should be noted that in step two, after combining the Laplace prior and the linear Gaussian observation model, the structural state parameters... The posterior distribution of a model typically no longer has a Gaussian closed-form, making it difficult to directly sample and obtain its posterior analytical representation. Therefore, based on the theory of Gaussian scaling mixture models, mixture variables are introduced. Each of its components is a positive number, and it is assumed that the mixture variable... It follows a product-type exponential distribution, making the structural state parameters In mixed variables Under certain conditions, the structure follows a zero-mean Gaussian distribution, therefore the structural state parameters can be... Laplace's a priori Equation (2.5) can be rewritten in Gaussian scale mixture form: (2.5); In the formula: The component density, also known as the structural state parameter. In mixed variables Conditional distribution under given conditions; It is a mixed variable; For mixed variables The prior distribution of .

[0031] In equation (2.5), the Laplace distribution has leptokurtic and heavy-tailed characteristics. It can be transformed into a conditional Gaussian distribution in a non-mixed form by introducing a scaling variable from the exponential distribution. Therefore, assuming a mixed variable... The prior distribution is a product-type exponential distribution, which can be determined according to equation (2.6): (2.6); In the formula: For mixed variables The rate parameter in the exponential prior distribution.

[0032] In equation (2.5), the component density It follows a zero-mean Gaussian distribution and can be determined according to equation (2.7): (2.7); In the formula: It is a diagonal matrix with each component of the mixed variable as its diagonal element.

[0033] It should be noted that in step two, due to the monitoring data... and mixed variables Given structural state parameters Under the condition that they are conditionally independent, the joint distribution of the three can be determined according to equation (2.8): (2.8); In the formula: Structural state parameters Mixed variables Monitoring data The joint distribution of .

[0034] Furthermore, based on the joint distribution in equation (2.8), the posterior joint distribution can be decomposed into the product of two conditional distributions, as shown in equation (2.9): (2.9); In the formula: Structural state parameters Mixed variables The posterior joint distribution; For the posterior component distribution, i.e., given a mixture of variables and monitoring data Structural state parameters under certain conditions The posterior distribution of; Given monitoring data Mixed variables under conditions The posterior distribution of; Based on Bayes' principle, the mixed variables in equation (2.9) are... By integrating, the structural state parameters can be obtained. The posterior mixture distribution is given by equation (2.10): (2.10); In the formula: Structural state parameters The posterior mixed distribution.

[0035] Posterior mixed distribution That is, structural state parameters The posterior distribution is simply a different way of expressing the same meaning.

[0036] In equation (2.10), under the condition that the linear Gaussian observation model as shown in equation (2.1) and the conditional Gaussian distribution as shown in equation (2.7) both hold, the posterior component density... The Gaussian form can be determined according to equation (2.11): (2.11); In the formula: for The inverse matrix, For mixed variables The posterior covariance matrix; This is the posterior mean; Wherein, the inverse matrix of the posterior covariance matrix and posterior mean They satisfy equations (2.12) and (2.13) respectively: (2.12); (2.13); In the formula: For mixed variables The covariance matrix of each component, for The inverse matrix.

[0037] In equation (2.10), the posterior distribution It can be expressed as follows (2.14): (2.14); (2.15); In the formula: For mixed variables The likelihood function; For mixed variables The vector corresponding to the mean of each component.

[0038] It should be noted that, according to equations (2.11)-(2.13) and (2.14)-(2.15), the structural state parameters posterior mixture distribution It can be rewritten as equation (2.16): (2.16).

[0039] The risks associated with bridge rotation construction are typically driven by the anomalous evolution of a few key parameters, with posterior changes in structural state parameters exhibiting sparse distribution or localized anomalies. This application employs a Laplace prior to describe structural state parameters, which can suppress disturbances in non-key parameters and highlight the responses of key parameters through sparsity contraction. This effectively suppresses coupled compensation of structural state parameters and improves the stability and interpretability of key state parameter identification.

[0040] However, when the Laplace prior is combined with the linear Gaussian observation model, the posterior probability density of the structural state parameters usually no longer has a Gaussian closed form, making direct sampling and posterior analytical expression difficult. This application, based on Gaussian scale mixture model theory, introduces mixture variables to rewrite the Laplace prior into a hierarchical Gaussian scale mixture form based on a conditional Gaussian distribution and a product-type exponential prior distribution. This transforms the non-Gaussian posterior problem into a conditional Gaussian inference framework problem, and by sampling the mixture variables, a posterior analytical expression of the structural state parameters can be achieved.

[0041] Step 3: Based on the mixed variables obtained in Step 2 prior distribution and posterior distribution Constructing mixed variables log-posterior distribution Based on this, we can derive its relationship with mixed variables. The first and second derivatives are used as the corresponding gradients and Hessian matrices.

[0042] It should be noted that the product-type Laplace prior mixed variables constructed based on equations (2.4) and (2.14) The log-posterior distribution is determined according to equation (3.1): (3.1); In the formula: for The identity matrix; For mixed variables Each component is a diagonal matrix composed of diagonal elements; A T Let A be the transpose of A; For monitoring data Column vectors; for The inverse matrix; Let be a normalization constant, so that the integral of the corresponding posterior distribution is 1.

[0043] It should be noted that the log-posterior distribution For mixed variables Taking the first derivative, we obtain the gradient as shown in equation (3.2): (3.2); (3.3); In the formula: Log-posterior distribution The gradient; To extract the main diagonal of the matrix and form column vectors; This is an element-wise squaring operation; It should be noted that... For mixed variables Taking the second derivative, we obtain the Hessian matrix as shown in equation (3.4): (3.4); In the formula, Let z be a diagonal matrix with each component of the variable z as its diagonal element.

[0044] Step 4: Employ the Laplace approximation method by maximizing the log-posterior density from Step 3. To solve for mixed variables Maximum posterior point Based on the maximum a posteriori point The r-dimensional positive components are used to divide the mixture variable into r-dimensional components and other components. Furthermore, based on the r-dimensional components and the Hessian matrix at the maximum posterior point, the r-dimensional approximate marginal posterior mixture distribution of the mixture variable is derived. ;according to Based on the prior distributions of the remaining components, derive the dimension-reduced approximate posterior mixture distribution of the overall mixture variables. ,use Replace the original Obtain structural state parameters Approximate posterior distribution Sample the r-dimensional component and the remaining components separately to obtain approximate posterior samples of the structural state parameters.

[0045] It should be noted that in the classical Laplace approximation method, the mean of the approximate distribution is given by the maximum a posteriori estimate, and its covariance matrix is ​​given by the negative log-posterior distribution at that maximum a posteriori point. The inverse of the Hessian matrix is ​​given. Based on the above theory, the maximum a posteriori estimate is constructed by solving the following maximum a posteriori optimization problem: (4.1); In the formula: For mixed variables The maximum a posteriori point; arg min is the value of the independent variable that minimizes the objective function.

[0046] Among them, the mixed variables The domain of definition from Expand to Component partitioning is performed based on the following set of indices: (4.2); In the formula: For mixed variables Midpoint and maximum a posteriori The component corresponding to the neutral component; For the selected The index set of each component (r is the number of positive components).

[0047] The selected component is denoted as The unselected components are denoted as ,in .

[0048] Through permutation matrix The following component decomposition methods can be defined: (4.3).

[0049] It should be noted that, after obtaining the maximum posterior point... Then, take the negative logarithmic posterior distribution Hessian matrix in the index set. The inverse of the corresponding submatrix, taken as the covariance matrix, can be represented as: (4.4); In the formula: This is the approximate posterior covariance matrix at the maximum posterior point; For matrix The inverse matrix; subscript This indicates retrieving the corresponding index set from the Hessian matrix. The rows and columns.

[0050] Based on the selected component Define the r-dimensional approximate marginal posterior mixture distribution according to equation (4.5). : (4.5); In the formula: It is an r-dimensional approximate marginal posterior mixture distribution; This is an indicator function.

[0051] For the remaining components not selected Preserve its prior distribution Unchanged, as in equation (4.6): (4.6); Therefore, the overall dimensionality reduction approximates the posterior distribution. It can be written as equation (4.7): (4.7); In the formula: It is a dimension-reduced approximate posterior distribution.

[0052] For the selected component According to its approximate marginal posterior distribution Sampling was conducted on the unselected components. According to its prior distribution Sampling is performed as shown in equation (4.8): (4.8); In the formula: n is the sample ordinal number, and N is the sample size.

[0053] Then, by inverse transformation of the permutation matrix, the aforementioned sampled samples are rearranged back to the original component order, resulting in the complete approximate posterior sample of the mixed variables as shown in equation (4.9): (4.9); It should be noted that, using Replace the original Then the approximate posterior distribution of the structural state parameter x can be written as equation (4.10): (4.10); Substituting the mixed variable posterior sample obtained from equation (4.9) into the conditional posterior sample distribution Sampling is performed to obtain posterior samples of structural state parameters. .

[0054] In Equation (2.15), the posterior mixture distribution simultaneously includes determinant terms, quadratic exponential terms, and implicit coupling dependencies of the mean and covariance on the mixture variables. This results in a non-standard, high-dimensional, and complex distribution, making direct sampling computationally expensive and inefficient. Step four of this application uses maximum a posteriori estimation to quickly locate the high-probability region of the posterior distribution and combines it with the local curvature information in the Hessian matrix located at the maximum a posteriori point to construct an approximate posterior distribution. This significantly reduces computational complexity while preserving the probability quantification capability of Bayesian inference.

[0055] Step 5: Based on the principle of probabilistic risk control, establish a graded early warning judgment criterion based on the probability of exceeding limits; based on the structural state parameters at the current moment... Approximate posterior samples and engineering threshold Calculate the probability of exceeding limits and determine the structural state parameters at the current moment according to the graded early warning discrimination criteria. The alert status is in effect.

[0056] Based on the principles of probabilistic risk control, three levels of early warning thresholds are set, and four early warning states are defined: Normal state: (5.1); Level 1 Warning: (5.2); Level 2 warning: (5.3); Alarm status: (5.4); In the formula: α1, α2, and α3 are the graded warning thresholds. α1, α2, and α3 can be referenced to the two-tailed probabilities corresponding to ±3σ, ±2σ, and ±1σ under the standard normal distribution, i.e., α1≈0.003, α2≈0.05, and α3≈0.32, and σ is the standard deviation of the standard normal distribution.

[0057] It should be noted that this is based on monitoring data at the current moment. Calculate structural state parameters posterior probability density Combined with engineering thresholds Define the probability of exceeding the limit in the interval ,Right now: (5.5).

[0058] Therefore, based on the approximate posterior samples of the structural state parameters... Discrete estimation of out-of-limit probability As shown in equation (5.6): (5.6); In the formula: This is an indicator function that takes the value 1 when the condition in parentheses is true, and 0 otherwise; N is the number of samples.

[0059] It should be noted that, based on each structural state parameter In the event of a warning status, the following appropriate measures can be taken: The normal status indicates that construction can continue according to the established rotation and monitoring plan; A Level 1 warning requires a reliability check of the monitoring data at the warning monitoring points to confirm whether the monitoring data is abnormal (i.e., a reliability check), to generate risk alerts and tracking records, and to record abnormal monitoring points to provide a basis for further risk management; the rotation construction continues; Level 2 warning requires a special investigation of the structural status parameters and corresponding adjustments to the rotation construction parameters. It is necessary to conduct a special investigation of key equipment such as support / lubrication contact, traction system, braking limit, and attitude synchronization control, and implement process control and parameter adjustments under safety constraints. The rotation construction continues to prevent the risk from further amplifying. If an alarm is triggered, stop the rotation construction and conduct a comprehensive risk assessment. Immediately implement mandatory safety measures according to the emergency plan, and quickly organize emergency verification and retesting to confirm the risk and pinpoint the cause. Work can only resume after the risk indicators have subsided and the conditions for resuming work have been met and approved.

[0060] Based on step four, this step can quickly calculate the structural state parameters. The probability of exceeding limits is calculated, and then the warning status of the structural state parameters at the current moment is judged according to the hierarchical warning judgment criteria. Structural state parameters can reflect the inherent mechanism of bridge rotation construction. Compared with relying on monitoring indicators and their empirical thresholds, the warning method based on structural state parameters and their probability of exceeding limits can better reflect the risk of bridge rotation construction, improving the accuracy of the warning. While retaining the probability quantification capability of Bayesian inference, the computational complexity is significantly reduced, which can significantly improve the timeliness of the warning status judgment, so as to take timely risk mitigation measures and improve the safety of the rotation construction. In other words, this invention significantly improves computational efficiency and warning efficiency while ensuring high warning accuracy, and realizes the transformation from traditional monitoring indicator threshold alarms to a refined warning mode driven by mechanism parameters.

[0061] Implementation method two; Please see Figure 3 This embodiment provides a bridge rotation construction status risk classification and early warning system, including: Data acquisition module 100 is used to collect monitoring data; The data processing module 200 is used to calculate the probability of exceeding the limits of structural state parameters and determine the corresponding early warning state; The interactive module 300 is used to select structural state parameters and display the warning status of the structural state parameters; The data storage module 400 is used to store monitoring data, probability of exceeding limits, and early warning status.

[0062] It should be noted that the input terminal of the data processing module 200 is connected to the data acquisition module 100, and the data processing module 200 is connected to the interaction module 300; the input terminal of the data storage module 400 is connected to the data acquisition module 100, and the data storage module 400 is connected to the data processing module 200. The interaction module 300 can be a touch screen, which can display multiple structural state parameters. The user can select the desired structural state parameters. The engineering threshold values ​​for the structural state parameters can be input into the interaction module 300, or they can be pre-set and stored in the data storage module 400, and then retrieved when calculating the probability of exceeding limits. The graded early warning discrimination criteria based on the probability of exceeding limits can also be pre-stored in the data storage module 400.

[0063] It should be noted that the method for calculating the probability of exceeding the limit of structural state parameters and the method for judging the early warning state have been described in detail in Implementation Method 1, and will not be elaborated on in this implementation method.

[0064] Through the above description of the embodiments, those skilled in the art can clearly understand that the above methods can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0065] The applicant further declares that while the above embodiments illustrate the implementation method of the present invention, the present invention is not limited to the above-described embodiments, meaning that the present invention does not necessarily rely on the above methods and structures for implementation. Those skilled in the art should understand that any improvements to the present invention, equivalent substitutions for the selected implementation methods, additions to steps, and selections of specific methods all fall within the protection and disclosure scope of the present invention.

Claims

1. A method for risk classification and early warning of bridge rotation construction status, characterized in that: Includes the following steps: Step 1: Determine the structural state parameters of the bridge rotation and set up a sensor monitoring network to acquire monitoring data during the rotation construction process; Step 2: Assuming that the structural state parameters follow a Laplace prior, and that the monitoring data and structural state parameters satisfy a linear Gaussian observation relationship, the posterior distribution of the structural state parameters is derived using Bayesian inference. A mixture variable is introduced, and it is assumed that the mixture variable follows a product-type exponential distribution, so that the Laplace prior distribution of the structural state parameters can be expressed as a Gaussian-scale mixture. Then, the posterior mixture distribution of the structural state parameters is derived. Step 3: Based on the prior and posterior distributions of the mixed variables, construct their log-posterior distribution, and derive the corresponding gradient and Hessian matrix accordingly. Step 4: Using the Laplace approximation method, maximize the log-posterior distribution of the mixed variables to solve for the maximum a posteriori point of the mixed variables. Based on the r-dimensional positive components of the maximum a posteriori point, the mixed variables are divided into r-dimensional components and the remaining components. Based on the r-dimensional components and the Hessian matrix at the maximum a posteriori point, construct the r-dimensional approximate marginal a posteriori mixture distribution of the mixed variables. Then, combined with the prior distribution of the remaining components, derive the dimension-reduced approximate a posteriori mixture distribution of the overall mixed variables. Replace the corresponding posterior distribution with this distribution to obtain the approximate posterior distribution of the structural state parameters. Sample the r-dimensional components and the remaining components separately to obtain the approximate posterior samples of the structural state parameters. Step 5: Based on the principle of probabilistic risk control, establish a graded early warning discrimination criterion based on the probability of exceeding limits; calculate the corresponding probability of exceeding limits based on the approximate posterior sample of the structural state parameters at the current moment and the engineering threshold, and determine the early warning status of the structural state parameters at the current moment according to the graded early warning discrimination criterion.

2. The bridge rotation construction status risk classification and early warning method according to claim 1, characterized in that: In step one, the structural state parameters include at least one of the following: friction characteristic parameters, stiffness parameters, rotational parameters, contact state parameters, geometric eccentricity parameters, and environmental coupling parameters.

3. The bridge rotation construction status risk classification and early warning method according to claim 1, characterized in that: In step two, the observation noise in the linear Gaussian observation relationship follows a zero-mean Gaussian distribution.

4. The bridge rotation construction status risk classification and early warning method according to claim 1, characterized in that: In step two, the posterior mixture distribution of the structural state parameters is derived as follows: The corresponding posterior distribution is derived from the prior distribution and likelihood function of the structural state parameters under mixed variable conditions. The corresponding posterior distribution is derived from the prior distribution and likelihood function of the mixed variables; Establish the joint posterior distribution of structural state parameters and mixed variables based on the two posterior distributions; The posterior mixed distribution is obtained by integrating the mixture variables in the posterior joint distribution.

5. The bridge rotation construction status risk classification and early warning method according to claim 1, characterized in that: In step three, the first derivative of the log-posterior distribution of the mixed variables is used as the gradient of the mixed variables, and the second derivative of the log-posterior distribution of the mixed variables is used as the Hessian matrix of the mixed variables.

6. The bridge rotation construction status risk classification and early warning method according to claim 1, characterized in that: In step four, the r-dimensional approximate marginal posterior mixture distribution of the mixture variables is multiplied by the prior distributions of the remaining components of the mixture variables to obtain the dimension-reduced approximate posterior mixture distribution of the mixture variables as a whole.

7. The bridge rotation construction status risk classification and early warning method according to claim 1, characterized in that: In step five, based on the principle of probabilistic risk control, three levels of early warning thresholds are set, and four early warning states are established accordingly as a tiered early warning discrimination criterion based on the probability of exceeding limits: Normal state: ; Level 1 Warning: ; Level 2 warning: ; Alarm status: ; In the formula: α1 represents the probability of structural state parameters exceeding the engineering threshold range; α2, α3 are graded early warning thresholds, and α1, α2, and α3 correspond to the two-tailed probabilities of 3 times, 2 times, and 1 times the standard normal distribution, respectively.

8. The bridge rotation construction status risk classification and early warning method according to claim 1, characterized in that: In step five, based on the approximate posterior sample of the structural state parameters and the engineering threshold, the probability of exceeding the limit of the structural state parameters is calculated using the following formula: ； In the formula: This is the engineering threshold; This represents the probability of exceeding the limit. For indicator functions; Let N be the nth approximate posterior sample, and N be the number of samples.

9. The bridge rotation construction status risk classification and early warning method according to any one of claims 1-8, characterized in that: In step five, based on the warning status of the structural state parameters, take the following measures: Under normal conditions, continue the rotation construction; Level 1 warning: Conduct reliability verification of monitoring data from warning monitoring points, confirm whether the monitoring data is abnormal and record abnormal monitoring points; continue the rotation construction; A Level II warning requires a special investigation of the structural status parameters that are under warning, and the rotation construction will continue after adjusting the corresponding rotation construction parameters. Alarm status detected. Stop rotation construction and conduct a comprehensive risk assessment.

10. A bridge rotation construction status risk classification and early warning system, characterized in that: include: The data acquisition module is used to collect monitoring data; The data processing module is used to calculate the probability of exceeding the limits of structural state parameters and determine the corresponding early warning status; The interactive module is used to select structural status parameters and display the warning status of structural status parameters; The data storage module is used to store monitoring data, probability of exceeding limits, and early warning status. To achieve the bridge rotation construction status risk classification and early warning method as described in any one of claims 1-8.

Images