Search

CN-122023921-A - Highway bridge damage identification method and system

CN122023921ACN 122023921 ACN122023921 ACN 122023921ACN-122023921-A

Abstract

The invention relates to the technical field of bridge structure health monitoring, in particular to a highway bridge damage identification method and system, comprising the following steps: S1, acquiring surface images and structure physical response data of a bridge structure through an unmanned plane and multiple types of sensors, and carrying out standardization and noise suppression on the acquired data. According to the invention, through converting bridge damage identification into an unsupervised multi-objective optimization problem integrating physical constraint and priori knowledge, and combining with the dynamic parameter regulation and control and hierarchical search strategy of an improved bat algorithm, the accurate inversion and optimal solution output of the real damage state of the bridge are realized, the accuracy and efficiency of damage identification are effectively improved, the reliability quantification of the identification result is completed based on historical data and maintenance records, the core parameters of the bat algorithm are dynamically optimized through a hierarchical feedback mechanism, closed loop iterative optimization is formed, and the stability, the self-adaptability and the engineering reliability of the identification performance of the system are continuously improved.

Inventors

  • Yu Fenyi
  • QIU YANHUI
  • YU SU

Assignees

  • 广州通辉工程有限公司

Dates

Publication Date
20260512
Application Date
20260203

Claims (10)

  1. 1. A method for identifying damage to a highway bridge, comprising the steps of: S1, acquiring surface images and structure physical response data of a bridge structure through an unmanned plane and a plurality of types of sensors, and carrying out standardization and noise suppression on the acquired data; S2, respectively extracting apparent damage characteristics of crack textures and spalling areas and structural dynamics damage characteristics of vibration modes and frequency changes from the preprocessed multi-source data, and fusing and constructing a high-dimensional damage characteristic vector set; S3, modeling a bridge damage identification problem as a multidimensional constraint optimization problem, constructing an unsupervised fitness function based on structural response reconstruction residual error, multisource feature physical consistency and damage sparsity priori, and intelligently searching an optimal damage combination solution in a high-dimensional feature space by utilizing an improved bat algorithm; S4, comparing the similarity of the optimal solution obtained by optimizing the algorithm with a preset typical damage database, and judging the damage type and quantitatively evaluating the severity level; And S5, combining historical monitoring data with actual maintenance records, performing reliability assessment on the damage identification result, constructing a layered feedback optimization mechanism which uses an identification error and an algorithm convergence speed as a multi-target fitness function based on the damage type and the grade, feeding back the assessment error to an algorithm model, and dynamically optimizing iteration parameters of a bat algorithm.
  2. 2. A highway bridge damage identification system according to claim 1, comprising a data acquisition and preprocessing unit (1), a multi-mode damage feature fusion unit (2), an unsupervised damage inversion optimization unit (3), a damage mode matching and quantization evaluation unit (4), and a credibility evaluation and parameter adaptive regulation unit (5), wherein: the data acquisition and preprocessing unit (1) is used for automatically acquiring bridge apparent images and structural physical response data through unmanned aerial vehicles and multiple types of sensors, and finishing standardized cleaning and noise suppression of the data; The multi-mode damage feature fusion unit (2) is used for automatically extracting and fusing apparent damage visual features and structural dynamics damage features from the preprocessed image and physical data to generate a unified bridge high-dimensional damage feature vector; The unsupervised damage inversion optimization unit (3) is used for constructing a damage identification problem into a multidimensional optimization model, defining an unsupervised fitness function based on a structural response reconstruction residual error, characteristic physical consistency and damage sparsity prior, driving an improved bat algorithm to execute intelligent search in a characteristic space, and outputting an optimal damage state solution; the damage mode matching and quantitative evaluation unit (4) is used for carrying out automatic similarity matching and comparison on the damage state solution output by the unsupervised damage inversion optimization unit (3) and a typical damage mode database, and carrying out judgment of damage types and quantitative evaluation of severity levels of the damage types; the credibility evaluation and parameter self-adaptive regulation and control unit (5) is used for carrying out credibility analysis on the identification result based on the historical data and maintenance records, constructing a multi-target feedback criterion according to the identification error and the convergence speed, and dynamically optimizing the core iteration parameters of the bat algorithm in the unsupervised damage inversion optimization unit (3) through a layered feedback mechanism.
  3. 3. A road bridge damage identification system according to claim 2, characterized in that the data acquisition and pre-processing unit (1) comprises a multi-source data acquisition module (11), a data cleaning and normalization module (12), wherein: The multi-source data acquisition module (11) is used for synchronously acquiring bridge apparent images and physical response data of strain and vibration by carrying high-definition camera equipment and multi-type structure sensors on an unmanned aerial vehicle; The data cleaning and normalizing module (12) is used for performing noise suppression processing on the collected multi-source data, performing normalization operation and outputting high-quality preprocessed data.
  4. 4. The highway bridge damage identification system according to claim 2, wherein the multi-modal damage signature fusion unit (2) comprises an apparent damage visual signature extraction module (21), a kinetic damage signature extraction module (22), a multi-modal signature fusion module (23), wherein: The apparent damage visual characteristic extraction module (21) is used for extracting visual characteristic vectors of apparent damage of crack textures and concrete stripping areas from the pretreated bridge image data; The dynamic damage characteristic extraction module (22) is used for extracting a characteristic vector of structural dynamic damage with vibration mode and frequency change from the preprocessed physical response data; the multi-mode feature fusion module (23) is used for carrying out fusion treatment on the apparent damage visual features and the dynamic damage features to generate a unified high-dimensional damage feature vector set.
  5. 5. The highway bridge damage identification system according to claim 2, wherein the unsupervised damage inversion optimization unit (3) comprises an optimization problem modeling module (31), an improved bat algorithm driving module (32), an optimal solution output module (33), wherein: The optimization problem modeling module (31) is used for converting the bridge damage identification problem into a multidimensional constraint optimization problem and constructing an unsupervised fitness function with structural response reconstruction residual error, characteristic physical consistency and damage sparsity priori; The improved bat algorithm driving module (32) is used for loading core logic of an improved bat algorithm, setting algorithm iteration parameters and driving the algorithm to perform intelligent searching in a high-dimensional feature space; The optimal solution output module (33) is used for outputting an optimal damage combined solution capable of representing the bridge damage state by the optimal solution obtained by the screening algorithm search.
  6. 6. The highway bridge damage identification system according to claim 5, wherein the optimization problem modeling module (31) converts the bridge damage identification problem into a multidimensional constraint optimization problem, and constructs an unsupervised fitness function with a structure response reconstruction residual error, characteristic physical consistency and damage sparsity prior, and the specific operations are as follows: A1 definition of injury variable The method comprises the steps of including damage position and damage degree parameters, converting bridge damage identification problems into multidimensional constraint optimization problems taking damage variables as optimization variables, and setting The value range of (1) is [0,1], wherein 0 represents no damage and 1 represents complete damage; A2 calculating the structural response reconstruction residual By actually measuring the physical response vector Y and the simulation response vector Root mean square error between by the following expression: ; Wherein, the Physical response data for the actual structure acquired for the ith sensor, To candidate solutions based on current injury Theoretical response data of the ith channel simulated by the model of (a), wherein n is the total dimension of the response data; A3, calculating a physical consistency index C of the features, quantifying the matching degree of the visual features of the apparent damage and the dynamic features of the structure, wherein the value range of C is [0,1], and the closer to 1, the stronger the consistency is represented; a4, introducing a damage sparsity priori index S, and applying sparsity constraint to damage variables based on the characteristics of 'local concentration and overall sparsity' of bridge damage, wherein the specific expression is as follows: ; wherein m is the number of damage variables; a5, constructing an unsupervised fitness function, wherein the specific expression is as follows: ; In the formula, And (3) for presetting the positive weight coefficient, performing multi-objective optimization with minimum reconstruction residual error, strongest characteristic consistency and optimal damage sparsity through the function.
  7. 7. The highway bridge damage identification system of claim 5, wherein the improved bat algorithm driver module (32) loads the core logic of the improved bat algorithm, sets algorithm iteration parameters, and the driver algorithm performs intelligent searching in the high-dimensional feature space, and specifically operates as follows: b1, loading improved bat algorithm core logic, and initializing algorithm iteration parameters including bat group size, initial flight step length, pulse emission probability, maximum iteration times and search space boundaries; b2, introducing bridge damage characteristic space priori knowledge, dynamically adjusting flying step length and pulse emission probability of bat individuals, and carrying out self-adaptive switching on global exploration and local optimization; wherein the dynamically adjusted pulse emission probability is as follows: ; In the formula, For the pulse emission probability of the ith bat at the t-th iteration, As the minimum value of the probability of pulse transmission, The maximum value of the pulse emission probability is represented by k, the attenuation coefficient is represented by t, and the current iteration times are represented by t; the dynamic adjustment flight step length is as follows: ; In the formula, For the flight step of the ith bat at the t-th iteration, As the weight of the inertia is given, For the flight step of the ith bat at the t-1 th iteration, For the current global optimal solution, For the i-th bat t-1 moment, Is a random number; B3, adopting a hierarchical search strategy, enabling a global exploration layer to randomly fly through the bat group to traverse the high-dimensional feature space, and enabling a local optimization layer to be based on Carrying out neighborhood refinement search, and optimizing a search path by combining the spatial distribution characteristics of the damage features; Wherein, the local optimization is as follows: ; In the formula, For the position of the ith bat at the t-th iteration, For the position of the ith bat at iteration t-1, As a factor of the random disturbance, The average flight step length of the group at the moment t; b4, after each iteration, updating bat individual positions and global optimal solution caches by taking an unsupervised fitness function value of an optimization problem modeling module (31) as an evaluation basis; and B5, stopping searching when the iteration times reach the maximum value or the fluctuation of the fitness function value of the optimal solution is smaller than a preset threshold value, and outputting a search result corresponding to the optimal damage characteristic combination.
  8. 8. A highway bridge damage identification system according to claim 2, wherein the damage pattern matching and quantification assessment unit (4) comprises a damage signature matching module (41), a damage class quantification assessment module (42), wherein: The damage characteristic matching module (41) is used for automatically matching the optimal damage combination solution with the characteristic templates in the preset typical damage mode database in a similarity mode; And the damage level quantitative evaluation module (42) is used for judging the damage type of the bridge according to the matching result and carrying out graded quantitative evaluation on the damage severity.
  9. 9. The highway bridge damage identification system according to claim 2, wherein the reliability evaluation and parameter adaptive regulation unit (5) comprises an identification result reliability analysis module (51), a multi-objective feedback criterion construction module (52), an algorithm parameter adaptive regulation module (53), wherein: The recognition result credibility analysis module (51) is used for combining the historical monitoring data and the actual maintenance record to quantitatively analyze the credibility of the damage discrimination and grade evaluation results; The multi-target feedback criterion construction module (52) is used for constructing a multi-target feedback optimization criterion by taking the recognition error and the algorithm convergence speed as core indexes; The algorithm parameter self-adaptive regulation and control module (53) is used for dynamically regulating core iteration parameters of the bat algorithm in the unsupervised damage inversion optimization unit (3) through a layered feedback mechanism.
  10. 10. The highway bridge damage identification system according to claim 9, wherein the algorithm parameter adaptive control module (53) dynamically adjusts core iteration parameters of the bat algorithm in the unsupervised damage inversion optimization unit (3) through a layered feedback mechanism, and specifically comprises the following operations: c1, receiving identification errors corresponding to different damage types and grades output by the credibility evaluation and parameter self-adaptive regulation and control unit (5) Convergence speed with algorithm ; C2, layering based on the damage type and the damage level, and calculating parameter adjustment weights of all layering; ; In the formula, Distributing coefficients for the weights; Setting a differential triggering threshold value for each layering, and when the identification error of a certain layering exceeds the threshold value, preferentially triggering parameter adjustment of the layering; c4 inertial weight by dynamically adjusting bat algorithm Other core iteration parameters of the attenuation coefficient k are adjusted in a similar way; ; In the formula, For adjusting the coefficients; c5, synchronizing the adjusted parameters to an improved bat algorithm driving module (32) for a next round of searching; and C6, tracking the performance of the algorithm after adjustment, and repeating the steps C2-C5 if the identification error does not reach the standard yet until the preset precision requirement is met.

Description

Highway bridge damage identification method and system Technical Field The invention relates to the technical field of bridge structure health monitoring, in particular to a highway bridge damage identification method and system. Background Highway bridges are structures built for highway traffic to cross natural or artificial obstacles (such as rivers, canyons, railways, other roads, etc.), and are important components of highway traffic infrastructure, and the structural integrity and the operation safety of the structures are directly related to the traffic efficiency of highway lines, the life and property safety of drivers and passengers, and the stable operation of regional traffic transport networks. In the full life cycle operation and maintenance process of the highway bridge, damage identification and disease monitoring are core links for guaranteeing the safety service of the bridge. However, the existing bridge damage identification method has the defects that the traditional scheme mainly depends on manual inspection or single sensor monitoring, consumes a large amount of manpower and material resources, is low in detection efficiency, is easily limited by severe environment and manual experience, has high omission rate and cannot realize real-time dynamic monitoring, and is difficult to meet the requirements of the fine operation and maintenance of the highway bridge under complex working conditions due to the fact that the physical characteristics of the bridge structure are not fully combined due to the fact that the fitness function design is partially adopted, the optimization algorithm is easy to fall into local optimization, the self-adaptive feedback mechanism for different damage types and grades is lack, and the like. Based on the above, the invention provides a method and a system for identifying damage to highway bridges, so as to solve the technical problems set forth above. Disclosure of Invention The invention aims to provide a highway bridge damage identification method and system, which are used for realizing accurate inversion and optimal solution output of a real damage state of a bridge by converting bridge damage identification into an unsupervised multi-objective optimization problem integrating physical constraint and priori knowledge and combining with a dynamic parameter regulation and layering search strategy for improving a bat algorithm, effectively improving the accuracy and efficiency of damage identification, completing the reliability quantification of an identification result based on historical data and maintenance records, dynamically optimizing core parameters of the bat algorithm through a layering feedback mechanism, forming closed loop iteration optimization, and continuously improving the stability, the self-adaptability and the engineering reliability of the identification performance of the system. In order to achieve the above purpose, the present invention provides the following technical solutions: the invention provides a highway bridge damage identification method, which comprises the following steps: S1, acquiring surface images and structure physical response data of a bridge structure through an unmanned plane and a plurality of types of sensors, and carrying out standardization and noise suppression on the acquired data; S2, respectively extracting apparent damage characteristics of crack textures and spalling areas and structural dynamics damage characteristics of vibration modes and frequency changes from the preprocessed multi-source data, and fusing and constructing a high-dimensional damage characteristic vector set; S3, modeling a bridge damage identification problem as a multidimensional constraint optimization problem, constructing an unsupervised fitness function based on structural response reconstruction residual error, multisource feature physical consistency and damage sparsity priori, and intelligently searching an optimal damage combination solution in a high-dimensional feature space by utilizing an improved bat algorithm; S4, comparing the similarity of the optimal solution obtained by optimizing the algorithm with a preset typical damage database, and judging the damage type and quantitatively evaluating the severity level; And S5, combining historical monitoring data with actual maintenance records, performing reliability assessment on the damage identification result, constructing a layered feedback optimization mechanism which uses an identification error and an algorithm convergence speed as a multi-target fitness function based on the damage type and the grade, feeding back the assessment error to an algorithm model, and dynamically optimizing iteration parameters of a bat algorithm. Based on the method, the invention also provides a highway bridge damage identification system, which comprises a data acquisition and preprocessing unit, a multi-mode damage characteristic fusion unit, an unsupervised damage inversion optimi