A bridge safety assessment method based on multi-intelligent algorithm fusion

Abstract

The application discloses a bridge safety evaluation method based on multi-intelligent algorithm fusion, and belongs to the technical field of bridge engineering and intelligent algorithm. In view of problems of insufficient multi-source data processing and missing physical constraints of traditional evaluation technology, the method integrates multidimensional indexes through eight steps of multi-source data fusion preprocessing, physical information neural network construction and self-adaptive deep forest feature learning, and generates a structure vulnerability index (SVI) to guide maintenance decision-making. The method has the following innovative points: a five-algorithm deep coupling architecture, a physically embedded digital twin, and a dynamic weight proof theory. The method has the following beneficial effects: high-precision identification of bridge damage, interpretable prediction and visualized decision-making, significantly improved evaluation accuracy and operation and maintenance efficiency, and is suitable for full life cycle management of various bridges.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of bridge engineering and intelligent algorithm technology, in particular to a bridge safety evaluation method based on multi-intelligent algorithm fusion. BACKGROUND

[0002] At present, as the core component of the traffic infrastructure, the safety evaluation of the bridge is crucial to public safety and economic operation. The traditional bridge safety evaluation technology mainly relies on single sensor (such as strain gauge, accelerometer) to monitor key parameters, and combines experience formula or finite element model to judge damage, which has the following significant defects:

[0003] Inadequacy of multi-source data fusion technology:

[0004] The existing method independently processes strain, vibration, visual and other data, and does not effectively solve the data heterogeneity and environmental noise interference (such as strain drift caused by temperature change, vibration noise caused by traffic load). For example, the traditional Hampel filter uses a fixed threshold, which cannot suppress the influence of daily temperature fluctuations on sensor data, resulting in a false positive rate of more than 15%.

[0005] The modal decomposition technology (such as traditional VMD) does not consider signal distribution matching and frequency interval constraint, which is easy to cause modal aliasing and difficult to accurately separate abnormal vibration modes caused by damage (such as the frequency overlap problem of cable relaxation and main beam vibration of a cable-stayed bridge).

[0006] Disconnection between physical constraints and data-driven models:

[0007] Although pure data-driven models (such as LSTM, CNN) can fit historical data, they lack structural mechanics constraints, and the prediction results under extreme working conditions (such as overload, earthquake) may violate physical laws (such as the nonlinear relationship between simply supported beam mid-span displacement and load is ignored), resulting in insufficient evaluation reliability.

[0008] The traditional finite element model relies on idealized material parameters and boundary conditions, and has weak real-time updating capability (the model updating period is usually ≥24 hours), and cannot reflect the material degradation (such as the decrease of elastic modulus caused by concrete carbonation) and geometric deformation (such as the decrease of cross-section stiffness caused by crack propagation) during the service of the bridge.

[0009] Single algorithm is difficult to deal with complex damage patterns:

[0010] The bridge damage types are various (such as steel structure fatigue crack, concrete crack, cable force loss), and the learning ability of the traditional single classifier (such as random forest) to high-dimensional nonlinear features (such as vibration signal kurtosis, crack fractal dimension) is limited, and the damage identification accuracy is generally lower than 80%.

[0011] The parameter optimization process usually adopts a single objective (such as only optimizing identification accuracy), ignores reasoning efficiency and model sensitivity, and leads to problems such as "high accuracy but long calculation time" or "high sensitivity leading to false alarm" in actual application. SUMMARY

[0012] The purpose of the present application is to provide a bridge safety evaluation method based on multi-intelligent algorithm fusion, which solves the problems of multi-source data processing, physical constraint modeling, multi-algorithm cooperation and visual decision of traditional technology.

[0013] To achieve the above purpose, the present application provides a bridge safety evaluation method based on multi-intelligent algorithm fusion, comprising the following steps:

[0014] S1, cleaning the bridge sensor data, visual data and environmental data, and performing modal decomposition on the sensor data through improved variational modal decomposition to generate a high-dimensional feature vector containing vibration modal features, image frequency spectrum features and environmental compensation strain;

[0015] S2, fusing sensor time sequence features and finite element physical constraints through a double-channel network, embedding a beam bending equation as a physical loss term, and outputting a structure response prediction value;

[0016] S3, using a four-level cascade forest structure, combining tree depth dynamic adjustment and feature importance feedback mechanism, and extracting high-dimensional nonlinear damage features;

[0017] S4, taking damage identification accuracy, reasoning time and vulnerability index sensitivity as objective functions, and generating a multi-objective optimal parameter combination through an improved firefly algorithm;

[0018] S5, establishing a hierarchical Bayesian model, and calculating the confidence interval of the prediction result through Hamilton Monte Carlo sampling;

[0019] S6, dynamically adjusting the multi-model weight according to the real-time prediction error, and proving the weight convergence through Lyapunov stability theory;

[0020] S7, real-time updating of the finite element model, combining CUDA parallel computing and phase field method simulation, realizing structure stress cloud rendering and crack evolution prediction;

[0021] S8, integrating structure strength, dynamic characteristics and fatigue life indicators, generating a comprehensive evaluation index through adaptive weight, and guiding hierarchical maintenance decision.

[0022] Preferably, in step S1, the abnormal value processing adopts an improved Hampel filter, and the threshold calculation formula is ;

[0023] wherein, The standard deviation is multiplied by the MAD, and t is time. The dynamic factor reflects the diurnal environmental impact.

[0024] Preferably, in step S1, the mathematical optimization model of the improved variational mode decomposition is:

[0025] ;

[0026] wherein, is a set of mode components obtained by decomposition, is a set of center frequencies of the mode components; is a mode component, is a center frequency, KL The divergence constraint simulates the signal distribution consistent with the observation signal distribution, and the frequency interval constraint avoids mode aliasing. is the total number of mode components, is a unit impulse function, is a partial differential operator, is an imaginary unit, is a complex exponential modulation factor for frequency modulation of the signal; is KL a weight coefficient of the divergence constraint term, is KL divergence, used to measure the difference between two probability distributions and ; is a weight coefficient of the frequency interval constraint term; The difference between the center frequencies of the adjacent two modes and is taken norm square.

[0027] Preferably, in step S2, the dual-channel network is a dual-channel architecture of the physical information neural network (PINN), which specifically includes:

[0028] Sensor data channel: extract time sequence features of strain and vibration signals through 1D-CNN;

[0029] Physical constraint channel: input finite element parameters, and embed beam bending equation as a physical loss term.

[0030] Preferably, in step S3, a four-level cascade forest structure is adopted, combined with a tree depth dynamic adjustment and a feature importance feedback mechanism, wherein the four-level cascade forest structure is:

[0031] First layer: 10 random forests, with a feature random sampling rate of 0.8;

[0032] Second layer: 8 completely random forests, with single feature segmentation;

[0033] Third layer: 6 gradient boosting forests, learning rate 0.05;

[0034] Fourth layer: 4 XGBoost forests, maximum depth 7.

[0035] Preferably, in step S4, a multi-objective optimal parameter combination is generated by improving the firefly algorithm with the damage identification accuracy, inference time and vulnerability index sensitivity as the objective function, specifically including:

[0036] ;

[0037] wherein, is the first optimization objective function, used to measure the accuracy loss of the model; is the second optimization objective function, used to measure the inference time consumption of the model; is the third optimization objective function, used to measure the sensitivity of the model to bridge damage; is the parameter to be optimized, is the input evaluation model, is the damage identification accuracy, is the inference time, and is the vulnerability index gradient.

[0038] Preferably, in step S5, a hierarchical Bayesian model is established, specifically as follows:

[0039] Observation layer: ;

[0040] wherein, is the observation data; is the model parameter; is the observation noise variance; is the normal distribution; is the model prediction value; is the unit matrix;

[0041] Parameter layer: ;

[0042] wherein, is the parameter dimension; is the precision hyperparameter, controlling the concentration of the prior distribution of the model parameter ; is the precision corresponding to the i-th parameter; is the product distribution; is the prior normal distribution, assuming that the model parameter obeys a normal distribution with mean 0 and variance ;

[0043] Hyperparameter layer: ;

[0044] where, is a gamma distribution, is a hyperparameter provides a prior distribution, controlling its statistical properties; is a gamma distribution hyperparameter, controls the shape of the distribution, controls the scale; is a gamma distribution hyperparameter, controlling the prior distribution of the observation noise precision; is the noise precision;

[0045] 95% confidence intervals of the posterior predictive distribution are computed by Hamiltonian Monte Carlo sampling.

[0046] Preferably, in step S7, the finite element model is updated in real time to be a three-dimensional digital twin, and the construction steps thereof include:

[0047] Real-time stress cloud rendering: based on CUDA parallel computing, the strain value is mapped to RGB color;

[0048] Crack evolution prediction: the phase field method is used to simulate crack propagation.

[0049] Preferably, in step S8, the formula integrating the structural strength, dynamic characteristic and fatigue life indicators is a structure vulnerability index SVI, and the expression thereof is:

[0050] ;

[0051] wherein, SSI is a structural strength indicator, DFI is a dynamic characteristic indicator, FLI is a fatigue life indicator, and the weights α, β and γ are adaptively adjusted through a linear model containing the indicator change rate and environmental severity.

[0052] Preferably, the strength, dynamic and fatigue indicators are integrated, the adaptive weights reflect the damage development speed and environmental influence, and a quantitative evaluation result is generated to guide maintenance decision.

[0053] Therefore, the bridge safety evaluation method based on the multi-intelligent algorithm fusion has the following beneficial effects:

[0054] (1) The present application proposes a "five-algorithm deep coupling" framework (PINN, ADF, MOFA, Bayesian model, and dynamic weight mechanism), which first dynamically integrates a physical information neural network with a four-level cascade forest and a multi-objective optimization algorithm, thereby breaking through the performance bottleneck of traditional single algorithms.

[0055] (2) Interpretability: PINN and digital twin ensure that the prediction results conform to the laws of mechanics and support damage mechanism analysis.

[0056] (3) Robustness: Adaptive mechanisms and uncertainty quantification adapt to complex working conditions, reducing the false alarm rate to 2.1%;

[0057] (4) Decision support: 3D visualization and the SVI index provide intuitive assessment results, reducing maintenance costs by 40% and increasing early warning time by 320%.

[0058] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a structural diagram of a bridge safety assessment method based on the fusion of multiple intelligent algorithms in the present invention. DETAILED DESCRIPTION

[0060] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0061] Unless otherwise defined, the technical or scientific terms used in the present invention shall have the usual meanings understood by persons of ordinary skill in the field to which the present invention belongs. The words "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Words such as "include" or "comprise" mean that the elements or objects preceding the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0062] Example

[0063] like Figure 1 As shown, the present invention provides a bridge safety assessment method based on the fusion of multiple intelligent algorithms, comprising the following steps:

[0064] S1. Cleaning bridge sensor data, visual data, and environmental data, and performing modal decomposition on the sensor data using improved variational modal decomposition to generate a high-dimensional feature vector containing vibration modal features, image spectrum features, and environmental compensation strain;

[0065] S2, through the dual-channel network, the sensor timing characteristics and the finite element physical constraints are integrated, the beam bending equation is embedded as the physical loss term, and the structural response prediction value is output;

[0066] S3, adopt four-level cascade forest structure, combined with tree depth dynamic adjustment and feature importance feedback mechanism, extract high-dimensional nonlinear damage features;

[0067] S4, taking damage identification accuracy, reasoning time and vulnerability index sensitivity as objective function, generating multi-objective optimal parameter combination through improved firefly algorithm;

[0068] S5, establish hierarchical Bayesian model, calculate the confidence interval of the prediction result by Hamilton Monte Carlo sampling;

[0069] S6, dynamically adjust the multi-model weight according to the real-time prediction error, and prove the weight convergence through Lyapunov stability theory;

[0070] S7, real-time update of finite element model, combined with CUDA parallel computing and phase field simulation, realize structure stress nephogram rendering and crack evolution prediction;

[0071] S8, integrate structure strength, dynamic characteristics and fatigue life index, generate comprehensive evaluation index through adaptive weight, guide hierarchical maintenance decision.

[0072] Among them, data collection: 200 strain sensors (100Hz), 50 vibration accelerometers (500Hz) real-time data acquisition, unmanned aerial vehicle to obtain crack image (0.1mm precision);

[0073] Pretreatment: improved Hampel filter to remove temperature drift, IVMD decomposition of vibration signal, identification of 3.2~4.0Hz abnormal resonance band (cable force loss warning). The improved Hampel filter is used for outlier processing, and its threshold calculation formula is ;

[0074] Among them, is the standard deviation multiple, MAD is the median absolute deviation, t is the time, and the dynamic factor reflects the daily cycle of environmental impact.

[0075] PINN modeling: double-channel network input sensor data and finite element parameters, inversion of cable force loss rate (such as No. 3 cable loss 18.7%), physical loss term to ensure that the inversion result meets the beam theory.

[0076] The double-channel architecture of physical information neural network (PINN) includes:

[0077] Sensor data channel: extract time series features of strain and vibration signals through 1D-CNN;

[0078] Physical constraint channel: input finite element parameters, embed beam bending equation as physical loss term.

[0079] The mathematical optimization model of the improved variational modal decomposition is:

[0080] ;

[0081] wherein, is a set of each modal component obtained by decomposition, is a set of center frequencies of each modal component; is a modal component, is a center frequency, KL the divergence constraint simulates the signal distribution consistent with the observation signal distribution, and the frequency interval constraint avoids modal aliasing; is the total number of modal components, is a unit impulse function, is a partial differential operator, is an imaginary unit, is a complex exponential modulation factor for frequency modulation of the signal; is KL a weight coefficient of the divergence constraint term, is KL divergence, used to measure the difference between two probability distributions and ; is a weight coefficient of the frequency interval constraint term; the difference between the center frequencies of two adjacent modes and is taken norm square.

[0082] ADF feature learning: four-level forest extracts crack width, vibration main frequency and other features, and gradient boosting forest focuses on micro crack identification. The adaptive deep forest (ADF) adopts a four-level cascade structure:

[0083] First layer: 10 random forests, feature random sampling rate 0.8;

[0084] Second layer: 8 completely random forests, single feature segmentation;

[0085] Third layer: 6 gradient boosting forests, learning rate 0.05;

[0086] Fourth layer: 4 XGBoost forests, maximum depth 7.

[0087] MOFA optimization: balance accuracy (98%) and inference time (180ms) to generate the optimal model parameter combination. The objective function of multi-objective glowworm optimization includes:

[0088] ;

[0089] wherein, For the first optimization objective function, which measures the precision loss of the model; For the second optimization objective function, which measures the inference time consumption of the model; For the third optimization objective function, which measures the sensitivity of the model to bridge damage; For the to-be-optimized parameters, For the input of the evaluation model, For the damage identification accuracy, For the inference time, and For the vulnerability index gradient.

[0090] BUQ analysis: Bayesian uncertainty quantification is achieved through hierarchical Bayesian models;

[0091] Observation layer:

[0092] Wherein, is the observed data; is the model parameter; is the observation noise variance; is the normal distribution; is the model prediction value; is the unit matrix;

[0093] Parameter layer:

[0094] Wherein, is the parameter dimension; is the precision hyperparameter, which controls the concentration of the prior distribution of the model parameter corresponding to the i-th parameter; is the precision; is the product distribution; is the prior normal distribution, which assumes that the model parameter obeys a normal distribution with mean 0 and variance ;

[0095] Hyperparameter layer:

[0096] Wherein, is the gamma distribution, which provides the prior distribution for the hyperparameter and controls its statistical characteristics; is the gamma distribution hyperparameter, which controls the shape of the distribution, and controls the scale; is the gamma distribution hyperparameter, which controls the prior distribution of the observation noise precision; is the noise precision;

[0097] The 95% confidence interval of the posterior predictive distribution is calculated through Hamilton Monte Carlo sampling.​​

[0098] The remaining life prediction confidence interval is [6.21, 8.15] years, providing probabilistic support for maintenance planning.

[0099] Digital twin: Unity3D engine real-time rendering stress cloud map, AR label damage cable position, guide accurate maintenance.

[0100] SVI calculation: The structural vulnerability index SVI calculation formula is:

[0101] ;

[0102] Where SSI is the structural strength index, DFI is the dynamic characteristic index, and FLI is the fatigue life index. The weights a, b, and g are adaptively adjusted through a linear model containing the index change rate and environmental severity.

[0103] Combined with SSI (strength degradation), DFI (frequency shift), and FLI (fatigue consumption), the adaptive weight generates SVI=0.62, triggering the load limit passage decision.

[0104] Finally, the strength, dynamic, and fatigue indexes are integrated, and the adaptive weight reflects the damage development speed and environmental influence to generate a quantitative evaluation result to guide maintenance decisions.

[0105] Taking a 300m span cable-stayed bridge as an example:

[0106] Data collection: 200 strain sensors (100Hz), 50 accelerometers (500Hz) for real-time sampling, and a drone for 0.1mm precision crack image acquisition.

[0107] Preprocessing verification: Improved Hampel filter eliminates strain drift (error from 18% to 3.5%) caused by daily temperature difference (20℃), and IVMD separates the cable relaxation characteristic frequency 3.6Hz.

[0108] PINN inversion: Dual-channel network inputs sensor data and finite element parameters, inverts cable force loss of No. 3 cable by 18.7%, and physical loss term ensures that the inversion result has an error of <1% compared with the measured cable force.

[0109] ADF feature extraction: Four-level forest extracts crack fractal dimension (1.72), vibration main frequency shift (4.2Hz→3.8Hz), and other features, with damage identification accuracy of 98.6%.

[0110] SVI decision: Integrating SSI (strength degradation 15%), DFI (frequency shift 9.3%), and FLI (fatigue consumption 28%), generates SVI=0.62, triggering the maintenance decision of limiting the load of the main span by 50%.

[0111] Therefore, the application realizes high-precision identification, interpretable prediction and visual decision of bridge damage through deep coupling of data preprocessing, physical information neural network, deep forest, multi-objective optimization, uncertainty quantification, digital twin and other technologies, significantly improves the evaluation accuracy and operation efficiency, and is suitable for the whole life cycle management of various bridges.

[0112] And the application forms a complete technical chain from multi-source data preprocessing (S1) to digital twin decision (S8):

[0113] Data layer: improve Hampel filter (dynamic threshold formula) and IVMD (KL divergence constraint) to suppress environmental noise;

[0114] Model layer: ADF four-level forest (random forest XGBoost) extracts 23-dimensional nonlinear features, and MOFA optimizes the objective function group (accuracy / reasoning time / SVI gradient);

[0115] Application layer: digital twin renders stress nephogram through CUDA, and SVI index (alpha, beta, gamma adaptive weight) guides hierarchical maintenance, and the integrity of the technical link is better than that of the traditional scheme.

[0116] Finally, it should be pointed out that: the above examples are only used to illustrate the technical solutions of the application and not to limit them, although the application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that: it can still modify or equivalently replace the technical solutions of the application, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the application.

Claims

1. A bridge safety assessment method based on multi-intelligent algorithm fusion, characterized in that: The following steps are involved: S1. Cleaning bridge sensor data, visual data, and environmental data, and performing modal decomposition on the sensor data using improved variational modal decomposition to generate a high-dimensional feature vector containing vibration modal features, image spectrum features, and environmental compensation strain; S2, through the dual-channel network, the sensor timing characteristics and the finite element physical constraints are integrated, the beam bending equation is embedded as the physical loss term, and the structural response prediction value is output; S3, using a four-level cascade forest structure, combined with dynamic tree depth adjustment and feature importance feedback mechanism to extract high-dimensional nonlinear damage features; S4, taking damage identification accuracy, inference time and vulnerability index sensitivity as objective functions, the multi-objective optimal parameter combination is generated by improving the firefly algorithm; S5. Establish a hierarchical Bayesian model and calculate the confidence interval of the prediction results through Hamiltonian Monte Carlo sampling; S6. Dynamically adjust the weights of multiple models according to the real-time prediction error and prove the convergence of weights through Lyapunov stability theory; S7, real-time update of finite element models, combined with CUDA parallel computing and phase field simulation, to achieve structural stress cloud rendering and crack evolution prediction; S8. Integrate structural strength, dynamic characteristics and fatigue life indicators, generate a comprehensive evaluation index through adaptive weighting, and guide hierarchical maintenance decisions.

2. The bridge safety assessment method based on multi-intelligent algorithm fusion according to claim 1 is characterized by: In step S1, the improved Hampel filter is used for outlier processing, and the threshold calculation formula is: ; in, is the multiple of standard deviation, MAD is the median absolute deviation, t is time, and the dynamic factor reflects the impact of the daily cycle environment.

3. The bridge safety assessment method based on multi-intelligent algorithm fusion according to claim 1 is characterized by: In step S1, the mathematical optimization model of the improved variational mode decomposition is: ； in, is the set of modal components obtained by decomposition, is the set of center frequencies of each modal component; is the modal component, is the center frequency, KL Divergence constraints ensure that the simulated signal distribution is consistent with the observed signal, and frequency interval constraints avoid modal aliasing; is the total number of modal components, is the unit impulse function, is the partial differential operator, is the imaginary unit, is a complex exponential modulation factor, used to frequency modulate the signal; for KL The weight coefficient of the divergence constraint term, for KL Divergence, used to measure the divergence of two probability distributions and the differences between; is the weight coefficient of the frequency interval constraint; For two adjacent modal center frequencies and Difference Norm squared.

4. The bridge safety assessment method based on multi-intelligent algorithm fusion according to claim 1 is characterized by: In step S2, the dual-path network is a dual-path architecture of a physical information neural network PINN, specifically including: Sensor data path: Extracting the temporal features of strain and vibration signals through 1D-CNN; Physical constraint path: Input finite element parameters and embed the beam bending equation as a physical loss term.

5. The bridge safety assessment method based on multi-intelligent algorithm fusion according to claim 1 is characterized by: In step S3, a four-level cascade forest structure is adopted, combining dynamic adjustment of tree depth and feature importance feedback mechanism, where the four-level cascade forest structure is: First layer: 10 random forests, feature random sampling rate 0.8; Second layer: 8 completely random forests, single feature segmentation; The third layer: 6 gradient boosting forests, learning rate 0.05; Fourth layer: 4 XGBoost forests with a maximum depth of 7.

6. The bridge safety assessment method based on multi-intelligent algorithm fusion according to claim 1 is characterized by: In step S4, the damage identification accuracy, inference time, and vulnerability index sensitivity are used as objective functions, and the improved firefly algorithm is used to generate a multi-objective optimal parameter combination, specifically including: ； in, It is the first optimization objective function, used to measure the accuracy loss of the model; It is the second optimization objective function, used to measure the inference time of the model; is the third optimization objective function, which measures the sensitivity of the model to bridge damage; is the parameter to be optimized, For input The evaluation model, is the damage identification accuracy, is the inference time, is the vulnerability index gradient.

7. The bridge safety assessment method based on multi-intelligent algorithm fusion according to claim 1 is characterized by: In step S5, a hierarchical Bayesian model is established, as shown below: Observation layer: ; in, is the observation data; are model parameters; is the observation noise variance; is a normal distribution; is the model prediction value; is the identity matrix; Parameter layer: ; in, is the parameter dimension; is the accuracy hyperparameter, controlling the model parameters The degree of concentration of the prior distribution, The precision corresponding to the i-th parameter; is the product distribution; Assume that the model parameters are normal distributions. The mean is 0 and the variance is Normal distribution; Hyperparameter layer: ; in, is the gamma distribution, and is the hyperparameter Provide a prior distribution and control its statistical properties; is the gamma distribution hyperparameter, Control the shape of the distribution, Control scale; is the gamma distribution hyperparameter, which controls the prior distribution of the observation noise accuracy; is the noise accuracy; Compute 95% confidence intervals for the posterior predictive distribution via Hamiltonian Monte Carlo sampling.

8. The bridge safety assessment method based on multi-intelligent algorithm fusion according to claim 1 is characterized by: In step S7, the finite element model is updated in real time to a three-dimensional digital twin, and the construction steps include: Real-time stress cloud rendering: Based on CUDA parallel computing, the strain value is mapped to RGB color; Crack evolution prediction: Phase field method is used to simulate crack propagation.

9. The bridge safety assessment method based on multi-intelligent algorithm fusion according to claim 1 is characterized by: In step S8, the formula integrating the structural strength, dynamic characteristics and fatigue life index is the structural vulnerability index SVI, which is expressed as follows: ； Among them, SSI is the structural strength index, DFI is the dynamic characteristics index, FLI is the fatigue life index, and the weight is Adaptive adjustment is made through a linear model that includes the indicator change rate and environmental severity.

10. The bridge safety assessment method based on multi-intelligent algorithm fusion according to claim 1 is characterized by: Integrating strength, dynamics, and fatigue indicators, adaptive weights reflect damage development speed and environmental impacts, and generate quantitative assessment results to guide maintenance decisions.

Images