A bridge superstructure safety risk assessment method and system and a storage medium

Abstract

The present disclosure relates to a bridge superstructure safety risk assessment method, system and storage medium. The method comprises: obtaining disease deduction information of the technical condition of the bridge superstructure; obtaining bridge superstructure technical condition score data, bridge superstructure deduction indicators, disease classification categories of the deduction indicators of a sample bridge, and analyzing common indicators in the deduction indicators based on the same; calculating the deduction values of the disease classification categories of the common indicators at each scale based on the LSHADE algorithm, and calculating the safety risk score of the bridge superstructure of the sample bridge; dividing the safety risk level; using the CatBoost integrated learning method to construct a bridge safety risk evaluation model, training the bridge safety risk evaluation model using the frequency reduction coefficient of the sample bridge; and using the trained bridge safety risk evaluation model to assess the safety risk level of the bridge superstructure. The present disclosure has high assessment efficiency, low cost and high accuracy.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of machine learning and structural engineering, in particular to a bridge superstructure safety risk assessment method, system and storage medium. BACKGROUND

[0002] As a key node of traffic engineering, bridges bear the heavy responsibility of maintaining road smoothness and driving safety. With the increasing service time, highway bridge structures are inevitably affected by multiple factors such as environmental erosion, vehicle load and material aging, leading to the gradual degradation of their safety performance. In view of this key problem, how to achieve accurate and rapid assessment of bridge safety has become a hot and difficult point of current research.

[0003] The existing bridge safety evaluation methods mainly rely on periodic manual inspection and limited monitoring means. Although these methods have played an important role in the early bridge operation and maintenance, with the increasing complexity of bridge structures, the continuous expansion of scale and the extension of service life, it is difficult to achieve comprehensive and accurate assessment of bridge safety by relying solely on manual inspection and basic monitoring data, which limits the scientificity and efficiency of bridge safety management.

[0004] In terms of bridge regular inspection, China mainly relies on the bridge technical condition assessment method given in the standard "Highway Bridge Technical Condition Assessment Standard" (JTG TH21-2011) (hereinafter referred to as the standard) to score and classify bridges. However, this method has obvious limitations: on the one hand, the assessment process highly depends on the apparent damage information obtained by manual visual inspection, which is highly subjective and difficult to ensure consistency of results; on the other hand, the assessment results only stay at the technical condition level, which is difficult to effectively reflect the actual mechanical performance of the bridge, resulting in inaccurate assessment of structural safety.

[0005] In order to solve the problems existing in the technical condition assessment method, existing research has explored from the aspects of assessment index extraction and evaluation method improvement, aiming to improve the objectivity of the evaluation results and the reflection ability of the structural safety.

[0006] For example, the paper "Highway Bridge Site Detection and Technical Condition Assessment" ([Shao Penglei. Highway Bridge Site Detection and Technical Condition Assessment [D]. Zhengzhou University, 2016.]) combines the analytic hierarchy process and group decision-making method to establish a beam bridge technical condition evaluation hierarchical model and proposes an improved in-service bridge technical condition assessment method. However, this method, although scientifically uses the analytic hierarchy process and group decision-making method to improve the relationship between technical condition assessment and safety, is inevitably affected by expert subjective judgment, which may lead to certain differences in bridge assessment results in different evaluation groups or application scenarios.

[0007] As disclosed in Chinese Patent No. CN112668149A, a "Beam Bridge Technical Condition Parameterized Structure Modeling and Intelligent Evaluation System" is proposed, which systematically integrates field collection and background evaluation, simplifies the operation process by using parameterized modeling, and considers the disease evolution trend in scoring, taking into account the evaluation specification and actual development status, improving the scientificity and operability of the evaluation.

[0008] As disclosed in Chinese Patent No. CN109102016A, a "Bridge Technical Condition Testing Method" is proposed, which uses a long short-term memory recurrent neural network (LSTMRNN) to build a bridge technical condition prediction model that can automatically correct the weights of influencing factors, solving the problem of inaccurate bridge technical condition prediction methods.

[0009] As disclosed in Chinese Patent No. CN118606847A, a "Bridge Technical Condition Prediction Method and System Based on Machine Learning" is proposed, which preprocesses, filters and balances bridge detection data, evaluates feature importance using random forest, solves class imbalance using SMOTETomek algorithm, and finally uses XGBoost model to predict bridge technical condition grade.

[0010] As disclosed in Chinese Patent No. CN107341282A, an "Improved Bridge Deterioration Evaluation Method Based on Previous Year Technical State" is proposed, which couples four key parameters: "bridge construction technical state score", "non-deterioration time", "statistical life of similar bridges", and "operational use time", to establish a bridge technical state deterioration process model that can describe the influence of environmental, load and material changes.

[0011] Although the above researches introduce various improvement methods in bridge technical condition evaluation, improving the scientificity and convenience of the evaluation, they still mainly rely on external defects or historical state data to build evaluation models, which is difficult to directly reflect the changes in structural mechanical properties, making the evaluation results often focus on surface defect features and ignore internal stress mechanisms, making it difficult to achieve comprehensive and accurate quantification of bridge safety.

[0012] In recent years, bridge monitoring technology based on sensor networks and big data analysis has been gradually applied to bridge safety management. By collecting dynamic data such as stress, vibration and deformation of bridges, combined with long-term static data such as material aging and environmental impact, not only the overall monitoring of bridge operation state is realized, but also key data support is provided for risk assessment.

[0013] As the paper "Bridge Health Monitoring Data Storage and Early Warning Method Based on Big Data" published in the journal of Science, Technology and Engineering, the paper constructs a bridge service performance evaluation model and realizes real-time safety state warning through multi-factor analysis method to mine the correlation between real-time sensor data.

[0014] As the invention patent with publication number CN114282398A proposes "Bridge Health Monitoring System and Method Based on Big Data", by integrating monitoring data, engineering data and social big data, using WEKA and other tools for data mining, realizing real-time damage prediction, model correction and multi-level warning of bridge.

[0015] As the invention patent with publication number CN117648734A proposes "Bridge Health Intelligent Monitoring Method and Evaluation System", combining theoretical simulation and monitoring point optimization, constructing a monitoring and evaluation model, realizing structure state recognition and damage prediction based on strain deviation analysis, improving the accuracy and intelligent level of bridge health evaluation.

[0016] As the invention patent with publication number CN118549532A proposes "Highway Bridge Safety Monitoring Method and Device", by automatically collecting key structural response data, combining time-varying bearing capacity calculation model, realizing real-time monitoring and safety evaluation of bridge operation state.

[0017] However, the existing bridge superstructure safety risk assessment method still has the problems of low assessment efficiency and low accuracy. In addition, for a large number of small and medium span bridges, the existing safety risk assessment means has high cost, which makes it difficult to realize large-scale popularization and application. SUMMARY

[0018] Therefore, it is necessary to provide a bridge superstructure safety risk assessment method, system and storage medium to solve the problems of low assessment efficiency, low accuracy and high cost.

[0019] To solve the above problems, the present disclosure adopts the following technical solutions:

[0020] In a first aspect, the present disclosure provides a bridge superstructure safety risk assessment method, comprising the following steps:

[0021] Obtaining disease deduction information of the technical condition of the bridge superstructure, the disease deduction information of the technical condition including technical condition evaluation indexes and deduction values of each of the technical condition evaluation indexes at each scale, the deduction values of each of the technical condition evaluation indexes at each scale being numerical values calculated according to the analytic hierarchy process and the group decision method;

[0022] Obtaining bridge dynamic characteristic parameters, bridge superstructure structure parameters, bridge superstructure technical condition score data, bridge superstructure deduction index and disease classification categories of the bridge superstructure deduction index of a sample bridge, and analyzing common indexes in the bridge superstructure deduction index based on the above data; calculating deduction values of the disease classification categories of the common indexes at each scale based on the LSHADE algorithm, and calculating a safety risk score of the bridge superstructure of the sample bridge according to the disease classification categories of the common indexes and the deduction values of the disease classification categories of the common indexes at each scale;

[0023] According to the safety risk score of the bridge superstructure of the sample bridge, using a clustering algorithm to divide the safety risk levels;

[0024] Using a CatBoost integrated learning method to construct a bridge safety risk evaluation model, using the frequency reduction coefficient of the sample bridge, the bridge superstructure structure parameters of the sample bridge, the disease classification categories of the common indexes of the sample bridge, the deduction values of the disease classification categories of the common indexes of the sample bridge, and the safety risk level to which the bridge superstructure of the sample bridge belongs to train the bridge safety risk evaluation model, and the output of the model being the safety risk level of the bridge superstructure;

[0025] Using the trained bridge safety risk evaluation model to evaluate the safety risk level of the bridge superstructure.

[0026] In a preferred embodiment, the calculation of the deduction values of the disease classification categories of the common indexes at each scale based on the LSHADE algorithm includes: taking the maximization of the correlation between the frequency reduction coefficient of the sample bridge and the bridge superstructure safety risk score as a fitness function, using the LSHADE algorithm to obtain a deduction value correction coefficient of the disease classification categories of the common indexes, and according to the disease deduction information of the technical condition of the bridge superstructure and the deduction value correction coefficient, obtaining the deduction values of the disease classification categories of the common indexes at each scale.

[0027] In a preferred embodiment, the specific steps of training the bridge safety risk evaluation model using the frequency reduction coefficient of the sample bridge, the bridge superstructure structure parameters of the sample bridge, the disease classification categories of the common indexes of the sample bridge, the deduction values of the disease classification categories of the common indexes of the sample bridge, and the safety risk level to which the bridge superstructure of the sample bridge belongs include:

[0028] determining a main disease classification category in the disease classification category of the common index of the sample bridge;

[0029] According to the main disease category, the main classification category of the common index of the sample bridge, and according to the frequency reduction coefficient corresponding to the main disease classification category of the sample bridge, the bridge superstructure structure parameter of the sample bridge, and the safety risk level to which the bridge superstructure structure of the sample bridge belongs, the bridge safety risk evaluation model is trained.

[0030] In a preferred embodiment, the specific steps of training the bridge safety risk evaluation model further include: evaluating and verifying the integrated learning model by using model evaluation indexes, and the model evaluation indexes include accuracy, precision, recall rate and F1 score.

[0031] In a preferred embodiment, the bridge dynamic characteristic parameter includes the first-order natural frequency of the damaged bridge and the first-order natural frequency of the intact bridge; the bridge superstructure structure parameter includes the cross-sectional form of the bridge superstructure; and the frequency reduction coefficient of the sample bridge is the ratio of the first-order natural frequency of the damaged bridge to the first-order natural frequency of the intact bridge.

[0032] In a preferred embodiment, the clustering algorithm is a K-means clustering algorithm, and the number of classifications of the K-means clustering algorithm is 4.

[0033] In a preferred embodiment, the analytic hierarchy process includes: giving at least two judgment matrices for the disease deduction information of the original technical condition, and the elements of the judgment matrix represent the ratio of the influence of two technical condition evaluation indexes on the bridge technical condition; performing consistency detection on each judgment matrix, determining the weight vector of the judgment matrix for the judgment matrix that meets the consistency check, and determining the weight value of each technical condition evaluation index according to the weight vector; the group decision-making method includes: calculating the reliability coefficient according to the weight vector obtained by the analytic hierarchy process; and the value calculated according to the analytic hierarchy process and the group decision-making method is specifically: a value calculated according to the weight value, the reliability coefficient and the disease deduction information of the original technical condition.

[0034] In a preferred embodiment, the loss function of the bridge safety risk evaluation model training is:

[0035]

[0036] wherein, represents the total loss function; is the total number of training samples; represents the loss of the i-th training sample; represents the input feature vector of the i-th training sample; is the i-th training sample; is the i-th training sample.​ The true value of each training sample; For the first The model's predicted values ​​for each training sample; This indicates the total number of decision trees; Indicates the sequence number of the decision tree; Indicates the first A decision tree, For the first The regularization term of a decision tree.

[0037] Secondly, this disclosure provides a risk assessment system for the safety of bridge superstructures, including,

[0038] The acquisition module is used to acquire the technical condition deduction information of the bridge superstructure. The technical condition deduction information includes technical condition assessment indicators and the deduction value of each technical condition assessment indicator at each scale. The deduction value of each technical condition assessment indicator at each scale is a value calculated by the analytic hierarchy process and the group decision method.

[0039] The calculation module is used to acquire the bridge dynamic characteristic parameters, bridge superstructure structural parameters, bridge superstructure technical condition score data, bridge superstructure deduction indicators, and the defect classification categories of the deduction indicators of the sample bridge, and analyze the common indicators among the bridge superstructure deduction indicators accordingly; it is used to calculate the deduction values ​​of the defect classification categories of the common indicators at each scale based on the LSHADE algorithm, and to calculate the safety risk score of the bridge superstructure of the sample bridge based on the defect classification categories of the common indicators and the deduction values ​​of the defect classification categories of the common indicators at each scale.

[0040] The risk level classification module is used to classify the safety risk level of the sample bridges based on the safety risk score of the bridge superstructure and using a clustering algorithm.

[0041] The model building and training module is used to build a bridge safety risk assessment model using the CatBoost ensemble learning method. It is used to train the bridge safety risk assessment model using the frequency reduction coefficient of the sample bridge, the structural parameters of the superstructure of the sample bridge, the disease classification categories of common indicators of the sample bridge, the deduction values ​​of the disease classification categories of common indicators of the sample bridge, and the safety risk level of the superstructure of the sample bridge. The output of the model is the safety risk level of the bridge superstructure.

[0042] The safety risk assessment module is used to assess the safety risk level of the bridge superstructure using a trained bridge safety risk assessment model.

[0043] In a third aspect, the disclosure provides a computer-readable storage medium comprising instructions that, when executed on a computer, cause the computer to perform a bridge superstructure safety risk assessment method as described in the first aspect.

[0044] The bridge superstructure safety risk assessment method, system and medium of the disclosure obtain common indicators and disease classification categories in sample bridges according to disease deduction information based on the technical condition of the analytic hierarchy process and group decision-making method, accurately calculate the deduction value of the disease classification category of the common indicators at each scale by using the LSHADE algorithm, i.e., obtain the safety risk scoring rule, calculate the safety risk score of the sample bridge based on the safety risk scoring rule, divide the safety risk level according to the calculation, introduce the frequency reduction coefficient to train the bridge safety risk evaluation model constructed by the CatBoost integrated learning method, and obtain the bridge safety risk evaluation model capable of outputting the safety risk level. The risk is assessed through the model. Such design makes the assessment efficiency of the disclosure high and the cost low, and the accuracy of the safety risk assessment is improved through the design of the whole process. For a large number of small and medium span bridges, the safety risk assessment is fast and low in cost. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 It is a flowchart of the method in one embodiment of the disclosure;

[0046] Figure 2 It is a common disease information statistical chart of the bridge superstructure in one embodiment of the disclosure;

[0047] Figure 3 It is an optimal parameter convergence curve chart in one embodiment of the disclosure;

[0048] Figure 4 It is a distribution chart of the optimal combination of each deduction value correction coefficient in one embodiment of the disclosure;

[0049] Figure 5 It is a bridge superstructure safety risk level classification boundary chart in one embodiment of the disclosure;

[0050] Figure 6 It is a confusion matrix diagram of the evaluation effect in one embodiment of the disclosure;

[0051] Figure 7 It is a ROC curve diagram of the evaluation effect in one embodiment of the disclosure;

[0052] Figure 8 It is a structure diagram of the system in one embodiment of the disclosure. DETAILED DESCRIPTION

[0053] The technical solutions of the present disclosure will be described in detail below in combination with the drawings and preferred embodiments.

[0054] Term explanation:

[0055] The technical condition evaluation index is a series of indexes for evaluating the performance of bridge structure, component or whole, which can be understood as the damage category considered in the bridge safety risk evaluation. The content of the index aims to reflect the current condition and possible future performance change of the bridge. For the bridge superstructure, the technical condition evaluation index of the superstructure including super load-bearing component, super general component and support, and the technical condition evaluation index of the super load-bearing component and super general component are all 11, respectively: including ① honeycomb, pitting, ② spalling, corner drop, ③ cavity, hole, ④ concrete cover thickness, ⑤ steel corrosion, ⑥ concrete carbonization, ⑦ concrete strength, ⑧ mid-span deflection, ⑨ structural displacement, ⑩ prestressed component damage, Main beam crack. The technical condition evaluation index of the support is all 3, respectively: ① aging, metamorphic, cracking, ② defect, out-drum, ③ position string, void. It can be understood that the technical condition evaluation index in this paper has the above 25 indexes, but the number is not limited to 25, and the index content is not limited to the examples below, which can be changed according to the technical condition evaluation standard and / or other actual conditions, etc.

[0056] Scale, a standard for classifying and evaluating the technical condition of the bridge, is also applicable to the safety risk evaluation of the bridge. In this paper, the scale is the scale of the technical condition evaluation index, common index and disease classification category. Commonly used, a total of 1 to 5, 5 scales, respectively called 1st, 2nd, 3rd, 4th and 5th. It can be understood that the specific scale division may change due to the update of the standard. Each of the technical condition evaluation indexes has its corresponding scale number. Most of the technical condition evaluation indexes have 5 scales, but some of the technical condition evaluation indexes have only 4 scales, and some of the technical condition evaluation indexes have 3 scales, for example, the index of “honeycomb, pitting” has 3 scales, only “1”, “2” and “3” three scales, i.e. 1st, 2nd and 3rd, for example, the index of “spalling, corner drop” has 4 scales, only “1”, “2”, “3” and “4” four scales, i.e. 1st, 2nd, 3rd and 4th. In this paper, since the common index and the disease classification category in this paper are based on the technical condition evaluation index, the explanation and example of the scale of the technical condition evaluation index in this paragraph are also applicable to the common index and the disease classification category.

[0057] Referring to Figure 1 The embodiment provides a bridge superstructure safety risk evaluation method, which comprises the following steps:

[0058] Obtaining disease deduction information of the technical condition of the bridge superstructure, the disease deduction information of the technical condition comprising technical condition evaluation indexes and deduction values of each of the technical condition evaluation indexes at each scale thereof, the deduction values of each of the technical condition evaluation indexes at each scale thereof being numerical values calculated according to the analytic hierarchy process and the group decision method;

[0059] Obtaining bridge dynamic characteristic parameters, bridge superstructure structure parameters, bridge superstructure technical condition score data, bridge superstructure deduction indexes, disease classification categories of the deduction indexes of a sample bridge, and analyzing common indexes in the bridge superstructure deduction indexes of the sample bridge based on the same; calculating deduction values of the disease classification categories of the common indexes at each scale thereof based on the LSHADE algorithm, and calculating a safety risk score of the bridge superstructure of the sample bridge according to the disease classification categories of the common indexes and the deduction values of the disease classification categories of the common indexes at each scale thereof;

[0060] Dividing safety risk levels by using a clustering algorithm according to the safety risk score of the bridge superstructure of the sample bridge;

[0061] Constructing a bridge safety risk evaluation model by using a CatBoost integrated learning method, training the bridge safety risk evaluation model by using a frequency reduction coefficient of the sample bridge, the bridge superstructure structure parameters of the sample bridge, the disease classification categories of the common indexes of the sample bridge, the deduction values of the disease classification categories of the common indexes of the sample bridge, and the safety risk level to which the safety risk score of the bridge superstructure of the sample bridge belongs, and outputting the model as a safety risk level of the bridge superstructure;

[0062] Evaluating the safety risk level of the bridge superstructure by using the trained bridge safety risk evaluation model.

[0063] The bridge superstructure safety risk evaluation method and effects thereof are described in detail below, and refer to Figure 2 , the method comprising:

[0064] Step 1: Obtaining disease deduction information of the technical condition of the bridge superstructure, the disease deduction information of the technical condition comprising technical condition evaluation indexes and deduction values of each of the technical condition evaluation indexes at each scale thereof, the deduction values of each of the technical condition evaluation indexes at each scale thereof being numerical values calculated according to the analytic hierarchy process and the group decision method;

[0065] The obtaining in the step 1 can be direct obtaining or can be obtained by calculation.

[0066] The analytic hierarchy process comprises: giving at least two judgment matrices for the disease deduction information of the original technical condition, elements of the judgment matrix representing the ratio of the influence of two technical condition evaluation indexes on the bridge technical condition; performing consistency detection on each judgment matrix, determining the weight vector of the judgment matrix for the judgment matrix meeting the consistency check, and determining the weight value of each technical condition evaluation index according to the weight vector;

[0067] The group decision method comprises: calculating the reliability coefficient according to the weight vector obtained by the analytic hierarchy process;

[0068] The value calculated according to the analytic hierarchy process and the group decision method is specifically: a value calculated according to the weight value, the reliability coefficient and the disease deduction information of the original technical condition. Specifically, the weight value of the technical condition evaluation index is multiplied by the reliability coefficient to obtain a product, which is called the first product, all first products of the technical condition evaluation index are added to obtain the deduction value coefficient of the technical condition evaluation index, and the deduction value coefficient is multiplied by the deduction value (referred to as the original deduction value) in the disease deduction information of the corresponding original technical condition to obtain the deduction value in the disease deduction information of the technical condition of the bridge superstructure. Generally, the disease deduction information of the technical condition of the bridge superstructure is presented in the form of a table.

[0069] Specifically, the disease deduction information of the technical condition of the bridge superstructure is obtained by using the paper "Highway Bridge Field Detection and Technical Condition Evaluation".

[0070] In step one, the result of the paper "Highway Bridge Field Detection and Technical Condition Evaluation" can be directly obtained as the disease deduction information of the technical condition of the bridge superstructure, or the disease deduction information of the technical condition of the bridge superstructure can be calculated by using the existing deduction value according to the analytic hierarchy process and the group decision method (the method is recorded in the paper "Highway Bridge Field Detection and Technical Condition Evaluation"). Preferably, the existing deduction value is the deduction value recorded in the specification "Highway Bridge Technical Condition Evaluation Standard"; it can be understood that the source of the existing deduction value is not limited.

[0071] The analytic hierarchy process (AHP) mainly includes the following steps: First, for each technical condition assessment indicator in the standard "Standard for Technical Condition Assessment of Highway Bridges," at least two judgment matrices are provided. Specifically, each element of the judgment matrix represents the ratio of the importance of two technical condition assessment indicators to the bridge's technical condition assessment, i.e., the ratio of their impact on the bridge's technical condition. Except for the diagonal elements, all elements represent the ratio of the importance of two different technical condition assessment indicators to the bridge's technical condition assessment; that is, the diagonal equals 1, meaning the ratio of the importance of the same technical condition assessment indicator to the bridge's technical condition assessment is 1. Different judgment matrices are determined based on the opinions of different experts. Then, a consistency check is performed on each judgment matrix to determine if it conforms to consistency. If not, the corresponding judgment matrix is ​​redefined, or the judgment matrix is ​​discarded. Finally, for the judgment matrices that conform to the consistency check, a weight vector is determined. The weight vector is a column vector. Based on the weight vector, the weight value of each technical condition assessment indicator is determined. Specifically, all weight vectors are summarized, and the weight value of each technical condition assessment indicator is obtained accordingly. Here, the number of experts required to achieve consistency in the judgment matrices of the upper load-bearing components, the upper general components, and the supports is [number missing]. .

[0072] Example 1: An expert's judgment matrix for the upper load-bearing component, as shown in the following formula:

[0073]

[0074] The first column, fourth row represents the ratio of the importance of the "concrete cover thickness" indicator to the bridge risk assessment to the importance of the "honeycomb and pitting" indicator to the bridge technical condition assessment.

[0075] The group decision-making method includes: obtaining the results for the upper load-bearing component / upper general component / support respectively based on the analytic hierarchy process. For simplicity, the weight vectors are not differentiated between the weight vectors of upper load-bearing components, upper general components, or supports. The weight vector is defined as follows: ,in, , This represents the total number of technical condition assessment indicators for the superstructure of bridges, including the upper load-bearing components, general components, and supports. Indicates the first Experts The first of the experts (position) of the superstructure of the bridge, specifically the superstructure of the upper load-bearing members / general members / supports. The deduction value (judgment value) for each technical condition assessment indicator. Indicates the first The sum of the standard deviation of each technical condition evaluation index of the upper bearing member / upper general member / support is:

[0076]

[0077] The difference coefficient is obtained by normalization , The difference of the judgment of a certain technical condition evaluation index of the upper bearing member / upper general member / support is represented, that is, the difference coefficient of the weight vector of the The difference coefficient of the weight vector of the The difference coefficient of the weight vector of the

[0078] Then, the similarity is calculated. Assuming that any two weight vectors of the upper bearing member / upper general member / support are and , the included angle between the weight vector and the weight vector is calculated. The cosine value of the included angle represents the similarity of the judgment of the two experts (the expert and the expert), and the similarity of the expert is represented by The similarity represents the sum of the cosine values of the weight vectors of the expert and all other experts minus 1. The is normalized, and the normalized is used as the similarity coefficient. The similarity coefficient of the weight vector of the expert to the upper bearing member / upper general member / support is represented by The similarity coefficient of the expert is three, corresponding to the upper bearing member, the upper general member, and the support.

[0079] Assuming that the reliability coefficient of the expert to the upper bearing member / upper general member / support is , then is:

[0080]

[0081] Obviously, the reliability coefficient of the expert is three, corresponding to the upper bearing member, the upper general member, and the support.

[0082] ​The weight value of the technical condition evaluation index obtained by the analytic hierarchy process is multiplied by all the reliability coefficients obtained by the group decision method, and the product is added again to obtain the deduction value coefficient of the technical condition evaluation index. By multiplying the deduction value coefficient with the corresponding original deduction value, a new deduction value can be obtained. Finally, the disease deduction information of all technical conditions is obtained as an example. The disease deduction information of the superstructure of the bridge finally obtained in this step is shown in Table 1.

[0083] Table 1

[0084]

[0085] Table 1 (continued)

[0086]

[0087] Step two, obtaining the bridge dynamic characteristic parameters, bridge superstructure structure parameters, bridge superstructure technical condition score data, bridge superstructure deduction indicators, disease classification categories of the deduction indicators, and analyzing common indicators in the bridge superstructure deduction indicators based on the disease deduction information of the technical condition of the bridge superstructure. With the maximum correlation of the frequency reduction coefficient of the sample bridge and the bridge superstructure safety risk score as the fitness function, the LSHADE algorithm is used to obtain the deduction value correction coefficient of the disease classification category of the common indicator. According to the disease deduction information and the deduction value correction coefficient, the deduction value of the disease classification category of the common indicator at each scale is obtained. According to the disease classification category of the common indicator and the deduction value of the disease classification category of the common indicator at each scale, the disease deduction information of the safety risk is established. According to the disease deduction information of the safety risk, the safety risk score of the bridge superstructure of the sample bridge is calculated. The disease deduction information of the safety risk includes the disease classification category of the common indicator, the deduction value of the disease classification category of the common indicator at each scale, and in some embodiments, the common indicator.

[0088] It can be understood that the deduction indicators are deduction technical condition indicators; the common indicators are common indicators in the deduction indicators, also known as common deduction indicators. Common indicators refer to technical condition evaluation indicators with an occurrence probability greater than a certain preset probability threshold in the bridge superstructure technical condition scores of all sample bridges. It can be understood that the disease classification category of the common indicator is used as an indicator of the safety risk score rule.

[0089] Here, the disease classification category of the common indicator in one or more deduction indicators may not have at least two specific disease classification categories, and the disease classification category of the common indicator is itself.

[0090] In this embodiment, the penalty value correction coefficient of the disease classification category of all common indicators in step two is modified. In some embodiments, only the penalty value correction is made for the disease classification category of indicators in the common indicators that have at least two specific disease classifications (i.e., the disease classification is not its own).

[0091] In this embodiment, the bridge dynamic characteristic parameters include the theoretical natural frequency of the bridge and the measured natural frequency of the bridge (i.e., the first-order natural frequency of the intact bridge and the first-order natural frequency of the damaged bridge), and the bridge superstructure structural parameters include the cross-sectional form of the bridge superstructure. It can be understood that different disease classification categories have different effects on structural safety. The disease classification category of the deducted indicator is the classification of the deducted indicator, including the cause classification, the trend classification and / or the form classification, such as the classification of the main beam crack, which includes horizontal crack, vertical crack, diagonal crack and network crack. The sample bridge is a bridge with bridge superstructure technical condition score data, which is used as a sample to calculate the penalty value correction coefficient using the LSHADE algorithm.

[0092] The maximum correlation between the frequency reduction coefficient of the sample bridge and the bridge superstructure safety risk score data is used as the fitness function, and the LSHADE algorithm (LSHADE, Linear Population Size Reduction Adaptive Differential Evolution, which is an improved differential evolution (DE) algorithm) is used to iteratively solve the optimal combination of the penalty value correction coefficient. According to the disease deduction information of the bridge superstructure and the optimal penalty value correction coefficient, the modified penalty value of the disease classification category of the common indicators in each scale is obtained. That is, according to the related information of the technical condition score, the penalty value of the disease classification category of the common indicators is modified to obtain the penalty value for safety risk assessment, and the safety risk scoring system is obtained.

[0093] A specific example of step two is given below. As an embodiment, the sample bridge comes from the bridge digital management platform developed by Guangxi Jiaoke Group Co., Ltd. In a certain period, the platform contains periodic detection data of 3202 bridges, mainly including bridge dynamic characteristic parameters, bridge superstructure structural parameters, bridge superstructure technical condition score data, bridge superstructure deducted indicators and disease classification categories of the deducted indicators. The statistical results of some common deducted indicators of the bridge superstructure of large bridges, medium bridges and small bridges are shown in Table 1. Figure 2 Figure 2 ​The first column corresponds to the main beam crack, and the second column's "surface defects" correspond to "stripping, corner chipping," "voids, holes," "honeycomb, pitting," and "support defects" correspond to the support's "defects." According to the statistics of the bridge digital management platform, for common reinforced concrete beam bridges: for the superstructure load-bearing components, D11 "main beam cracks," D2 "stripping, corner chipping," D3 "voids, holes," and D5 "reinforcement corrosion" are the most common deduction indicators, which are common defects; the superstructure general components mainly include diaphragms, wet joints, and hinge joints, with D12 "honeycomb, pitting," D13 "stripping, corner chipping," D14 "voids, holes," D16 "reinforcement corrosion," and D22 "main beam cracks" being the most common deduction indicators; for plate support defects, D23, D24, and D25 are all common deduction indicators.

[0094] Currently, the standards specify the same deduction values ​​for different types of cracks. However, cracks can be divided into structural cracks and non-structural cracks. Further, the classification categories mainly include transverse cracks, longitudinal cracks, diagonal cracks, and network cracks. In reality, different types of cracks do not have entirely the same impact on structural safety. Therefore, it is necessary to redefine the deduction value correction coefficient and the deduction value for safety risk scoring for different types of cracks. To more objectively correct the deduction values ​​for various defect classification categories, a frequency reduction coefficient is introduced (…). This parameter reflects the safety of the bridge, and the frequency reduction factor is defined as follows:

[0095]

[0096] in, , where is the first-order natural frequency of the bridge after damage, and is the frequency obtained from the actual bridge dynamic characteristic test, i.e., the actual natural frequency. is the first-order natural frequency of the intact bridge structure, and is the theoretical natural frequency obtained from numerical simulation.

[0097] To reduce computational load, when calculating the safety risk score of the superstructure of the sample bridges, a subset of sample bridges can be selected, with the requirement that the selected subset of bridges should cover as many technical condition levels as possible.

[0098] In addition, the sample bridges for determining the common index and the sample bridges for determining the deduction value of the disease classification category of the common index are not required to be the same, and the sample bridges for calculating the deduction value are usually part of the sample bridges for determining the common index. When calculating the safety risk score of the superstructure of the sample bridge according to the deduction value, the number of the sample bridges is not limited, and whether the sample bridges are derived from the sample bridges for determining the common index and the deduction value of the disease classification category of the common index is not limited. Preferably, the superstructure technical condition score data of all the sample bridges used in the step of calculating the safety risk score of the superstructure of the sample bridge can cover all the technical condition grades or cover the first four types of technical condition grades. The existing technical condition grades are divided into five grades (the bridge technical condition assessment grades are divided into five types: type one, type two, type three, type four and type five), that is, the technical condition grade division in the standard “Highway Bridge Technical Condition Assessment Standard” (JTGT H21-2011).

[0099] Here, 239 bridges are selected from the 3202 bridges in the bridge digital management platform 3202 for dynamic characteristic testing, including 92 type one bridges, 76 type two bridges, 44 type three bridges and 27 type four bridges. According to Table 1 in step one, the disease classification categories of the common deduction index and the corresponding initial deduction values are obtained, as shown in Table 2. For the super load-bearing member, the super general member and the support, the disease classification categories of the common deduction index are 19 types in total, and the 19 disease classification categories, the codes of the disease classification categories and the deduction values before the LSHADE algorithm correction are shown in Table 2.

[0100] Table 2

[0101]

[0102] The LSHADE differential evolution algorithm is used to construct an optimization model of the deduction value correction coefficient. By setting a fitness function, the correlation coefficient (Pearsen correlation coefficient, i.e., Spearman correlation coefficient) between the deduction value of the disease classification category of the common index and the actual safety risk score of the superstructure of the sample bridge is maximized as the optimization target, and the optimal combination of the deduction value correction coefficient is obtained through multiple runs. The specific steps are as follows:

[0103] (1) For each component of the superstructure of the bridge, a disease classification category deduction value matrix, also known as a bridge disease quantification score matrix, is constructed. The row vector represents the disease classification category, and the column vector corresponds to the deduction value of different scales. According to Table 2, the bridge disease quantification score matrix is constructed by using the matrix expression method, and the superstructure technical condition score data of the sample bridge is used as the row vector, and the disease classification category of the sample bridge is used as the column vector. ​​​The disease classification category deduction value matrix of the upper load-bearing member, the upper general member, and the support is shown below.

[0104]

[0105]

[0106]

[0107] (2) For the initial disease classification category deduction value matrix, a deduction value correction coefficient is given for each disease classification category, which is referred to as the initial disease classification category deduction value matrix. The deduction value correction coefficients of the disease classification categories of the upper load-bearing member are respectively , , , , , 、 , The deduction value correction coefficients of the disease classification categories of the upper general member are respectively , , , , , 、 , The deduction value correction coefficients of the disease classification categories of the support are respectively , , . Since the focus of this part of the study is to adjust the deduction values of different types of cracks, reference is made to the deduction value setting for non-structural cracks in the “Standard for Technical Condition Evaluation of Highway Bridges (Draft for Solicitation of Opinions)” (JTG H21-2017), and the deduction value correction coefficient range of the crack disease classification category is set to [0.5, 1.5]. At the same time, in order to moderately correct the deduction value correction coefficients of other disease classification categories, the deduction value correction coefficient range is limited to [0.8, 1.2]. It can be understood that the setting range is also followed in (3) below.

[0108] (3) For the bridge superstructure, according to the calculation process of the bridge technical condition score in the specification, the safety risk score of the sample bridge is quickly calculated by using the Matlab software, that is, the deduction values of the disease classification categories of the common indicators at their respective scales are used to calculate the score of the safety risk score of the bridge superstructure, and the safety risk score is obtained. According to the initial disease classification category deduction value matrix, the initial safety risk score is obtained.

[0109] The calculation process is as follows: first, the deduction value of the disease classification category is sorted according to the deduction value size, the deduction value of each component of the sample bridge is calculated Then, the component score is calculated according to the component quantity Finally, the score of the entire superstructure is determined according to the weight of the component.

[0110] The main calculation formula is as follows:

[0111]

[0112]

[0113]

[0114]

[0115] Wherein, represents the component number of the bridge superstructure, The value is 1, 2, 3, The value 1 corresponds to the upper load-bearing component, The value 2 corresponds to the upper general component, The value 3 corresponds to the support; represents The number of components on the component, that is, the The component; And Both represent the number of disease classification categories of common indicators on the The component of the The component; Is an intermediate variable from 1 to ; The deduction value of the The component of the The component of the The disease classification category; Represents the deduction value of the The component of the The disease classification category, Represents the converted deduction value of the The component of the The disease classification category; Represents the total number of disease classification categories of common indicators on the The component of the The component; The score of the The component of the The bridge superstructure, The component of the Safety risk score for each component; For the bridge superstructure Safety risk score for this type of component; For the first The average safety risk score of each component in the class of components; For the first The minimum safety risk score of each component in a class of parts; For the first A coefficient that determines the number of components in a given type of component.

[0116] (4) Based on the initial safety risk score of each sample bridge calculated in (3), the control parameters (scaling factor) are dynamically adjusted using the LSHADE algorithm through a historical memory mechanism. F and crossover probability CR The adaptive population reduction strategy effectively balances global exploration and local exploitation capabilities, significantly improving the convergence speed and solution accuracy of complex optimization problems. First, the deduction value correction coefficient vector is set as follows:

[0117]

[0118] Initially, the deduction value correction coefficient vector This is the vector of parameters to be optimized.

[0119] Given the first Features of individual sample bridges The output score of the model equipped with the LSHADE algorithm is :

[0120]

[0121] in, Indicates the first Safety risk score of the superstructure of the sample bridge. The function representing the calculation of the safety risk score of the bridge superstructure. This represents the total number of bridges in the sample. Indicates the sample bridge number, i.e., the first bridge. One sample bridge, features This includes information on the classification of defects and the number of components of the sample bridges. In this embodiment, 239 bridges are used as sample bridges.

[0122] Record of actual measurements The sequence is , Indicates the first A sample bridge After removing invalid bridge samples, calculate the Spearman correlation coefficient:

[0123]

[0124] wherein, denotes the calculated Spearman correlation coefficient, denotes the calculation function of the Spearman correlation coefficient.

[0125] The mean value of the deduction value correction coefficient vector is modified and the standard deviation The regularization is defined as:

[0126]

[0127]

[0128] wherein, denotes the total number of the bridge superstructure disease classification categories, which is 19 in this embodiment, denotes the serial number of the bridge superstructure disease classification category, and denotes the disease classification category of the first bridge superstructure disease classification category, denotes the first bridge superstructure disease classification category.

[0129] The mean value and the standard deviation are combined by weight to form the fitness function to be minimized , and the calculation formula is as follows:

[0130]

[0131] wherein, is the rank correlation penalty, is the regularization penalty, and the coefficient 0.05 is used to balance the regularization strength to prevent the parameter vector from excessive oscillation in the optimization process.

[0132] The fitness function mainly balances the prediction accuracy and parameter stability by minimizing the fitness function and using a global optimization method to optimize the bridge scoring model parameters. The convergence condition mainly aims at the optimization stability of the 19 deduction value correction coefficients, which is controlled by two core parameters: one is the function tolerance (Function Tolerance=1e -4 ), and when the average change of the fitness function in the continuous default 50 generations is less than this threshold, it is determined to be converged; the other is the maximum iteration number (MaxGenerations=10000), which is forcibly limited to prevent infinite loop. In order to improve the accuracy of parameter optimization, a total of 30 model runs are performed, and the running results are as follows: Figure 3The optimal deduction value correction coefficient results of each disease classification type are shown in Table 2 Figure 4 As shown in Table 2, Figure 4 In Table 2, the blue column chart represents the mean value of each deduction value correction coefficient, and the black vertical line represents the standard deviation Because the model is run 30 times, the black vertical line represents the fluctuation range of the deduction value correction coefficient corresponding to the disease.

[0133] (5) According to the final deduction value correction coefficient determined by the LSHADE algorithm, the deduction value of each scale of the disease classification category of the common index is calculated, and according to the disease classification category of the common index and the deduction value of each scale of the disease classification category of the common index, the bridge superstructure safety risk score of the sample bridge is recalculated.

[0134] Step three, according to the relevant data, determine the number of bridge safety risk levels, and set it as the clustering number of K-means algorithm, perform clustering analysis on the recalculated bridge superstructure safety risk score, and divide each safety risk level according to the clustering result, that is, obtain the score interval corresponding to each safety risk level, and complete the risk level assessment of the sample bridge based on the interval where the safety risk score of each sample bridge is located, thereby constructing the safety risk level data set of the bridge superstructure; each safety risk level is a data set, and each data set includes the frequency reduction coefficient of the sample bridge, the bridge superstructure structure parameter of the sample bridge, the disease classification category of the common index of the sample bridge, the deduction value of the disease classification category of the common index of the sample bridge, the modified bridge superstructure score of the sample bridge, and obviously also includes the information of the safety risk level to which the modified bridge superstructure (safety risk score) of the sample bridge belongs.

[0135] Determine the bridge safety risk level interval. After obtaining the modified safety risk score, refer to the relationship between the bridge technical condition assessment classification and the bearing capacity degradation in the “Highway Bridge and Culvert Maintenance Specification” (JTG 5120-2021) to determine the bridge safety risk evaluation level and the corresponding characteristics, and divide the bridge safety risk evaluation level into five levels. The K-means clustering algorithm is used to set the classification number according to the division level to obtain the score interval under different safety risk levels, and the sample bridge is divided into safety levels according to the modified safety risk score, thereby constructing the risk level data set of the bridge superstructure. The formula of K-means clustering is:

[0136]

[0137] Among them, Total squared error is represented by E, The number of clusters is represented by K; The number of clusters is represented by K; The sample point set contained in the cluster is represented by Ck; For the first the center of the cluster (the mean of all samples within the cluster); For the sample data points.

[0138] The sample bridges in this embodiment mainly focus on bridges with technical condition grades one to four. Therefore, as an embodiment but not limitation, the maximum bridge safety risk assessment grade in the sample interval is set to four in this step, so the number of categories is set to "4" when performing sample clustering, and clustering analysis is performed by the K-means clustering algorithm to obtain the scoring interval threshold under different safety risk grades. The safety risk grade classification effect is as shown in Figure 5 For the interval range below the minimum score of the sample, it is directly determined as the fifth level of safety risk, that is, the total number of safety risk grades is five. It is worth noting that when new sample bridges containing technical condition grade five are supplemented, the entire risk assessment interval range will be dynamically modified. The bridge safety risk assessment standard is shown in Table 3.

[0139] Table 3

[0140]

[0141] Step four, a CatBoost (‌Categorical Boosting‌) integrated learning method is used to construct a bridge safety risk evaluation model. The frequency reduction coefficient of the sample bridge, the bridge superstructure structural parameters of the sample bridge, the disease classification category of the common index of the sample bridge, the disease classification category of the common index of the sample bridge, the safety risk grade to which the safety risk score of the bridge superstructure of the sample bridge belongs are used to train the bridge safety risk evaluation model. The output of the model is the safety risk grade of the bridge superstructure;

[0142] In this step, a CatBoost integrated learning method is used to construct an integrated learning model, and the integrated model is used to evaluate the bridge safety risk grade.

[0143] First, the Optuna optimization method (Optuna, a super parameter optimization framework) is used to optimize the super parameters of the CatBoost model, including the depth of the tree, the learning rate, and the regularization coefficient. The frequency reduction coefficient of the sample bridge , the bridge superstructure structural parameters of the sample bridge, the disease classification category of the common index of the sample bridge, the disease classification category of the common index of the sample bridge, and the safety risk grade to which the safety risk score of the bridge superstructure of the sample bridge belongs are used to train the bridge safety risk evaluation model. The bridge section forms in the sample bridges in this embodiment mainly include solid plate girder, hollow plate girder, box girder, and T girder. The disease classification categories of the common indexes are selected from the transverse crack disease of the upper load-bearing member, the transverse crack disease of the upper general member, and the disengagement and out-of-limit disease of the support.

[0144] To improve the applicability of the input parameters, the concept of disease scale proportion is proposed, and the calculation formula is as follows:

[0145]

[0146]

[0147]

[0148] It can be understood that each type of component has a number of components, represents the number of components with the first disease in the upper bearing component, is the total number of components in the upper bearing component, represents the number of components with the first disease in the upper general component, is the total number of components in the upper general component, represents the number of components with the first disease in the support, is the total number of components in the support.

[0149] Set a proportion threshold, select the disease classification category with high proportion (higher than the proportion threshold) as the main disease classification category, that is, the disease classification category that meets the proportion threshold in the common index disease classification category is selected as the main disease classification category. The main disease classification category replaces the common index disease classification category as the input of the integrated learning model. With dynamic characteristics as the main index and apparent disease as the auxiliary index, the safety of the bridge is determined. However, there are many diseases in actual bridges, so it is necessary to screen diseases that have a greater impact on bridge safety to improve the simplicity of bridge evaluation.

[0150] Based on the bridge superstructure risk level data set obtained in step three, the gradient boosting decision tree framework is trained iteratively, and the objective function is optimized to reduce the model residual. The training objective function of the bridge safety risk evaluation model is a multiclass log loss function (Multiclass Log Loss). The loss function of the bridge safety risk evaluation model training is defined as follows:

[0151]

[0152] wherein, represents the total loss function; is the total number of training samples; represents the loss of the th training sample; represents the input feature vector (independent variable) of the i-th training sample; true value of the i-th training sample; true value of the i-th training sample; model predicted value of the i-th training sample; model predicted value of the i-th training sample; total number of decision trees; serial number of the decision tree; i-th base learner, i.e., i-th decision tree; i-th decision tree; i-th decision tree; regularization term of the i-th decision tree. In this step, the difference between the model predicted value and the true value is measured. For a multi-classification problem, suppose there are K safety risk level classifications, then the multi-classification log loss function can be expressed as:

[0153]

[0154]

[0155] wherein, is an indicator function, which takes the value 1 if and only if the true label of the i-th sample is y i, otherwise it takes the value 0; is the class index number, from 0 to K-1; is the raw score of the i-th class output by the model for the i-th sample, which means that the error measurement model directly outputs on each safety risk level classification category; is also the class index number, from 0 to K-1; is the raw score of the i-th class output by the model for the i-th sample. The loss function can also be expressed as: wherein, is the raw score of the i-th class output by the model for the i-th sample, is the probability that the i-th sample is predicted to be the i-th class.

[0156] The regularization term is used to control the model complexity to avoid overfitting, and the calculation formula is as follows:

[0157]

[0158] wherein, is the raw score of the i-th class output by the model for the i-th sample, is the probability that the i-th sample is predicted to be the i-th class.

[0159] The regularization term is used to control the model complexity to avoid overfitting, and the calculation formula is as follows: ​​​​​​​​​​

[0160]

[0161] in, The penalty coefficient representing the number of leaf nodes in the decision tree; Indicates the first The total number of leaf nodes in the decision tree; express L 2. Regularization weights; express L 1. Regularization weights; Indicates the first The first decision tree The output weights of each leaf node.

[0162] Then, the optimized decision tree model is integrated into the CatBoost model. During the integration process, the prediction results of multiple decision tree models are integrated through gradient boosting to obtain the final ensemble learning model. The final ensemble model can accurately evaluate the safety risk level of bridges, making the evaluation results more objective and reducing the potential risks of bridges.

[0163] Furthermore, following step four, the following steps are also included:

[0164] Step 5: Evaluate and validate the ensemble learning model using model evaluation metrics, including but not limited to accuracy, precision, recall, and F1 score. The model that meets the requirements of the corresponding evaluation metrics will be used as the trained bridge safety risk assessment model.

[0165] In this embodiment, the prediction results for the model evaluation metrics are: accuracy 0.946, precision 0.945, recall 0.929, and F1 score 0.937. Additionally, the confusion matrix and ROC curve (Receiver Operating Characteristic curve) of the prediction results are as follows: Figure 6 and Figure 7 As shown, Figure 6 The gradient colors represent different accuracy rates, with darker colors indicating higher accuracy. The percentage represents the proportion of correctly classified samples out of the total number of samples. Figure 7The relationship between the true rate and the false positive rate of the security risk level, wherein "Average" represents the average value of the curves of the security risk levels one, two, three and four in the figure, represents the overall average performance obtained by uniformly calculating all the sample bridge, can reflect the discrimination ability of the model on the overall sample bridge, the red line segment of the security risk level three is covered by the security risk level four, and the AUC (Area Under the Curve) represents the area under the ROC curve, the value range is between 0~1, is a measure of the discrimination ability of the classification model to the sample, the larger represents the better classification ability of the model.

[0166] Step six, using the trained bridge safety risk evaluation model to evaluate the safety risk level of the bridge superstructure. The input is the frequency reduction coefficient of the bridge superstructure to be evaluated, the bridge superstructure structure parameter, the main disease classification category / common index disease classification category, the main disease classification category / common index disease classification category deduction value, and the output is the safety risk level of the bridge superstructure.

[0167] Referring to Figure 8 , a bridge superstructure safety risk evaluation system is provided, comprising:

[0168] An acquisition module is configured to acquire disease deduction information of the technical condition of the bridge superstructure, wherein the disease deduction information of the technical condition comprises technical condition evaluation indexes and deduction values of each of the technical condition evaluation indexes at each scale, and the deduction values of each of the technical condition evaluation indexes at each scale are numerical values calculated according to the analytic hierarchy process and the group decision method;

[0169] A calculation module is configured to acquire bridge dynamic characteristic parameters, bridge superstructure structure parameters, bridge superstructure technical condition score data, bridge superstructure deduction indexes and disease classification categories of the deduction indexes of a sample bridge, analyze common indexes in the bridge superstructure deduction indexes based on the above parameters, calculate deduction values of the disease classification categories of the common indexes at each scale based on the LSHADE algorithm, and calculate a safety risk score of the bridge superstructure of the sample bridge according to the disease classification categories of the common indexes and the deduction values of the disease classification categories of the common indexes at each scale;

[0170] A grade division module is configured to divide the safety risk level by using a clustering algorithm based on the safety risk score of the bridge superstructure of the sample bridge;

[0171] The model construction training module is configured to construct a bridge safety risk evaluation model by using a CatBoost integrated learning method, and is configured to train the bridge safety risk evaluation model by using a frequency reduction coefficient of a sample bridge, a bridge superstructure structure parameter of the sample bridge, a disease classification category of a common index of the sample bridge, a deduction value of the disease classification category of the common index of the sample bridge, and a safety risk level to which a bridge superstructure of the sample bridge belongs.

[0172] The safety risk assessment module is configured to assess the safety risk level of the bridge superstructure by using the trained bridge safety risk evaluation model.

[0173] In the embodiment, the method for calculating the deduction value of the disease classification category of the common index at each scale based on the LSHADE algorithm includes: taking the maximum correlation of the frequency reduction coefficient of the sample bridge and the bridge superstructure safety risk score as a fitness function, using the LSHADE algorithm to obtain a deduction value correction coefficient of the disease classification category of the common index, and using the disease deduction information of the technical condition of the bridge superstructure and the deduction value correction coefficient to obtain the deduction value of the disease classification category of the common index at each scale.

[0174] In the embodiment, the model construction training module includes:

[0175] The first determination unit is configured to determine a main disease classification category in the disease classification category of the common index of the sample bridge.

[0176] The training unit is configured to train the bridge safety risk evaluation model according to the main disease category, the deduction value of the main classification category of the common index of the sample bridge, and according to the frequency reduction coefficient of the sample bridge corresponding to the main disease classification category, the bridge superstructure structure parameter of the sample bridge, and the safety risk level to which the bridge superstructure of the sample bridge belongs.

[0177] In the embodiment, the model construction training module is further configured to evaluate and verify the integrated learning model by using a model evaluation index, and the model evaluation index includes accuracy, precision, recall, and F1 score.

[0178] In the embodiment, the bridge dynamic characteristic parameter includes a first-order natural frequency of the damaged bridge and a first-order natural frequency of the intact bridge; the bridge superstructure structure parameter includes a cross-sectional form of the bridge superstructure; and the frequency reduction coefficient of the sample bridge is a ratio of the first-order natural frequency of the damaged bridge to the first-order natural frequency of the intact bridge.

[0179] In the embodiment, the clustering algorithm is a K-means clustering algorithm, and the number of categories of the K-means clustering algorithm is 4.

[0180] In this embodiment, the analytic hierarchy process (AHP) includes: providing at least two judgment matrices for the original technical condition defect deduction information, where each element of the judgment matrix represents the ratio of the impact of two technical condition assessment indicators on the bridge's technical condition; performing consistency checks on each judgment matrix; determining the weight vector of the judgment matrix for those that meet the consistency check; and determining the weight value of each technical condition assessment indicator based on the weight vector. The group decision method includes: calculating a reliability coefficient based on the weight vector obtained from the AHP; specifically, the numerical values ​​calculated by the AHP and group decision method are numerical values ​​calculated based on the weight values, reliability coefficients, and the original technical condition defect deduction information.

[0181] In this embodiment, the loss function for training the bridge safety risk assessment model is:

[0182]

[0183] in, Represents the total loss function; The total number of training samples; Indicates the first Loss per training sample; Indicates the first The input feature vector of each training sample; For the first The true value of each training sample; For the first The model's predicted values ​​for each training sample; This indicates the total number of decision trees; Indicates the sequence number of the decision tree; Indicates the first A decision tree, For the first The regularization term of a decision tree.

[0184] In specific implementation, the risk assessment system for the safety of bridge superstructure can be implemented by referring to one of the risk assessment methods for the safety of bridge superstructure in any of the above embodiments. The specific implementation steps will not be repeated here.

[0185] An electronic device can be implemented according to the method of this disclosure, the electronic device comprising: a memory; one or more processors; one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing a risk assessment method for the safety of a bridge superstructure according to any of the above embodiments.

[0186] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic components, discrete hardware components, etc. The general-purpose processor can be a microprocessor or can also be any conventional processor.

[0187] The memory can include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage. In addition, the memory can include a combination of any computer-readable storage media, and the memory can be a semiconductor memory chip, a magnetic disk, an optical disk.

[0188] The memory stores executable code, which, when processed by the processor, can cause the processor to perform part or all of the above-mentioned method.

[0189] The present disclosure also provides a computer-readable storage medium comprising instructions which, when executed on a computer, cause the computer to perform each step of the method of risk assessment of bridge superstructure safety according to any one of the embodiments described above.

[0190] The method, system and medium for risk assessment of bridge superstructure safety according to the present disclosure have the following effects:

[0191] According to the disease deduction information based on the technical condition of the analytic hierarchy process and the group decision method, the common indicators and disease classification categories in the sample bridge are obtained, and the LSHADE algorithm is used to accurately calculate the deduction value of the disease classification category of the common indicators in each scale, i.e. the safety risk scoring rule is obtained, the safety risk score of the sample bridge is calculated based on the safety risk scoring rule, the safety risk grade is divided according to the calculation, the bridge safety risk evaluation model capable of outputting the safety risk grade is obtained by training the bridge safety risk evaluation model constructed by the CatBoost integrated learning method. The present disclosure forms a bridge superstructure safety risk assessment method, sets the deduction value corresponding to the safety risk based on the accurate disease deduction information of the technical condition, determines the disease classification category of the common indicators, which takes into account the calculation efficiency and the accuracy of the deduction value, trains the safety risk evaluation model, and performs risk assessment through the model. By introducing the frequency reduction coefficient as a mechanical performance indicator, the evaluation result of the bridge is more objective and accurate. The assessment efficiency of the present disclosure is high and the cost is low, and the accuracy of the safety risk assessment is improved through the design of the whole process. For a large number of small and medium span bridges, the safety risk assessment is fast and low in cost, and is easy to realize large-scale popularization and application.

[0192] In the present disclosure, the scoring rule based on the technical condition is used to obtain the scoring rule of the safety risk analysis, specifically, the scoring rule based on the technical condition of the bridge superstructure, by obtaining the relevant information of the sample bridge and based on the LSHADE algorithm, the accurate scoring rule suitable for the safety risk analysis is obtained, which fills the gap in the prior art. The safety risk score of the sample bridge is calculated by using the designed scoring rule of the safety risk, and the safety risk is classified based on the calculated safety risk score; the bridge safety risk evaluation model is constructed and trained, and the safety risk level can be quickly and accurately evaluated based on the model.

[0193] Specifically, the scoring system for reflecting the safety of the bridge is constructed. The hierarchical analysis method and the group decision method are used to reconstruct the disease deduction table by referring to the disease deduction rule of the bridge technical condition in the existing specification, and the subjective judgment of the experts is quantified in the form of scientific weight. On this basis, the deduction value of the disease classification category of the common index is further corrected by using the LSHADE differential evolution algorithm, and an objective optimization mechanism driven by data is introduced, so that a more reasonable and accurate safety risk scoring standard is formed. This method combining subjective and objective factors can comprehensively reflect the influence of diseases on the safety of the structure and provide more scientific and reasonable deduction suggestions to provide reliable data support for the bridge safety risk assessment.

[0194] Specifically, the bridge technical condition assessment method in the existing specification is difficult to accurately reflect the actual safety risk of the bridge to a certain extent. In order to make up for this deficiency, the present disclosure introduces the concept of bridge safety risk level, divides the safety risk level into five levels according to the relevant standards, and uses the K-means clustering algorithm to cluster the modified scoring data to divide the scoring interval corresponding to each risk level.

[0195] Specifically, the precise and rapid evaluation of the safety of the bridge superstructure is realized, and the bridge superstructure safety risk evaluation model constructed by the present disclosure realizes the precise evaluation of the structure safety mainly based on the dynamic characteristics and supplemented by the apparent diseases. On the one hand, the interference of subjective factors in the traditional artificial judgment method is effectively reduced; on the other hand, by introducing the frequency reduction coefficient as a quantitative index of the structure performance, the damage and degradation degree of the bridge superstructure can be more comprehensively reflected, thereby improving the accuracy and reliability of the bridge superstructure safety risk evaluation.

[0196] The technical features of the above-described embodiments can be combined arbitrarily, and in order to make the description concise, not all possible combinations of the technical features in the above-described embodiments are described, however, as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present disclosure.

[0197] The above-described embodiments are merely illustrative of several embodiments of the present disclosure, which are described in a more specific and detailed manner, but should not be construed as limiting the scope of the patent. It should be noted that for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present disclosure, and these are all within the scope of the present disclosure. Therefore, the scope of protection of the patent of the present disclosure should be subject to the appended claims.

Claims

1. A risk assessment method for the safety of bridge superstructure, characterized in that, Includes the following steps: Obtain the technical condition deduction information of the bridge superstructure. The technical condition deduction information includes technical condition assessment indicators and the deduction value of each technical condition assessment indicator at each scale. The deduction value of each technical condition assessment indicator at each scale is a value calculated by the analytic hierarchy process and the group decision method. The process involves acquiring the bridge dynamic characteristic parameters, bridge superstructure structural parameters, bridge superstructure technical condition score data, deducted indicators for the bridge superstructure, and the defect classification categories of the deducted indicators for the sample bridges. Based on this, common indicators among the deducted indicators for the bridge superstructure are analyzed. Common indicators refer to technical condition assessment indicators that have a probability greater than a certain preset probability threshold in the technical condition scores of the bridge superstructures of all sample bridges. The deduction values ​​of the defect classification categories of the common indicators at each scale are calculated based on the LSHADE algorithm. Finally, the safety risk score of the bridge superstructure of the sample bridges is calculated based on the defect classification categories of the common indicators and their deduction values ​​at each scale. Based on the safety risk scores of the superstructure of the sample bridges, a clustering algorithm was used to classify the safety risk levels. A bridge safety risk assessment model is constructed using the CatBoost ensemble learning method. The model is trained by using the frequency reduction coefficient of the sample bridge, the structural parameters of the superstructure of the sample bridge, the disease classification categories of common indicators of the sample bridge, the deduction values ​​of the disease classification categories of common indicators of the sample bridge, and the safety risk level of the superstructure of the sample bridge. The output of the model is the safety risk level of the bridge superstructure. The safety risk level of the bridge superstructure is assessed using the trained bridge safety risk assessment model.

2. The risk assessment method for the safety of bridge superstructure according to claim 1, characterized in that, The calculation of the deduction values ​​of the common indicators' disease classification categories on each scale based on the LSHADE algorithm includes: using the maximum correlation between the frequency reduction coefficient of the sample bridge and the safety risk score of the bridge superstructure as the fitness function, using the LSHADE algorithm to obtain the correction coefficient of the deduction value of the common indicators' disease classification categories, and based on the disease deduction information of the technical condition of the bridge superstructure and the correction coefficient of the deduction value, obtaining the deduction values ​​of the common indicators' disease classification categories on each scale.

3. The risk assessment method for the safety of bridge superstructure according to claim 1, characterized in that, The specific steps for training the bridge safety risk assessment model using the frequency reduction factor of the sample bridge, the structural parameters of the superstructure of the sample bridge, the defect classification categories of common indicators of the sample bridge, the deduction values ​​of the defect classification categories of common indicators of the sample bridge, and the safety risk level of the superstructure of the sample bridge include: Determine the main disease classification category among the common disease classification categories of the sample bridge. The main disease classification category refers to the disease classification category with a proportion higher than the proportion threshold. Based on the main disease categories, the deduction values ​​of the main classification categories of common indicators of sample bridges, the frequency reduction coefficient of sample bridges corresponding to the main disease classification categories, the structural parameters of the superstructure of sample bridges, and the safety risk level of the superstructure of sample bridges, a bridge safety risk assessment model is trained.

4. The risk assessment method for the safety of bridge superstructure according to claim 3, characterized in that, The specific steps for training the bridge safety risk assessment model also include: evaluating and validating the ensemble learning model using model evaluation metrics, which include accuracy, precision, recall, and F1 score.

5. The risk assessment method for the safety of bridge superstructure according to claim 1, characterized in that, The bridge dynamic characteristic parameters include the first-order natural frequency of the damaged bridge and the first-order natural frequency of the intact bridge; the bridge superstructure structural parameters include the cross-sectional shape of the bridge superstructure; the frequency reduction factor of the sample bridge is the ratio of the first-order natural frequency of the damaged bridge to the first-order natural frequency of the intact bridge.

6. The risk assessment method for the safety of bridge superstructure according to claim 1, characterized in that, The clustering algorithm is the K-means clustering algorithm, and the number of clusters in the K-means clustering algorithm is 4.

7. The risk assessment method for the safety of bridge superstructure according to claim 1, characterized in that, The analytic hierarchy process (AHP) includes: providing at least two judgment matrices for the original technical condition deduction information, where each element of the judgment matrix represents the ratio of the impact of two technical condition assessment indicators on the bridge's technical condition; performing consistency checks on each judgment matrix; determining the weight vector of the judgment matrix for those that meet the consistency check; and determining the weight value of each technical condition assessment indicator based on the weight vector. The group decision method includes: calculating a reliability coefficient based on the weight vector obtained from the AHP; specifically, the numerical values ​​calculated using the AHP and group decision methods are calculated based on the weight values, reliability coefficients, and the original technical condition deduction information.

8. The risk assessment method for the safety of a bridge superstructure according to claim 1, characterized in that, The loss function for training the bridge safety risk assessment model is: ； in, Represents the total loss function; The total number of training samples; Indicates the first Loss per training sample; Indicates the first The input feature vector of each training sample; For the first The true value of each training sample; For the first The model's predicted values ​​for each training sample; This indicates the total number of decision trees; Indicates the sequence number of the decision tree; Indicates the first A decision tree, For the first The regularization term of a decision tree.

9. A risk assessment system for the safety of bridge superstructure, characterized in that, include, The acquisition module is used to acquire the technical condition deduction information of the bridge superstructure. The technical condition deduction information includes technical condition assessment indicators and the deduction value of each technical condition assessment indicator at each scale. The deduction value of each technical condition assessment indicator at each scale is a value calculated by the analytic hierarchy process and the group decision method. The calculation module is used to acquire the bridge dynamic characteristic parameters, bridge superstructure structural parameters, bridge superstructure technical condition score data, bridge superstructure deduction indicators, and the defect classification categories of the deduction indicators of the sample bridges. Based on this, it analyzes the common indicators among the deduction indicators of the bridge superstructure. Common indicators refer to technical condition assessment indicators that have a probability greater than a certain preset probability threshold in the technical condition scores of the bridge superstructures of all sample bridges. It is used to calculate the deduction values ​​of the defect classification categories of the common indicators at each scale based on the LSHADE algorithm. It is also used to calculate the safety risk score of the bridge superstructure of the sample bridges based on the defect classification categories of the common indicators and the deduction values ​​of the defect classification categories of the common indicators at each scale. The risk level classification module is used to classify the safety risk level of the sample bridges based on the safety risk score of the bridge superstructure and using a clustering algorithm. The model building and training module is used to build a bridge safety risk assessment model using the CatBoost ensemble learning method. It is used to train the bridge safety risk assessment model using the frequency reduction coefficient of the sample bridge, the structural parameters of the superstructure of the sample bridge, the disease classification categories of common indicators of the sample bridge, the deduction values ​​of the disease classification categories of common indicators of the sample bridge, and the safety risk level of the superstructure of the sample bridge. The output of the model is the safety risk level of the bridge superstructure. The safety risk assessment module is used to assess the safety risk level of the bridge superstructure using a trained bridge safety risk assessment model.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes instructions that, when executed on a computer, cause the computer to perform a risk assessment method for the safety of a bridge superstructure as described in any one of claims 1 to 8.

Images