Beam bridge cluster structure state rapid diagnosis method based on kalman filter

Abstract

The application discloses a kind of based on Kalman filter beam bridge cluster structure state rapid diagnosis method, the method utilizes beam bridge cluster structure acceleration response monitoring data, constructs cluster single bridge AR model, calculates single bridge AR model coefficient, adopts FCM algorithm, divides beam bridge cluster single bridge class, establishes Kalman filter algorithm model, calculates the correlation function of Kalman filter prediction response residual, constructs the damage diagnosis index of each type single bridge in cluster based on residual correlation function difference, sets the structure damage early warning value of different class single bridge in beam bridge cluster, realizes beam bridge cluster structure state rapid diagnosis.The application can be according to the acceleration response data of beam bridge cluster structure collected, complete cluster single bridge class division, set early warning threshold suitable for the state diagnosis of each type single bridge, realize beam bridge cluster structure state rapid diagnosis, applicable to beam bridge cluster structure state diagnosis and monitoring.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of bridge structure disease monitoring in actual operation, and relates to a beam bridge cluster structure state diagnosis method, in particular to a beam bridge cluster structure state rapid diagnosis method based on a Kalman filter. BACKGROUND

[0002] The application of acceleration response data in bridge engineering covers multiple aspects, from structural health monitoring to damage detection, which provides valuable information for engineers. First of all, acceleration response data plays a crucial role in the daily monitoring and maintenance of bridge structures. By monitoring the acceleration response data of the bridge in real time, engineers can detect the damage of the bridge structure in time and achieve a comprehensive assessment of the structural health. This helps to take timely maintenance measures to prevent minor damage deterioration and improve the service life of the bridge. Secondly, acceleration response data plays a key role in structural damage detection. By collecting the acceleration response data of the bridge structure during the bridge test, analyzing the change of the bridge structure stiffness, engineers can identify whether the structure is damaged and judge whether its performance meets the design requirements. This method has high sensitivity and can detect minor structural changes, providing a reliable basis for accurate damage assessment.

[0003] In the practical application of acceleration response data, the collected acceleration response data will inevitably be affected by various environmental factors. For beam bridge cluster structures, the acceleration response data of each individual bridge is analyzed one by one to achieve cluster structure state diagnosis. This process requires multiple repeated analysis of the influence of environmental factors on the acceleration response monitoring data of each individual bridge structure, greatly prolonging the time required for the entire structural state diagnosis process, severely reducing the efficiency of cluster structure state diagnosis, and making it difficult to achieve rapid warning of beam bridge cluster structure damage. Therefore, in order to ensure the safety of beam bridge cluster structures during operation, by utilizing the environmental similarity within the beam bridge cluster structure, some technical means are used to cluster the bridges within the cluster, set damage warning thresholds suitable for various bridges in the cluster structure, and reduce the number of environmental factor analyses in the diagnosis process, which undoubtedly can achieve rapid and effective diagnosis of beam bridge cluster structure risk. SUMMARY

[0004] In view of the deficiencies of the existing beam bridge cluster structure state diagnosis method, the present application provides a beam bridge cluster structure state rapid diagnosis method based on a Kalman filter. This method can achieve rapid and effective diagnosis of beam bridge cluster structure state, solving the problem of long diagnosis time and low diagnosis accuracy of existing bridge structure state diagnosis methods in beam bridge cluster structure state diagnosis.

[0005] The purpose of the present application is achieved by the following technical solutions:

[0006] A beam bridge cluster structure state rapid diagnosis method based on Kalman filter, comprising the following steps:

[0007] Step one: introducing the beam bridge cluster structure acceleration response monitoring data, constructing the AR model (Auto regressive model) of the single bridge in the cluster, calculating the AR model coefficient of the single bridge, and constructing the AR model coefficient matrix of the beam bridge cluster;

[0008] Step two: according to the AR model coefficient matrix of the beam bridge cluster constructed in step one, the FCM algorithm is used to divide the single bridge categories in the beam bridge cluster;

[0009] Step three: according to the beam bridge cluster structure acceleration response monitoring data introduced in step one, the single bridge structure state equation and observation equation based on the Kalman filter algorithm are established, the correlation function of the acceleration prediction response residual is calculated, and the residual correlation function matrix is constructed;

[0010] Step four: according to the residual correlation function matrix constructed in step three, the damage diagnosis index of each type of single bridge in the cluster based on the difference of the residual correlation function is constructed, and the structure damage early warning value of the single bridge of different categories in the beam bridge cluster is set.

[0011] Compared with the prior art, the present application has the following advantages:

[0012] The present application utilizes the beam bridge cluster structure acceleration response monitoring data, constructs the AR model of the single bridge in the cluster, calculates the AR model coefficient of the single bridge, uses the FCM algorithm to divide the single bridge categories in the beam bridge cluster, establishes the Kalman filter algorithm model, calculates the correlation function of the Kalman filter prediction response residual, constructs the damage diagnosis index of each type of single bridge in the cluster based on the difference of the residual correlation function, sets the structure damage early warning value of the single bridge of different categories in the beam bridge cluster, and realizes the rapid diagnosis of the beam bridge cluster structure state. The present application can complete the classification of the single bridge in the cluster according to the collected beam bridge cluster structure acceleration response data, set the early warning threshold suitable for the state diagnosis of each type of single bridge, realize the rapid diagnosis of the beam bridge cluster structure state, and is suitable for the state diagnosis and monitoring of the beam bridge cluster structure. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 It is the flow chart of the beam bridge cluster structure state rapid diagnosis algorithm based on Kalman filter.

[0014] Figure 2 It is the schematic diagram of the finite element model of the single bridge in the beam bridge cluster.

[0015] Figure 3 It is the schematic diagram of the structure acceleration response and the structure predicted acceleration response.

[0016] Figure 4 This is a schematic diagram showing the early warning values ​​for structural damage of individual bridges within a beam bridge cluster.

[0017] Figure 5 This is a comparison chart of diagnosis completion times.

[0018] Figure 6 This is a comparison chart of diagnostic accuracy rates. Detailed Implementation

[0019] The technical solution of the present invention will be further described below with reference to the accompanying drawings, but it is not limited thereto. Any modifications or equivalent substitutions to the technical solution of the present invention that do not depart from the spirit and scope of the technical solution of the present invention should be covered within the protection scope of the present invention.

[0020] This invention provides a rapid diagnostic method for the structural condition of beam bridge clusters based on Kalman filters. The method constructs an AR model coefficient matrix for the beam bridge cluster based on collected acceleration response monitoring data, classifies individual bridges within the cluster, establishes damage diagnostic indicators for individual bridges using Kalman filter algorithm theory, and sets early warning values ​​for damage to various types of individual bridges within the cluster, thereby achieving rapid diagnostics of the structural condition of the beam bridge cluster. Figure 1 As shown, the specific steps include the following:

[0021] Step 1: Introduce acceleration response monitoring data of the beam bridge cluster structure, construct AR models (Auto regressive models) of individual bridges within the cluster, calculate the AR model coefficients of individual bridges, and construct the AR model coefficient matrix of the beam bridge cluster. The specific steps are as follows:

[0022] Step 11: Based on the acceleration response monitoring data of the beam bridge cluster, construct AR models of individual bridges within the cluster and formulate the AIC criterion expression:

[0023]

[0024] In the formula, This represents the acceleration prediction value of the AR model at the t-th measurement point; For the first AR model coefficients of order, p is the order of the AR model; For the first Acceleration monitoring values ​​at the measuring points

[0025]

[0026] In the formula, ρ k Let be the k-th transition coefficient, where k = 1, 2, 3, ..., p.

[0027]

[0028] where R k is the autocovariance function; R k =E[a t a t-k ]; a t is the acceleration monitoring value of the single bridge structure in the cluster of beam bridges at the tth monitoring point; E[·] is the expectation function.

[0029]

[0030] where AIC is the AIC criterion function; N is the length of the acceleration response monitoring data of the single bridge structure in the cluster of beam bridges for establishing the AR model; is the variance of the AR model residual e t ,

[0031] Step 1-2: According to the AIC criterion expression constructed in step 1-1, the AR model order of the single bridge is calculated:

[0032]

[0033] where AIC min is the minimum value of the criterion function AIC in the interval p ∈ [1, 1000], and the model order p min is taken at this time.

[0034] Step 1-3: According to the AR model order of the single bridge calculated in step 1-2, the cluster AR model coefficient matrix is constructed:

[0035] Φ = [Φ1 Φ2 Φ3 … Φ m ] (6)

[0036] where Φ is the AR model coefficient matrix of the cluster of beam bridges; Φ j is the AR model coefficient vector of the jth single bridge, j = 1, 2, 3, …, m; and m is the number of single bridges in the cluster.

[0037]

[0038] where is the i th order AR model coefficient of the jth single bridge, i = 1, 2, 3, …, p min .

[0039] Step 2: According to the AR model coefficient matrix of the cluster of beam bridges constructed in step 1, the FCM algorithm is used to divide the single bridge categories in the cluster of beam bridges. The specific steps are as follows:

[0040] Step two one: set the number of single bridge categories in the bridge cluster, and establish the initial membership matrix:

[0041]

[0042] In the formula, u qj is the membership of the jth single bridge to the qth bridge category; c is the number of bridge categories.

[0043]

[0044] In the formula, U is the initial membership matrix.

[0045] Step two two: according to the initial membership matrix established in step two one, calculate the clustering center vector of each type of single bridge in the bridge cluster:

[0046]

[0047] In the formula, C q is the clustering center vector of the qth single bridge, q = 1, 2, 3, …, c; θ is the clustering parameter, and θ = 2.

[0048] Step two three: according to the clustering center vector of each type of single bridge in the bridge cluster obtained in step two two, reconstruct the membership matrix:

[0049]

[0050] In the formula, C ι is the clustering center vector of the ith single bridge, i = 1, 2, 3, …, c.

[0051] Step two four: according to the reconstructed membership matrix in step two three, calculate the FCM algorithm objective function:

[0052]

[0053] In the formula, u ιj is the membership of the jth single bridge to the ith bridge category.

[0054] Step two five: repeat steps two one to two four until the FCM algorithm objective function meets the judgment condition, and complete the classification of single bridges in the bridge cluster:

[0055] |J (n) -J (n+1) |<ε (13)

[0056] In the formula, J (n) is the FCM algorithm objective function calculated for the nth time; J (n+1)The FCM algorithm objective function calculated in the n+1th time is repeated; ε is a judgment condition, and ε=0.01 is taken.

[0057] Step three: according to the acceleration response monitoring data of the beam bridge cluster structure introduced in step one, the state equation and observation equation of the single bridge structure based on the Kalman filter algorithm are established, the correlation function of the acceleration prediction response residual is calculated, and the residual correlation function matrix is constructed. The specific steps are as follows:

[0058] Step three one: according to the acceleration response monitoring data of the beam bridge cluster structure introduced in step one, the state equation and observation equation of the single bridge structure based on the Kalman filter algorithm are established:

[0059]

[0060] In the formula, is the priori estimation value of the acceleration response of the single bridge in the beam bridge cluster at s+1 moment; is the priori estimation value of the acceleration response of the single bridge in the beam bridge cluster at s moment; is the correction value of the acceleration response of the single bridge in the beam bridge cluster at s+1 moment; y s is the monitoring value of the acceleration response of the single bridge in the beam bridge cluster at s moment; A is a state transition matrix; is the Kalman gain; C is the state output matrix of the single bridge in the beam bridge cluster, and I is a unit matrix.

[0061] Step three two: according to the state equation and observation equation of the single bridge structure based on the Kalman filter algorithm established in step three one, the acceleration prediction response residual correlation function of the single bridge structure in the beam bridge cluster is constructed:

[0062]

[0063] In the formula, h s is the acceleration prediction response residual correlation function of the single bridge structure in the beam bridge cluster at s moment, s=1, 2, 3, …, v, v is the number of acceleration response monitoring moments of the single bridge structure in the beam bridge cluster; Δt is the sampling time interval of the acceleration response monitoring data of the single bridge structure in the beam bridge cluster; D d (w) = DTFT(D(z)), and DTFT(·) is a discrete time state Fourier transform.

[0064] D(z) = E(e(z)e(z) * ) (16)

[0065] In the formula, E(·) is an expectation; e(z) * is the conjugate transpose of e(z).

[0066]

[0067] In the formula, z(·) is the z-transform; This is the correction value for the acceleration response of a single bridge within the beam bridge cluster at time s; y represents the prior estimate of the acceleration response of an individual bridge within the beam bridge cluster at time s-1; s-1 The value represents the acceleration response of a single bridge within the beam bridge cluster at time s-1.

[0068] Step 33: Based on the residual correlation function of the acceleration prediction response of individual bridge structures within the beam bridge cluster constructed in Step 32, construct the residual correlation function matrix:

[0069] H = [h1 h2 … h] v (18)

[0070] In the formula: H is the residual correlation function matrix.

[0071] Step 4: Based on the residual correlation function matrix constructed in Step 3, construct damage diagnosis indicators for various types of individual bridges within the cluster based on the difference in residual correlation functions, and set early warning values ​​for structural damage of different categories of individual bridges within the beam bridge cluster. The specific steps are as follows:

[0072] Step 41: Based on the acceleration response monitoring data of the beam bridge cluster introduced in Step 1, the acceleration response monitoring data of individual bridges within the cluster are divided into three categories according to the collection time: health status, reference status, and status to be diagnosed.

[0073] Θ+Υ+Ψ=v (19)

[0074] In the formula, Θ represents the number of monitoring times for the acceleration response monitoring data of a single bridge under healthy conditions; Υ represents the number of monitoring times for the acceleration response monitoring data of a single bridge under reference conditions; and Ψ represents the number of monitoring times for the acceleration response monitoring data of a single bridge under the condition to be diagnosed.

[0075] Step 42: Based on the residual correlation function matrix constructed in Step 3 and the classification of acceleration response monitoring data categories for individual bridges within the cluster in Step 41, construct the Hankle matrix for each individual bridge within the cluster:

[0076]

[0077] In the formula, For the cth node in the cluster i In the j-th class of single-unit bridges i The g-th Hankle matrix of a single bridge under reference condition, where c i =12,3,…,c; g=12,3,…,β, Floor(·) is a function that rounds down; sub(·) is a function that calculates the angle between two vectors; For the cth node in the cluster i In the j-th class of single-unit bridges i The difference in the correlation function of the acceleration response residuals of individual bridges at time o2 under reference conditions, where o2 = 12, 3, ..., Y.

[0078]

[0079] In the formula, For the cth node in the cluster i In the j-th class of single-unit bridges i The correlation function of the acceleration response residual of a single bridge at time o2 under reference conditions; For the cth node in the cluster i In the j-th class of single-unit bridges i Mean value of the acceleration response residual correlation function of a single bridge under healthy conditions.

[0080]

[0081] In the formula, For the cth node in the cluster i In the j-th class of single-unit bridges i The correlation function of the acceleration response residuals of a single bridge at time o1 under the condition of good health, where o1 = 1, 2, 3, ..., Θ.

[0082] Step 43: Based on the Hankle matrices of each individual bridge within the cluster constructed in Step 42, construct damage diagnosis indices for various types of individual bridges within the cluster based on the difference of residual correlation functions:

[0083]

[0084] In the formula, For the cth node in the cluster i In the j-th class of single-unit bridges i The g-th group of damage diagnostic indicators for a single bridge under reference conditions; For the cth node in the cluster i In the j-th class of single-unit bridges i The g-th damage diagnosis matrix for a single bridge under reference conditions; S o for The right null space, and satisfies norm(·) is a function for calculating the norm of a matrix.

[0085]

[0086] In the formula: For the cth node in the clusteri the jth i monomer bridge in the reference state.

[0087]

[0088] wherein: the cth i monomer bridge in the cluster based on the residual correlation function difference; the cth i monomer bridge in the cluster based on the residual correlation function difference; i the jth i monomer bridge in the reference state, wherein j i = 1, 2, 3, …, a, a is the number of the cth

[0089] Step four four: according to the damage diagnosis index of each type of monomer bridge in the cluster based on the residual correlation function difference constructed in step four three, set the structure damage early warning value of different types of monomer bridges in the bridge cluster:

[0090]

[0091] wherein, is the structure damage early warning value of the cth 90％ type of monomer bridge in the bridge cluster; Rank (·) is a sorting function that sorts the elements in the vector from large to small and removes the value at the 90% position.

[0092] The existing bridge cluster structure state diagnosis method often follows such a train of thought: constructing the finite element model of each monomer bridge in the cluster, analyzing the environmental factors affecting the acceleration response data of the monomer bridge structure, counting the environmental characteristics of each monomer bridge in the cluster, designing the acceleration response monitoring data denoising algorithm of each monomer bridge, and setting the structure damage threshold according to the early collected acceleration response monitoring data of each monomer bridge. However, due to the existence of a large number of bridge structures in the bridge cluster, setting the damage diagnosis threshold by analyzing the environmental characteristics of each bridge structure one by one will make the process of cluster structure damage diagnosis very slow, seriously affecting the operation and maintenance efficiency of the entire cluster structure, and increasing the operation and maintenance cost of the cluster structure. Therefore, the present application starts from the perspective of information clustering, divides the bridge categories in the cluster by using the environmental similarity existing in the cluster structure, and sets the damage diagnosis index based on the residual correlation function difference through the Kalman filter algorithm theory, and then realizes the rapid diagnosis of the bridge cluster structure state, and guarantees the safe operation and maintenance of the bridge cluster structure during the operation period.

[0093] The following test is used to verify the effect of the present application:

[0094] The test utilizes a finite element analysis method to establish a main beam finite element model of a beam bridge cluster structure, and to verify the effectiveness of the method. The acceleration response monitoring data is simulated by extracting the acceleration response data of the finite element model.

[0095] The test is as follows:

[0096] Step one: establish a main beam finite element model of a beam bridge cluster structure, simulate the damage of the main beam structure by changing the element stiffness and cross-sectional size of the finite element model at some positions, and simultaneously apply random traffic loads to the model to obtain acceleration response monitoring data under the load. The finite element model of a single bridge in the beam bridge cluster is as shown in Figure 2 .

[0097] Step two: based on the obtained acceleration response monitoring data, construct an AR model of a single bridge in the cluster, calculate the AR model coefficients of the single bridge, and construct an AR model coefficient matrix of the beam bridge cluster.

[0098] Step three: based on the constructed AR model coefficient matrix of the beam bridge cluster, use the FCM algorithm to classify the single bridges in the beam bridge cluster.

[0099] Step four: based on the obtained acceleration response monitoring data, establish the state equation and observation equation of a single bridge structure based on the Kalman filter algorithm, calculate the correlation function of the prediction response residual of the Kalman filter, and construct a residual correlation function matrix. The structure acceleration response and the predicted acceleration response are as shown in Figure 3 .

[0100] Step five: based on the constructed residual correlation function matrix, construct a damage diagnosis index of each type of single bridge in the cluster based on the residual correlation function difference, and set the single bridge structure damage warning value of different types of single bridges in the beam bridge cluster. As shown in Figure 4 , when damage occurs, the damage diagnosis index will rapidly increase and exceed the set single bridge structure damage warning value, achieving rapid diagnosis of the state of the beam bridge cluster structure. The diagnosis results of the present application and the diagnosis results of the existing beam bridge cluster structure state diagnosis method are compared, as shown in Figure 5 and Figure 6 , when the number of bridges in the cluster is small, the time required for diagnosis and the diagnosis effect of the present application are slightly better than those of the existing beam bridge cluster structure state diagnosis method, but as the number of bridges in the cluster increases, the time required for diagnosis and the diagnosis effect of the present application are significantly better than those of the existing beam bridge cluster structure state diagnosis method, verifying the effectiveness of the method of the present application.

[0101] The present application sets up damage diagnosis indexes based on residual correlation function difference through Kalman filter algorithm theory from the perspective of information clustering, divides the bridge categories in the cluster according to the existing environmental similarity in the cluster structure, and further realizes the rapid diagnosis of the state of the bridge cluster structure, thereby guaranteeing the safe operation and maintenance of the bridge cluster structure during the operation period.

Claims

1. A Kalman filter based method for rapid diagnosis of the state of a cluster of beam bridges, characterized in that The method comprises the following steps: Step one: introducing the acceleration response monitoring data of the girder bridge cluster structure, constructing the AR model of the single bridge in the cluster, calculating the AR model coefficient of the single bridge, and constructing the AR model coefficient matrix of the girder bridge cluster; Step two: according to the AR model coefficient matrix of the girder bridge cluster constructed in step one, the FCM algorithm is used to divide the single bridge categories in the girder bridge cluster, and the specific steps are as follows: Step two one: setting the number of single bridge categories in the girder bridge cluster, and establishing the initial membership matrix: wherein is the membership of the jth monomer bridge to the qth bridge class; is the number of bridge classes. In the formula, is the initialization of the membership matrix; Step two two: according to the initial membership matrix established in step two one, the clustering center vector of each single bridge category in the girder bridge cluster is calculated: wherein is the cluster center vector for the qth class of monomer bridges, ; is the clustering parameter; is the AR model coefficient vector for the jth monomer bridge, ; is the number of monomer bridges within the cluster. Step two three: according to the clustering center vector of each single bridge category in the girder bridge cluster obtained in step two two, the membership matrix is reconstructed: In the formula, is the first cluster center vector of the cluster of monomer bridges, ; Step two four: according to the reconstructed membership matrix in step two three, the FCM algorithm objective function is calculated: In the formula, is the membership of the jth monomer bridge to the th bridge class. Step two five: repeating steps two one to two four until the FCM algorithm objective function meets the judgment condition, and the single bridge category division in the girder bridge cluster is completed: In the formula, is the FCM algorithm objective function calculated for the nth time; is the FCM algorithm objective function calculated for the n+1th time; is a judgment condition; Step three: according to the acceleration response monitoring data of the girder bridge cluster structure introduced in step one, the state equation and observation equation of the single bridge structure based on the Kalman filter algorithm are established, the correlation function of the acceleration prediction response residual is calculated, and the residual correlation function matrix is constructed; Step four: according to the residual correlation function matrix constructed in step three, the damage diagnosis index of each single bridge category in the cluster based on the residual correlation function difference is constructed, and the structure damage early warning value of the single bridge of different categories in the girder bridge cluster is set.

2. The Kalman filter based rapid diagnosis method for the state of a cluster structure of beam bridges according to claim 1, characterized in that The specific steps of step one are as follows: Step one one: according to the acceleration response monitoring data of the girder bridge cluster structure, the AR model of the single bridge in the cluster is constructed, and the AIC criterion expression is constructed: In the formula, is the AIC criterion function; is the length of the acceleration response monitoring data of the single bridge structure in the beam bridge cluster for establishing the AR model; is the residual of the AR model is the variance of the residual of the AR model, ; is the acceleration monitoring value of the acceleration response monitoring data of the single bridge structure in the beam bridge cluster for establishing the AR model at the tth measurement point; is the acceleration prediction value of the AR model at the tth measurement point; is the order of the AR model; Step one two: according to the AIC criterion expression constructed in step one one, the AR model order of the single bridge is calculated: In the formula, is the interval The minimum value of the inner criterion function The model order at this time is taken The model order of the single-span bridge AR model; Step one three: according to the AR model order of the single bridge calculated in step one two, the cluster AR model coefficient matrix is constructed: In the formula, is the AR model coefficient matrix of the cluster of beam bridges; is the AR model coefficient vector of the jth single bridge, ; is the number of single bridges in the cluster.

3. The Kalman filter based rapid diagnosis method for the state of a cluster structure of beam bridges according to claim 2, characterized in that In each of the steps, The calculation formula is: In the formula, AR model coefficients of the 1st order, , AR model order; AR model coefficients of the 1st acceleration monitoring value of the measuring point, ; In the formula, is the kth transition coefficient, .

4. The Kalman filter based rapid diagnosis method for the state of a cluster structure of beam bridges according to claim 3, characterized in that The The calculation formula is: wherein is a self-covariance function; ; is an expectation function.

5. The Kalman filter based rapid diagnosis method for the state of a cluster structure of beam bridges according to claim 2, characterized in that In step one three, The calculation formula is: In the formula, is the jth monomer bridge i-th order AR model coefficient, .

6. The Kalman filter based rapid diagnosis method for the state of a cluster structure of beam bridges according to claim 2, characterized in that The specific steps of step three are as follows: Step three one: according to the acceleration response monitoring data of the girder bridge cluster structure introduced in step one, the state equation and observation equation of the single bridge structure based on the Kalman filter algorithm are established: In the formula, is the prior estimate value of the acceleration response of the single bridge in the group of beam bridges at time s+1; is the prior estimate value of the acceleration response of the single bridge in the group of beam bridges at time s; is the correction value of the acceleration response of the single bridge in the group of beam bridges at time s+1; is the monitoring value of the acceleration response of the single bridge in the group of beam bridges at time s; is the state transition matrix; is the Kalman gain; , C is the state output matrix of the single bridge in the group of beam bridges, and I is the unit matrix. Step three two: according to the state equation and observation equation of the single bridge structure based on the Kalman filter algorithm established in step three one, the residual correlation function of the acceleration prediction response of the single bridge in the girder bridge cluster is constructed: In the formula, is the residual correlation function of the acceleration response prediction of the single bridge structure in the beam bridge cluster at the s th moment, v is the number of monitoring moments of the acceleration response of the single bridge structure in the beam bridge cluster; is the sampling time interval of the acceleration response monitoring data of the single bridge structure in the beam bridge cluster; , is the discrete-time state Fourier transform; wherein is desired; is the conjugate transpose; In the formula, is the z transform; is the modified value of the acceleration response of the single bridge in the beam bridge cluster at time s; is the prior estimate value of the acceleration response of the single bridge in the beam bridge cluster at time s-1; is the monitoring value of the acceleration response of the single bridge in the beam bridge cluster at time s-1; Step three three: according to the residual correlation function of the acceleration prediction response of the single bridge in the girder bridge cluster constructed in step three two, the residual correlation function matrix is constructed: In the formula, H is the residual correlation function matrix.

7. The Kalman filter based rapid diagnosis method for the state of a cluster structure of beam bridges according to claim 6, characterized in that The specific steps of step four are as follows: Step four one: according to the acceleration response monitoring data of the girder bridge cluster structure introduced in step one, the acceleration response monitoring data of the single bridge in the cluster is divided into three categories of health status, reference status and to-be-diagnosed status according to the collection time sequence: In the formula, is the number of monitoring time points of the acceleration response monitoring data of the monomer bridge under the healthy condition; is the number of monitoring time points of the acceleration response monitoring data of the monomer bridge under the reference condition; is the number of monitoring time points of the acceleration response monitoring data of the monomer bridge under the condition to be diagnosed; Step four two: according to the residual correlation function matrix constructed in step three and the single bridge acceleration response monitoring data categories divided in step four one, the Hankle matrix of each single bridge in the cluster is constructed: wherein is the gth Hankle matrix for the th monolithic bridge in the cluster of monolithic bridges under the reference condition; is the gth Hankle matrix for the th monolithic bridge in the cluster of monolithic bridges under the reference condition; , , is a floor function; is a function to calculate the angle between two vectors; is the gth Hankle matrix for the th monolithic bridge in the cluster of monolithic bridges under the reference condition; is the gth Hankle matrix for the th monolithic bridge in the cluster of monolithic bridges under the reference condition; is the gth Hankle matrix for the Step four three: according to the Hankle matrix of each single bridge in the cluster constructed in step four two, construct the damage diagnosis index of each single bridge in the cluster based on the difference of residual correlation function: In the formula, is the gth damage diagnosis index of the kth monomer bridge in the cluster under the reference condition; is the gth damage diagnosis index of the kth monomer bridge in the cluster under the reference condition; is the gth damage diagnosis index of the kth monomer bridge in the cluster under the reference condition; is the gth damage diagnosis index of the kth monomer bridge in the cluster under the reference condition; is the gth damage diagnosis index of the kth monomer bridge in the cluster under the reference condition; is the gth damage diagnosis index of the kth monomer bridge in the cluster under the reference condition; is the right null space of , and satisfies , is a function for calculating the matrix norm; In the formula: is the damage diagnosis index vector of the i-th single bridge in the reference condition in the cluster; is the damage diagnosis index vector of the i-th single bridge in the reference condition in the cluster; is the damage diagnosis index vector of the i-th single bridge in the reference condition in the cluster; In the formula: is the damage diagnosis index of the i-th single bridge in the cluster under the reference condition, wherein is the damage diagnosis index of the i-th single bridge in the cluster based on the residual correlation function difference; is the damage diagnosis index of the i-th single bridge in the cluster based on the residual correlation function difference; is the damage diagnosis index of the i-th single bridge in the cluster under the reference condition, wherein is the damage diagnosis index of the i-th single bridge in the cluster under the reference condition, wherein , is the damage diagnosis index of the i-th single bridge in the cluster under the reference condition, wherein is the number of the i-th single bridge in the cluster. Step four four: according to the damage diagnosis index of each single bridge in the cluster based on the difference of residual correlation function constructed in step four three, set the structure damage early warning value of different types of single bridges in the bridge cluster: In the formula, is the structural damage early warning value of the first type single bridge in the beam bridge cluster; is the sorting function of the elements in the vector from large to small and the value at the 90% position.

8. The Kalman filter based rapid diagnosis method for the state of a cluster structure of beam bridges according to claim 7, characterized in that In step four two, The calculation formula is: wherein is the acceleration response residual correlation function for the kth monolithic bridge in the cluster at the ith time instant under the reference condition; is the acceleration response residual correlation function for the kth monolithic bridge in the cluster at the ith time instant under the reference condition; is the acceleration response residual correlation function for the kth monolithic bridge in the cluster at the ith time instant under the reference condition; is the acceleration response residual correlation function for the kth monolithic bridge in the cluster at the ith time instant under the reference condition; is the acceleration response residual correlation function for the kth monolithic bridge in the cluster at the ith time instant under the reference condition; is the acceleration response residual correlation function for the kth monolithic bridge in the cluster at the ith time instant under the reference condition; is the acceleration response residual correlation function for the kth monolithic bridge in the cluster at the ith time instant 9. The Kalman filter based rapid diagnosis method for the state of a cluster structure of beam bridges according to claim 8, characterized in that The The calculation formula is: In the formula, For the first in the cluster In the single-unit bridge category, the first The first individual bridge under healthy conditions The correlation function of the acceleration response residual at time 1. .

Images