Bridge deck deformation prediction method based on double models

Abstract

The application discloses a kind of bridge deck deformation prediction methods based on double model, S1: obtaining stable change data set and final influence change data set;S2: construct stable prediction model, predict future stable change under the beam displacement deformation and bridge tower skew angle deformation;S3: construct influence prediction model, predict future external influence under the beam displacement deformation and bridge tower skew angle deformation;S4: the beam displacement deformation under future stable change and bridge tower skew angle deformation and the beam displacement deformation under external influence and bridge tower skew angle deformation are added respectively, to obtain future low tower cable-stayed bridge deck settlement deformation data.The application retains the corresponding change characteristics of the historical measurement data set of bridge deck deformation, while fully considering the influence of the change of influencing factors on the bridge deck settlement deformation, greatly improving the applicability of the bridge deck settlement deformation prediction under complex conditions, providing a fast, accurate and scientific prediction method for bridge deck settlement deformation.

Description

Technical Field

[0001] This invention relates to the field of bridge risk control technology, and in particular to a method for predicting bridge deck deformation based on a dual-model approach. Background Technology

[0002] The trend of bridge deck subsidence is the result of multiple factors. Currently, most bridge deck settlement predictions are based on previously monitored bridge deck deformation data and use a single prediction model to predict bridge deck deformation. This ignores the relevant trends of historical bridge deck deformation displacement and the results caused by external influencing factors. Furthermore, existing models have poor generalization ability and are difficult to effectively predict data with small changes, resulting in large prediction errors. Summary of the Invention

[0003] This invention provides a bridge deck deformation prediction method based on a dual-model approach to overcome the technical problems of existing prediction models ignoring the relevant trends of historical bridge deck deformation displacement and the effects of external influencing factors, as well as the poor generalization ability of existing models, which makes it difficult to effectively predict data with small changes, resulting in large prediction errors.

[0004] To achieve the above objectives, the technical solution of the present invention is as follows:

[0005] A bridge deck deformation prediction method based on a dual-model approach, characterized by the following specific steps:

[0006] S1: Obtain the historical monitoring dataset of the low-tower cable-stayed bridge, and process the historical monitoring dataset to obtain the stable change dataset and the final impact change dataset;

[0007] S2: Construct a stable prediction model and use the stable prediction model and stable change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future stable changes.

[0008] S3: Construct an impact prediction model, and use the impact prediction model and the final impact change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future external influences.

[0009] S4: Add the beam displacement deformation and bridge tower tilt angle deformation under the future stable changes to the beam displacement deformation and bridge tower tilt angle deformation under external influences, respectively, to obtain the future settlement deformation data of the low-tower cable-stayed bridge deck.

[0010] Further, in S1, the process of acquiring the historical monitoring dataset of the low-tower cable-stayed bridge and processing the historical monitoring dataset to obtain the stable change dataset and the final impact change dataset is as follows:

[0011] S11: Establish a historical monitoring dataset that includes datasets of external influencing factors and datasets of bridge deck settlement and deformation;

[0012] S12: Based on whether the bridge deck settlement deformation is affected by external factors, the bridge deck settlement deformation dataset is divided into a stable change dataset and an impact change dataset, and the impact change dataset is combined with the external impact factor dataset to obtain the final impact change dataset.

[0013] Furthermore, in S11, the acquired historical monitoring dataset is represented as V = (H, J);

[0014] Where H = (h1, h2...h O ) T H represents the dataset of external influencing factors within O months, h1, h2...h O These represent the monitoring data of external influencing factors for each month;

[0015] in, The values ​​represent the vehicle operating load C, bridge self-weight load D, regional seasonal temperature E, near-bridge air humidity F, and wind disturbance G in month O, respectively, in units of t, t, ℃, %RH, and N.

[0016] Where J = (j1, j2, ... j O J represents the bridge deck settlement and deformation dataset corresponding to H within O months, j1, j2...j O These represent the bridge deck settlement and deformation monitoring data for each month;

[0017] in, The values ​​represent the deformation of beam displacement A and the deformation of bridge tower tilt angle B in month O, respectively, in mm and °.

[0018] In S12, depending on whether the bridge deck settlement deformation is affected by external factors, the bridge settlement deformation monitoring data J = (j1, j2...j O The dataset is divided into two categories: a stable dataset unaffected by external factors and a dataset affected by external factors. The stable dataset is... Combining the aforementioned impact change dataset with the aforementioned external influencing factor dataset yields the final impact change dataset.

[0019] Furthermore, in S2, the process of constructing a stable prediction model and using the stable prediction model and the stable change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future stable changes is as follows:

[0020] S21: Construct the minimum optimization function, expressed as:

[0021]

[0022] In the formula, λ is the optimal solution output by the optimization function matrix, μ is the normal vector coefficient of the height plane, and θ is the standard normalization coefficient. To minimize the error of the optimization function, To project nonlinear functions into a high-dimensional space, It is a constant;

[0023] S22: Construct a Lagrange polynomial based on equation (1) and input it into a stable changing dataset. get:

[0024]

[0025] in,

[0026]

[0027] In the formula, Θ is a Lagrange polynomial; These are the system state coefficients; for The sum of beam displacement deformation and bridge tower tilt angle deformation over a month;

[0028] S23: For μ, respectively Taking the partial derivative, we obtain the following results;

[0029]

[0030] S24: Substitute the result obtained in S23 into the solution of the simplified matrix to get:

[0031]

[0032] in,

[0033]

[0034] In the formula, E is the identity matrix; Z is the kernel function;

[0035] S25: Obtained from S24 and A stability prediction function is constructed, and the beam displacement deformation and bridge tower tilt angle deformation under future stable changes are calculated based on the stability prediction function. The formula is as follows:

[0036]

[0037] Furthermore, in S3, the process of constructing an impact prediction model and using the impact prediction model and the final impact change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future external influences is as follows:

[0038] S31: Use the data in the final impact change dataset as learning samples, and divide the learning samples according to the continuous monitoring time. One operation interval;

[0039] S32: Construct the selection sequence formula and substitute it with the abstract relation matrix of the final impact change dataset of the ι-th operation interval. The final impact change of the dataset ω with respect to the (i+1)th operation interval ι+1 , Represented as:

[0040]

[0041] Where, α ι+1 The data is selected using the formula for the sequence, where ψ is the sigmoid function and Λ is the number of elements in the sequence. α I α They are respectively the coefficients of the linear relationship in the formula for the sequence. To select the linear relationship error of the sequence;

[0042] in,

[0043]

[0044] In the formula, CXY is an abstract function, and ∧ is the relation matrix of the final dataset affecting changes. To account for the final error affecting the changed dataset, ω ι This represents the final impact change dataset for the ι-th operation interval;

[0045] S33: Construct the imported sequence formula and input the abstract relation matrix of the final impact change dataset for the ι-th operation interval. The final impact change of the dataset ω with respect to the (i+1)th operation interval ι+1 , is represented as:

[0046]

[0047] In the formula, β ι+1 , χ ι+1 These represent the left and right import formulas to maximize the input of model data, respectively. β I β Λ χ I χ These are the coefficients of the linear relationship in the imported sequence formula. These are the errors in the linear relationship of the imported sequence;

[0048] S34: Construct the derived sequence formula and substitute it into the abstract relation matrix of the final influence change dataset for the ι-th operation interval. And the final impact change of the dataset ω in the (1+1)th operation interval. ι+1 , is represented as:

[0049]

[0050] In the formula, ε ι+1 To derive the data output by the sequence formula, Λ ε I ε These are the coefficients for deriving the linear relationship in the sequence formula. To derive the error in the linear relationship of the sequence, This is the abstract relation matrix for the (i+1)th operation interval;

[0051] S35: Construct the impact prediction formula and input the abstract relation matrix of the (i+1)th operation interval. Represented as:

[0052]

[0053] in,

[0054]

[0055] ∫ ι+1 =χ ι+1 *β ι+1 +α ι+1 *∫ ι (18)

[0056] In the formula, ∫ ι and ∫ ι+1 κ represents the model data shape coefficients for the ι-th and ι+1-th operation intervals, respectively; ι+1 This refers to the beam displacement deformation and the bridge tower tilt angle deformation under future external influences; Λ κ The linear relationship coefficients for the result prediction function. The error represents the linear relationship of the prediction function.

[0057] Furthermore, in S4, the beam displacement deformation and tower tilt angle deformation under the future stable changes are added to the beam displacement deformation and tower tilt angle deformation under external influences, respectively, to obtain the future settlement deformation data of the low-tower cable-stayed bridge deck, expressed as:

[0058]

[0059] In the formula, This is the total predicted set of data on the settlement and deformation of the bridge deck of a low-tower cable-stayed bridge.

[0060] Beneficial effects: This invention establishes a stable prediction model and an influence prediction model. It uses the stable prediction model and a stable change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future stable changes. It uses the influence prediction model and the final influence change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future external influences. This method has shorter computation time and more accurate prediction results, greatly improving the efficiency of bridge deck deformation prediction. It can retain the corresponding change characteristics of historical measurement data sets of bridge deck deformation by using prediction functions and models, while fully considering the impact of changes in influencing factors on bridge deck descent deformation. This greatly improves the applicability to predict bridge descent deformation under complex conditions, providing a fast, accurate, and scientific prediction method for bridge deck descent deformation. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 This is a flowchart of a bridge deck deformation prediction method based on a dual-model approach in this invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] This embodiment provides a bridge deck deformation prediction method based on a dual-model approach, such as... Figure 1 As shown, the specific steps include:

[0065] S1: Obtain the historical monitoring dataset of the low-tower cable-stayed bridge, and process the historical monitoring dataset to obtain the stable change dataset and the final impact change dataset;

[0066] In a specific embodiment, S1 involves acquiring the historical monitoring dataset of the low-tower cable-stayed bridge and processing the historical monitoring dataset to obtain the stable change dataset and the final impact change dataset.

[0067] S11: Establish a historical monitoring dataset that includes datasets of external influencing factors and datasets of bridge deck settlement and deformation;

[0068] In a specific embodiment, in S11, the acquired historical monitoring dataset is represented as V = (H, J);

[0069] Where H = (h1, h2...h O ) T H represents the dataset of external influencing factors within O months, h1, h2...h O These represent the monitoring data of external influencing factors for each month; specifically, in this embodiment, month O can be a series of consecutive months or a series of non-consecutive months with equal intervals.

[0070] in, The values ​​represent the vehicle operating load C, bridge self-weight load D, regional seasonal temperature E, near-bridge air humidity F, and wind disturbance G in month O, respectively, in units of t, t, ℃, %RH, and N.

[0071] Where J = (j1, j2, ... j O J represents the bridge deck settlement and deformation dataset corresponding to H within O months, j1, j2...j O These represent the bridge deck settlement and deformation monitoring data for each month;

[0072] in, The values ​​represent the deformation of beam displacement A and the deformation of bridge tower tilt angle B in month O, respectively, in mm and °.

[0073] S12: Based on whether the bridge deck settlement deformation is affected by external factors, the bridge deck settlement deformation dataset is divided into a stable change dataset and an impact change dataset, and the impact change dataset is combined with the external impact factor dataset to obtain the final impact change dataset.

[0074] Specifically, in S12, based on whether the bridge deck settlement deformation is affected by external factors, the bridge settlement deformation monitoring data J = (j1, j2...j O The dataset is divided into two categories: a stable dataset unaffected by external factors and a dataset affected by external factors. The stable dataset is... Combining the aforementioned impact change dataset with the aforementioned external influencing factor dataset yields the final impact change dataset.

[0075] Specifically, in order to more accurately predict bridge deck deformation, the historical monitoring dataset is first decomposed, and then the mapping relationship between the historical data of bridge deck deformation and between various influencing factors and bridge deck deformation is explored by using a stability prediction model and an impact prediction model, so as to make a scientific and reasonable prediction of future bridge deck deformation data.

[0076] S2: Construct a stable prediction model and use the stable prediction model and stable change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future stable changes.

[0077] In a specific embodiment, S2, the process of constructing a stability prediction model and using the stability prediction model and the stability change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future stability changes is as follows:

[0078] S21: Construct the minimum optimization function, expressed as:

[0079]

[0080] In the formula, λ is the optimal solution output by the optimization function matrix, μ is the normal vector coefficient of the height plane, and θ is the standard normalization coefficient. To minimize the error of the optimization function, To project nonlinear functions into a high-dimensional space, It is a constant;

[0081] S22: Construct a Lagrange polynomial based on equation (1) and input it into a stable changing dataset. get:

[0082]

[0083] in,

[0084]

[0085] In the formula, Θ is a Lagrange polynomial; These are the system state coefficients; for The sum of beam displacement deformation and bridge tower tilt angle deformation over a month;

[0086] S23: For μ, respectively Find the partial derivative, and thus the derivative with respect to μ, Dimensionality reduction optimization yielded the following results;

[0087]

[0088] S24: Substitute the result obtained in S23 into the solution of the simplified matrix to get:

[0089]

[0090] in,

[0091]

[0092] In the formula, E is the identity matrix; Z is the kernel function. Specifically, in this embodiment, the kernel function includes, but is not limited to, the radial basis kernel function, the polynomial kernel function, and the linear kernel function.

[0093] S25: Obtained from S24 and A stability prediction function is constructed, and the beam displacement deformation and bridge tower tilt angle deformation under future stable changes are calculated based on the stability prediction function. The formula is as follows:

[0094]

[0095] S3: Construct an impact prediction model, and use the impact prediction model and the final impact change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future external influences.

[0096] In a specific embodiment, S3, the process of constructing an impact prediction model and using the impact prediction model and the final impact change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future external influences is as follows:

[0097] S31: Use the data in the final impact change dataset as learning samples, and divide the learning samples according to the continuous monitoring time. One operation interval;

[0098] S32: Construct the selection sequence formula and substitute it with the abstract relation matrix of the final impact change dataset of the ι-th operation interval. The final impact change of the dataset ω with respect to the (i+1)th operation interval ι+1 , Represented as:

[0099]

[0100] Where, α ι+1 The data is selected using the formula for the sequence, where ψ is the sigmoid function and Λ is the number of elements in the sequence. α I α They are respectively the coefficients of the linear relationship in the formula for the sequence. To select the linear relationship error of the sequence;

[0101] in,

[0102]

[0103] In the formula, CXY is an abstract function, and ∧ is the relation matrix of the final dataset affecting changes. To account for the final error affecting the changed dataset, ω ι This represents the final impact change dataset for the ι-th operation interval;

[0104] S33: Construct the imported sequence formula and input the abstract relation matrix of the final impact change dataset for the ι-th operation interval. The final impact change of the dataset ω with respect to the (i+1)th operation interval ι+1 , is represented as:

[0105]

[0106] In the formula, β ι+1 , χ ι+1 These represent the left and right import formulas to maximize the input of model data, respectively. β I β Λ χ I χ These are the coefficients of the linear relationship in the imported sequence formula. These represent the error in the linear relationship of the imported sequence; tanh is the hyperbola tangent expression function.

[0107] S34: Construct the derived sequence formula and substitute it into the abstract relation matrix of the final influence change dataset for the ι-th operation interval. And the final impact change of the dataset ω in the (1+1)th operation interval. ι+1 , is represented as:

[0108]

[0109] In the formula, ε ι+1 To derive the data output by the sequence formula, Λ ε I ε These are the coefficients for deriving the linear relationship in the sequence formula. To derive the error in the linear relationship of the sequence, This is the abstract relation matrix for the (i+1)th operation interval;

[0110] S35: Construct the impact prediction formula and input the abstract relation matrix of the (i+1)th operation interval. Represented as:

[0111]

[0112] in,

[0113]

[0114] ∫ ι+1 =χ ι+1 *β ι+1 +α ι+1 *∫ ι (18)

[0115] In the formula, ∫ ι and ∫ ι+1 κ represents the model data shape coefficients for the ι-th and ι+1-th operation intervals, respectively; ι+1 This refers to the beam displacement deformation and the bridge tower tilt angle deformation under future external influences; Λ κ The linear relationship coefficients for the result prediction function. The error represents the linear relationship of the prediction function.

[0116] Specifically, since the beam displacement A and tower tilt angle B of the deformed low-tower cable-stayed bridge have a certain mapping relationship with the vehicle running load C, bridge self-weight load D, regional seasonal temperature E, near-bridge air humidity F, and wind disturbance G, and because these influencing factors are constantly changing, this mapping relationship often has nonlinear characteristics and is difficult to represent with simple explicit equations. Therefore, in this embodiment, an influence prediction model is constructed, and the influence prediction model and the final influence change dataset are used to predict the beam displacement deformation and tower tilt angle deformation under future external influences.

[0117] S4: Add the beam displacement deformation and bridge tower tilt angle deformation under the future stable changes to the beam displacement deformation and bridge tower tilt angle deformation under external influences, respectively, to obtain the future settlement deformation data of the low-tower cable-stayed bridge deck.

[0118] In a specific embodiment, in S4, the beam displacement deformation and bridge tower tilt angle deformation under the future stable changes are added to the beam displacement deformation and bridge tower tilt angle deformation under external influences, respectively, to obtain the future settlement deformation data of the low-tower cable-stayed bridge deck, which is expressed as:

[0119]

[0120] In the formula, This is the total predicted set of data on the settlement and deformation of the bridge deck of a low-tower cable-stayed bridge.

[0121] In this embodiment, a monitoring dataset V = (H, J) with a capacity of 30 months was constructed. This dataset is derived from the historical measurement data of bridge deck deformation of bridges in service for 30 months, as shown in Table 1. The last two columns correspond to H = (h1, h2... h... 30 ) T The measurements of beam displacement A and tower tilt angle B over 30 months were obtained from actual measurements taken by on-site sensors.

[0122] Table 1:

[0123]

[0124]

[0125] Specifically, in order to further prove the accuracy of the method in this embodiment, five time points were selected, and the corresponding impact change data and the content of external influencing factor data are shown in Table 2.

[0126] Table 2:

[0127] factor A(mm) B(°) C(N) D(N) E(℃) F(%RH) G(t) τ(mm) κ(°) Test 1 0.72 0.72 56342 37550 -2 48 22 0.75 0.77 Test 2 0.75 0.77 56309 37543 7 50 24 0.81 0.83 Test 3 0.81 0.83 56309 37543 19 52 18 0.83 0.89 Test 4 0.83 0.89 56309 37543 35 53 21 0.85 0.95 Test 5 0.85 0.95 56309 37543 29 57 25 0.90 0.98

[0128] Specifically, Table 3 summarizes the prediction errors of the five test sample groups shown in Table 2. As can be seen from the table, the maximum absolute errors of τ and κ are 0.14 and 0.11, respectively, and the maximum relative errors are 13.5% and 10.1%, respectively. The accuracy is relatively high and can meet the prediction requirements of bridge deck deformation.

[0129] Table 3:

[0130] Test 1 Test 2 Test 3 Test 4 Test 5 τ measured value 0.81 0.87 0.92 0.98 1.04 κ measured value 0.82 0.89 0.96 1.04 1.09 τ predicted value 0.75 0.81 0.83 0.85 0.90 κ predicted value 0.77 0.83 0.89 0.95 0.98 τ absolute error 0.06 -0.06 0.09 0.13 0.14 κ absolute error 0.05 0.06 0.07 0.09 0.11 τ relative error 7.4％ 6.9％ 9.8％ 13.3％ 13.5％ κ relative error 6.1％ 6.7％ 7.3％ 8.7％ 10.1％

[0131] The beneficial effects of this invention are as follows: Based on two prediction models, accurate prediction of beam displacement deformation and bridge tower tilt angle deformation during bridge operation is achieved, avoiding the reliance on the subjective experience of construction personnel in previous methods and improving the accuracy of the calculation results. Specifically:

[0132] (1) There is a certain mapping relationship among historical bridge deck deformation displacement, influencing factors, and future bridge deck deformation displacement, but this relationship is often implicit and relatively complex. The multi-model parallel prediction method established in this invention can make full use of limited information to perform a large amount of calculation, giving full play to the advantages of mathematical statistics, and predicting bridge settlement deformation caused by external influencing factors and bridge settlement deformation not caused by external influencing factors, respectively, so that the prediction results are more stable and accurate. At the same time, the method is simple to implement, and after training, it can obtain the prediction relationship between historical bridge deck deformation displacement, influencing factors, and future bridge deck deformation displacement. In particular, for imbalanced datasets, the method can balance the error and maintain high computational accuracy even when some features are lost.

[0133] (2) In the process of predicting bridge deck settlement and deformation, extreme influencing factors often occur, affecting the model's accurate prediction of bridge deformation. The prediction model given in this invention screens and detects the input data of the model by using relevant function sequences. This not only scientifically avoids the influence of extreme external influencing factors on the model prediction results, but also effectively solves the time lag effect of external influencing factors on bridge deck deformation and displacement, thereby obtaining scientific and accurate prediction results.

[0134] (3) This invention has developed two prediction models. By relying on the models to summarize and process the monitoring data, the problem of data loss and confusion is avoided, and the standardized storage of bridge deformation data is realized.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting bridge deck deformation based on a dual-model approach, characterized in that, The specific steps include: S1: Obtain the historical monitoring dataset of the low-tower cable-stayed bridge, and process the historical monitoring dataset to obtain the stable change dataset and the final impact change dataset; S2: Construct a stable prediction model and use the stable prediction model and stable change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future stable changes. In S2, the process of constructing a stable prediction model and using the stable prediction model and the stable change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future stable changes is as follows: S21: Construct the minimum optimization function, expressed as: (1) In the formula, To optimize the optimal solution output by the function matrix, The normal vector coefficients of the high plane are... For standard normalization coefficients, To minimize the error of the optimization function, To project nonlinear functions into a high-dimensional space, It is a constant; S22: Construct a Lagrange polynomial based on equation (1) and input it into a stable changing dataset. ,get: (2) in, (3) In the formula, It is a Lagrange polynomial; These are the system state coefficients; for The sum of beam displacement deformation and bridge tower tilt angle deformation over a month; S23: respectively for Taking the partial derivative, we obtain the following results; (4) (5) (6) =0 (7) S24: Substitute the result obtained in S23 into the solution of the simplified matrix to get: (8) in, ， x=y =1,2,..., (9) In the formula, It is the identity matrix; For kernel functions; S25: Obtained from S24 and A stability prediction function is constructed, and the beam displacement deformation and bridge tower tilt angle deformation under future stable changes are calculated based on the stability prediction function. The formula is as follows: (10) S3: Construct an impact prediction model, and use the impact prediction model and the final impact change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future external influences. S4: Add the beam displacement deformation and bridge tower tilt angle deformation under the future stable changes to the beam displacement deformation and bridge tower tilt angle deformation under external influences, respectively, to obtain the future settlement deformation data of the low-tower cable-stayed bridge deck.

2. The bridge deck deformation prediction method based on a dual-model according to claim 1, characterized in that, In S1, the process of acquiring the historical monitoring dataset of the low-tower cable-stayed bridge and processing the historical monitoring dataset to obtain the stable change dataset and the final impact change dataset is as follows: S11: Establish a historical monitoring dataset that includes datasets of external influencing factors and datasets of bridge deck settlement and deformation; S12: Based on whether the bridge deck settlement deformation is affected by external factors, the bridge deck settlement deformation dataset is divided into a stable change dataset and an impact change dataset, and the impact change dataset is combined with the external impact factor dataset to obtain the final impact change dataset.

3. The bridge deck deformation prediction method based on a dual-model according to claim 2, characterized in that, In S11, the acquired historical monitoring dataset is represented as follows: ; in, , H express Data set of external influencing factors within one month These represent the monitoring data of external influencing factors for each month; in, ; ; ; They represent the first The monthly vehicle operating load (C), bridge self-weight load (D), regional seasonal temperature (E), near-bridge air humidity (F), and wind disturbance (G) content, respectively, are as follows: ; in, , J express Within a month H The corresponding bridge deck settlement and deformation dataset, These represent the bridge deck settlement and deformation monitoring data for each month; in, ; ; ; They represent the first The deformation of beam displacement A and the deformation of bridge tower tilt angle B over one month, respectively, in units of mm and °; In S12, the bridge settlement deformation monitoring data are analyzed based on whether the bridge deck settlement deformation is affected by external factors. The dataset is divided into two categories: a stable change dataset unaffected by external factors and a change dataset affected by external factors. The stable change dataset is... The impact change dataset is combined with the external influencing factor dataset to obtain the final impact change dataset. .

4. The bridge deck deformation prediction method based on a dual-model according to claim 3, characterized in that, In S3, the process of constructing an impact prediction model and using the model and the final impact change dataset to predict the beam displacement deformation and bridge tower tilt angle deformation under future external influences is as follows: S31: Use the data in the final impact change dataset as learning samples, and divide the learning samples according to the continuous monitoring time. One operation interval; S32: Construct a selection sequence formula and substitute it with the first... The abstract relation matrix of the final impact changes of each operation interval on the dataset. With the The final impact change of the dataset for each computational interval , ={1,2,3,…, -1} is represented as: (11) in, For data filtered using the selected sequence formula, For the sigmoid function, They are respectively the coefficients of the linear relationship in the formula for the sequence. To select the linear relationship error of the sequence; in, (12) In the formula, For abstract functions, The final relational matrix of the impact change dataset. To account for the error in the final impact on the changed dataset, For the first The final impact changes of the dataset across each computational interval; S33: Construct the formula for importing the sequence and substitute it into the first... The abstract relation matrix of the final impact changes of each operation interval on the dataset. With the The final impact change of the dataset for each computational interval , represented as: (13) (14) In the formula, These represent the methods for maximizing the input of model data using left-side and right-side import sequence formulas, respectively. These are the coefficients of the linear relationship in the imported sequence formula. These are the errors in the linear relationship of the imported sequences; S34: Construct the formula for deriving the sequence and substitute it into the first... The abstract relation matrix of the final impact changes of each operation interval on the dataset. and the The final impact change of the dataset for each computational interval , represented as: (15) In the formula, To derive the data output by the sequence formula, These are the coefficients for deriving the linear relationship in the sequence formula. To derive the error in the linear relationship of the sequence, For the first An abstract relation matrix of operation intervals; S35: Construct the impact prediction formula and substitute it into the first... Abstract relational matrix of each operation interval , represented as: (16) in, (17) (18) In the formula, and The first The operation interval and the first Model data shape coefficients for each computational interval; This refers to the beam displacement and deformation, as well as the bridge tower tilt angle and deformation, under future external influences. The linear relationship coefficients for the result prediction function. The error represents the linear relationship of the prediction function.

5. The bridge deck deformation prediction method based on a dual-model according to claim 4, characterized in that, In S4, the beam displacement deformation and bridge tower tilt angle deformation under the future stable changes are added to the beam displacement deformation and bridge tower tilt angle deformation under external influences, respectively, to obtain the future settlement deformation data of the low-tower cable-stayed bridge deck, expressed as: (19) In the formula, This is the total predicted set of data on the settlement and deformation of the bridge deck of a low-tower cable-stayed bridge.

Images