Long-life composite plate girder bridge vertical temperature gradient fatigue load model and construction method

Abstract

The invention discloses a vertical temperature gradient fatigue load model of a long-service-life composite plate girder bridge. The vertical temperature gradient fatigue load model is of a three-fold line type. According to the method, long-term monitoring data of a vertical temperature field of the composite plate girder is utilized, a long-short memory neural network deep learning algorithm is adopted to predict a temperature history within a designed service life, and an equal damage equivalence principle is combined, so that a construction method of a vertical temperature gradient fatigue load model of the long-life composite plate girder bridge is provided; and a vertical temperature gradient fatigue load model of the composite plate girder bridge with the designed service life of 200 years is constructed. The model realizes the rapid calculation of the temperature fatigue stress process of the composite plate girder bridge, and solves the problem of no load model during the temperature fatigue damage analysis of the composite plate girder bridge.

Description

Technical Field

[0001] The invention belongs to the technical field of bridge engineering, and specifically relates to a vertical temperature gradient fatigue load model and a construction method for a long-life composite plate-girder bridge. Background Art

[0002] Composite plate girder bridges usually use I-section steel beams. The thermal conductivity coefficients of the composite layer and the steel beams are somewhat different. Under the action of solar radiation, a complex nonlinear vertical temperature gradient will be formed in the composite beam section. As the temperature gradient changes alternately during the day and night, the composite beam section will experience periodic cyclic temperature fatigue stress, which contributes to the cumulative fatigue damage of the structure. Existing studies have shown that the coupling effect of temperature load and vehicle load will significantly increase the cumulative fatigue damage of the composite beam bridge details, and the fatigue damage under the action of temperature and vehicle alone cannot be simply linearly superimposed. Therefore, it is necessary to carry out long-term monitoring of the composite beam temperature field, use a reasonable deep learning algorithm to predict the temperature history within the design service life, and combine the damage equivalence principle to propose a method for constructing a vertical temperature gradient fatigue load model for long-life composite plate girder bridges. The vertical temperature gradient fatigue load model for long-life composite plate girder bridges is constructed, and the temperature fatigue stress history of composite plate girder bridges is calculated. It is sequentially coupled with the vehicle fatigue stress to conduct temperature-vehicle coupled fatigue damage analysis. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a vertical temperature gradient fatigue load model for a long-life composite plate girder bridge.

[0004] Another technical problem to be solved by the present invention is to provide a construction method for a vertical temperature gradient fatigue load model of a long-life composite plate girder bridge.

[0005] The vertical temperature gradient fatigue load model of the long-life composite plate girder bridge provided by the present invention is composed of formula (1):

[0006]

[0007] In formula (1), y is the vertical distance from the bottom surface of the bottom plate, h is the height of the composite plate beam, and h s is the height of the steel plate beam, h c is the height of the combined layer, T1 is the representative value of the temperature at the top plate of the combined plate-beam section, in °C, and T2 is the distance from the top plate to the combined plate-beam section h c The representative value of the temperature at the position, in °C, T3 is the temperature at which the composite plate beam section is 0.7h from the top plate s The representative value of the temperature at the position, in °C, T4 is the representative value of the temperature at the bottom plate of the composite plate beam section, in °C, β T (h), β T (b), β T (B), β T(k) is the intermediate variable, b is the width of the lower flange of the composite plate beam, in m, B is the width of the concrete deck of the composite plate beam bridge, in m, k is the number of I-beams in the composite plate beam, T1 0 is the initial temperature value at the top plate of the composite plate-beam section, in °C. h is the distance between the composite plate beam section and the top plate c Initial temperature value at position, in °C, T3 0 The distance between the composite plate beam section and the top plate is 0.7h s The initial temperature value at position, in °C, is the initial temperature of the bottom plate of the composite plate-beam section, in °C, N d It is the design service life, and its value is 100, 150 or 200, in years.

[0008] In formula (1), preferably, T1 0 The value range is [4.2,5.4], the unit is ℃, the value range of h is [1.2,4.0], the unit is m, the value of b is [0.3,0.6], the unit is m, the value of B is [1.5,30], the unit is m, the value of k is [2,6], and k is an integer.

[0009] More preferably, T1 in formula (1) 0 The values ​​of h, b, B, k are: T1 0 is 4.8℃, h is 1.41m, b is 0.5m, B is 6m, and k is 4.

[0010] The method for constructing a vertical temperature gradient fatigue load model of a long-life composite plate girder bridge provided by the present invention comprises the following steps:

[0011] (1) Temperature monitoring and preliminary data processing

[0012] Temperature measurement points are arranged inside the concrete composite slab and on the top, bottom and web of the steel plate beam of the composite slab beam bridge to collect temperature, denoted as T A (y), the collection time interval is 1 minute to 20 minutes, and the temperature gradient time history T is constructed A (y,t), where t is any moment in the measured temperature history.

[0013] (2) Standardization of raw data

[0014] For the temperature gradient time history T A (y, t) is processed as follows to obtain the processing standard data T sta (y,t):

[0015]

[0016] Where: E(·) is the mean of the measured temperature data, and D(·) is the variance of the measured temperature data.

[0017] (3) Constructing a long short-term memory neural network

[0018] The long short-term memory neural network includes a forget gate, an input gate, and an output gate. The forget gate is connected in parallel with the input gate and then in series with the output gate.

[0019] The construction method of long short-term memory neural network is as follows:

[0020] f t =σ(W f ×(h t-1 ,x t )+b f )

[0021] I t =σ(W I ×(h t-1 ,x t )+b I )

[0022] c t =f t oc t-1 +i t otanh(W c ×(h t-1 ,x t )+b c )

[0023] o t =σ(W o ×(h t-1 ,x t )+b o )

[0024] h t =o t otanh(c t )

[0025] In the formula, f t represents the forget gate, I t represents the input gate, c t Represents the cell state, the initial cell state matrix is ​​set to an empty matrix, o t represents the output gate, h t represents the final output value, the o symbol represents the prime multiplication of the unit; σ(·) is the sigmoid function expression, tanh(·) is the hyperbolic tangent function expression, W f is the weight matrix of the forget gate, which is initially set to the identity matrix, b f is the bias term of the forget gate, initially set to a zero matrix, W Iis the weight matrix of the input gate, b I is the bias term of the input gate, W c is the weight matrix of the unit state, b c is the bias term of the unit state, W o is the weight matrix of the output gate, b o is the bias term of the output gate.

[0026] (4) Training long short-term memory neural network

[0027] The standard data T sta (y, t), weight matrix, and bias term are input into the long short-term memory neural network for training, with an initial learning rate of α, α∈[0.001, 0.01], and the number of learning times num2∈[1000, 12000], until the error between the weight matrix and bias term of the output gate is less than 0.01, and the training is completed.

[0028] The error L of the weight matrix of the output gate W According to the following formula:

[0029]

[0030] In the formula, is the output gate weight matrix obtained from the u-th training.

[0031] The error L of the output gate bias term bo According to the following formula:

[0032]

[0033] In the formula, is the bias term obtained from the u-th training.

[0034] (5) Prediction of temperature gradient time course during design service life

[0035] Based on the constructed long short-term memory neural network, the last temperature data collected is used as the initial point of prediction and input into the long short-term memory neural network to calculate the design service life N. d All data within are predicted to obtain the design service life N d Corresponding data standard value The design service life N is obtained by the following formula: d Corresponding temperature gradient time history

[0036]

[0037] (6) Determine the equivalent fatigue temperature at typical locations

[0038] The temperature stress history is obtained by the following formula:

[0039]

[0040] In the formula, σ c (y) is the concrete composite layer stress, σ s (y) is the cross-sectional stress of the steel box girder, E s is the elastic modulus of steel, α s is the thermal expansion coefficient of steel, E c is the elastic modulus of the combined layer material, α c is the thermal expansion coefficient of the combined layer material, b s (y) is the thickness of the steel beam at position y, b c (y) is the thickness of the combined layer at position y, I s is the moment of inertia of the steel beam section, I c is the moment of inertia of the combined layer section, ρ is the structural system coefficient, M so 、N so 、M co 、M so is an intermediate variable.

[0041] The temperature stress time history is processed based on the rain flow counting method to obtain the temperature stress amplitude vector corresponding to the design service life.

[0042] 1) For the stress amplitude vector A within the steel beam range ys :

[0043]

[0044] In the formula, is the m1th stress amplitude at the y position, where m1 represents the number of elements in the vector and is a positive integer.

[0045] 2) For the stress amplitude vector A within the combined layer yc :

[0046]

[0047] In the formula, is the m2th stress amplitude at the y position, where m2 represents the number of elements in the vector and is a positive integer.

[0048] The equivalent temperature stress is obtained by using the above stress amplitude vector as follows:

[0049] 1) For the equivalent temperature stress within the steel beam range

[0050]

[0051] In the formula, is the vector A ys The i-th element in .

[0052] 2) For the equivalent temperature stress within the combined layer

[0053]

[0054] In the formula, is the vector A yc The jth element in .

[0055] The equivalent temperature at a typical location is constructed as follows

[0056] 1) For the equivalent temperature within the steel beam range

[0057]

[0058] 2) For the equivalent temperature within the combined layer

[0059]

[0060] (7) Constructing vertical temperature gradient fatigue load model

[0061] Using Matlab software, we can analyze the different positions y and different design service life N. d The equivalent temperature is fitted by the least square method, and the vertical temperature gradient fatigue load model of the long-life composite plate girder bridge is established as shown in formula (1).

[0062] The beneficial effects of the present invention are as follows:

[0063] 1. The present invention adopts the long short-term memory neural network deep learning algorithm and the equivalent fatigue damage principle to propose a method for constructing a vertical temperature gradient fatigue load model for a long-life composite plate-girder bridge, realizing the extension of the short-term temperature history to the long-life design service life.

[0064] 2. The present invention uses the long-term measured data of the temperature field of the highway composite plate girder bridge to construct a vertical temperature gradient fatigue load model that can be used for the fatigue design and evaluation of the composite plate girder bridge with a design service life of 200 years. The constructed vertical temperature gradient fatigue load model takes into account the composite plate girder section height h, the concrete bridge deck width B, the bottom plate width b, the number of steel plate girders k, and the design service life N. d impact. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 Distribution of representative temperature values ​​for the vertical temperature gradient fatigue load model of composite plate-girder bridge.

[0066] Figure 2 The figure is a flow chart for constructing the model of the present invention.

[0067] Figure 3 This is the temperature time history prediction result of long short-term memory neural network.

[0068] Figure 4 This is the layout diagram of the temperature measurement points for the composite plate girder.

[0069] Figure 5 This is the measured temperature history curve of the composite slab-girder bridge.

[0070] Figure 6 It is the stress amplitude histogram of the measured temperature field of the composite plate girder bridge. DETAILED DESCRIPTION

[0071] The present invention is further described in detail below with reference to the accompanying drawings and embodiments, but the present invention is not limited to these embodiments.

[0072] Example 1

[0073] This embodiment involves a vertical temperature gradient fatigue load model for a long-life composite plate girder bridge, which is composed of formula (1):

[0074]

[0075] In formula (1), y is the vertical distance from the top plate of the section at the typical position, h is the height of the composite plate beam, and h s is the height of the steel plate beam, h c is the height of the combined layer, T1 is the representative value of the temperature at the top plate of the combined plate-beam section, in °C, and T2 is the distance from the top plate to the combined plate-beam section h c The representative value of the temperature at the position, in °C, T3 is the temperature at which the composite plate beam section is 0.7h from the top plate s The representative value of the temperature at the position, in °C, T4 is the representative value of the temperature at the bottom plate of the composite plate beam section, in °C, β T (h), β T (b), β T (B), β T (k) is the intermediate variable, b is the width of the bottom plate of the composite plate beam, in m, B is the width of the concrete deck of the composite plate beam bridge, in m, k is the number of I-beams in the composite plate beam, T1 0 is the initial temperature value at the top plate of the composite plate-beam section, in °C. h is the distance between the composite plate beam section and the top plate c Initial temperature value at position, in °C, T3 0 The distance between the composite plate beam section and the top plate is 0.7h s The initial temperature value at the position, in °C, is the initial temperature of the bottom plate of the composite plate-beam section, in °C, N dIt is the design service life, and its value is 100, 150 or 200, in years.

[0076] In the present embodiment, the beam height h is selected as 1.41 m in formula (1), wherein the steel plate beam height h s is 1.2m, the combined layer height h c is 0.21m, the concrete bridge deck width B is 6.0m, the bottom plate width b is 0.5m, the number of I-beams k is 4, T1 0 The value of is 4.8℃, and the corresponding vertical temperature gradient fatigue load model of composite plate girder bridge is constructed. The temperature representative values ​​corresponding to the long-life design service life of 100 years, 150 years, and 200 years are shown in Table 1. The distribution of temperature representative values ​​in the vertical temperature gradient fatigue load model constructed according to this parameter is shown in Figure 1 shown.

[0077] Table 1 Representative values ​​of temperature for vertical temperature gradient fatigue load model of long-life composite plate girder bridge

[0078]

[0079] like Figure 2 As shown, the method for constructing the vertical temperature gradient fatigue load model of the long-life composite plate girder bridge is composed of the following steps:

[0080] (1) Temperature monitoring and preliminary data processing

[0081] Temperature measurement points are arranged inside the concrete composite slab and on the top, bottom and web of the steel plate beam of the composite slab beam bridge to collect temperature, denoted as T A (y), the collection time interval is 1 minute to 20 minutes, and the temperature gradient time history T is constructed A (y, t), where t is any time in the measured temperature history, which is 5 minutes in this embodiment.

[0082] (2) Standardization of raw data

[0083] For the temperature gradient time history T A (y, t) is processed as follows to obtain the processing standard data T sta (y,t):

[0084]

[0085] Where: E(·) is the mean of the measured temperature data, and D(·) is the variance of the measured temperature data.

[0086] (3) Constructing a long short-term memory neural network

[0087] The long short-term memory neural network includes a forget gate, an input gate, and an output gate. The forget gate is connected in parallel with the input gate and then in series with the output gate.

[0088] The construction method of long short-term memory neural network is as follows:

[0089] f t =σ(W f ×(h t-1 ,x t )+b f )

[0090] I t =σ(W I ×(h t-1 ,x t )+b I )

[0091] c t =f t oc t-1 +i t otanh(W c ×(h t-1 ,x t )+b c )

[0092] o t =σ(W o ×(h t-1 ,x t )+b o )

[0093] h t =o t otanh(c t )

[0094] In the formula, f t represents the forget gate, I t represents the input gate, c t Represents the cell state, the initial cell state matrix is ​​set to an empty matrix, o t represents the output gate, h t represents the final output value, the o symbol represents the prime multiplication of the unit, σ(·) is the sigmoid function expression, tanh(·) is the hyperbolic tangent function expression, W f is the weight matrix of the forget gate, which is initially set to the identity matrix, b f is the bias term of the forget gate, initially set to a zero matrix, W I is the weight matrix of the input gate, b I is the bias term of the input gate, W c is the weight matrix of the unit state, b c is the bias term of the cell state, W o is the weight matrix of the output gate, b o is the bias term of the output gate.

[0095] (4) Training long short-term memory neural network

[0096] The standard data T sta (y, t), weight matrix, bias term are input into the long short-term memory neural network for training, with an initial learning rate α, α∈[0.001, 0.01], and the number of learning times num2∈[1000, 12000], until the error between the weight matrix and the bias term of the output gate is less than 0.01, and the training is completed. In this embodiment, α=0.01, num2=1200.

[0097] The error L of the weight matrix of the output gate W According to the following formula:

[0098]

[0099] In the formula, is the output gate weight matrix obtained from the u-th training.

[0100] The error L of the output gate bias term bo According to the following formula:

[0101]

[0102] In the formula, is the bias term obtained from the u-th training.

[0103] (5) Prediction of temperature gradient time course during design service life

[0104] Based on the constructed long short-term memory neural network, the last temperature data collected is used as the initial point of prediction and input into the long short-term memory neural network to calculate the design service life N. d All data within are predicted to obtain the design service life N d Corresponding data standard value The prediction results are as follows Figure 3 As shown. The design service life N is obtained by the following formula d Corresponding temperature gradient time history

[0105]

[0106] (6) Determine the equivalent fatigue temperature at typical locations

[0107] The temperature stress time history is obtained by the following formula:

[0108]

[0109] In the formula, σ c (y) is the concrete composite layer stress, σ s (y) is the cross-sectional stress of the steel box girder, Es is the elastic modulus of steel, α s is the thermal expansion coefficient of steel, E c is the elastic modulus of the combined layer material, α c is the thermal expansion coefficient of the combined layer material, b s (y) is the thickness of the steel beam at position y, b c (y) is the thickness of the combined layer at position y, I s is the moment of inertia of the steel beam section, I c is the moment of inertia of the combined layer section, ρ is the structural system coefficient, M so 、N so 、M co 、M so is the intermediate variable, I c is the moment of inertia of the combined layer section, ρ is the structural system coefficient, M so 、N so 、M co 、M so is an intermediate variable. In this embodiment, E s 2.06×10 5 MPa, α s 1.1×10 -5 ℃ -1 , E c 3.45×10 4 MPa, α s 1.1×10 -5 ℃ -1 .

[0110] The temperature stress time history is processed based on the rain flow counting method to obtain the temperature stress amplitude vector corresponding to the design service life.

[0111] 1) For the stress amplitude vector A within the steel beam range ys :

[0112]

[0113] In the formula, is the m1th stress amplitude at the y position, where m1 represents the number of elements in the vector and is a positive integer.

[0114] 2) For the stress amplitude vector A within the combined layer yc :

[0115]

[0116] In the formula, is the m2th stress amplitude at the y position, where m2 represents the number of elements in the vector and is a positive integer.

[0117] The equivalent temperature stress is obtained by using the above stress amplitude vector as follows:

[0118] 1) For the equivalent temperature stress within the steel beam range

[0119]

[0120] In the formula, is the vector A ys The i-th element in .

[0121] 2) For the equivalent temperature stress within the combined layer

[0122]

[0123] In the formula, is the vector A yc The jth element in .

[0124] The equivalent temperature at a typical location is constructed as follows

[0125] 1) For the equivalent temperature within the steel beam range

[0126]

[0127] 2) For the equivalent temperature within the combined layer

[0128]

[0129] (7) Constructing vertical temperature gradient fatigue load model

[0130] Using Matlab software, we can analyze the different positions y and different design service life N. d The equivalent temperature is fitted by the least square method, and the vertical temperature gradient fatigue load model of long-life composite plate girder bridge is established as shown in formula (1).

[0131] Example 2

[0132] The vertical temperature gradient fatigue load model of the long-life composite plate girder bridge involved in this embodiment is shown in formula (1), and the construction method is the same as that of Example 1.

[0133] In formula (1), the beam height h is selected as 1.2m, where the steel plate beam height h s is 1.0m, and the combined layer height h c is 0.2m, the concrete bridge deck width B is 1.5m, the bottom plate width b is 0.3m, the number of I-beams k is 2, T1 0The value of is 4.2℃, and the other parameters are the same as those in Example 1. The corresponding vertical temperature gradient fatigue load model of the long-life composite plate girder bridge is constructed, and the temperature representative values ​​corresponding to the design service life of 100 years, 150 years, and 200 years are shown in Table 2. The temperature representative values ​​of the vertical temperature gradient fatigue load model constructed according to the parameters are as follows: Figure 1 shown.

[0134] Table 2 Representative values ​​of temperature for vertical temperature gradient fatigue load model of long-life composite plate girder bridge

[0135]

[0136] Example 3

[0137] The vertical temperature gradient fatigue load model of the long-life composite plate girder bridge involved in this embodiment is shown in formula (1), and the construction method is the same as that of Example 1.

[0138] In formula (1), the beam height h is selected as 4.0m, where the steel plate beam height h s is 3.5m, the combined layer height h c is 0.5m, the concrete bridge deck width B is 30.0m, the bottom plate width b is 0.6m, the number of I-beams k is 6, T1 0 The value of is 5.4℃, and the other parameters are the same as those in Example 1. The corresponding vertical temperature gradient fatigue load model of the long-life composite plate girder bridge is constructed, and the temperature representative values ​​corresponding to 100 years, 150 years, and 200 years are shown in Table 3. The temperature representative values ​​of the vertical temperature gradient fatigue load model constructed according to the parameters are as follows Figure 1 As shown.

[0139] Table 3 Representative values ​​of temperature for vertical temperature gradient fatigue load model of long-life composite plate girder bridge

[0140]

[0141] Test 1

[0142] In order to verify the effect of the vertical temperature gradient fatigue load model of the long-life composite plate girder bridge, the inventors used the vertical temperature gradient fatigue load model of the composite plate girder bridge in Example 1 and its construction method, arranged vertical temperature sensors on the composite plate test beam of the Weishui Campus of Chang'an University and carried out long-term temperature field monitoring. The actual bridge photos are shown in Figure 2.

[0143] Figure 4 shown.

[0144] 1. Long-term monitoring instruments

[0145] The monitoring instrument is a 60-channel JM3813 multifunctional data acquisition instrument produced in Yangzhou, Jiangsu Province, and the temperature measurement point uses a three-wire Pt100 sensor.

[0146] 2. Temperature measurement point arrangement

[0147] Temperature measuring points are arranged on the top plate, bottom plate and web of a test steel plate beam and temperature is collected. The time interval for each temperature collection is set to 5 minutes. The specific arrangement of the measuring points is as follows:

[0148] Figure 4 As shown in the figure, specifically: the temperature measuring points arranged on the steel webs on both sides of the composite steel plate beam in the height direction are represented by their vertical distances from the upper surface of the top plate, which are 0.00m, 0.12m, 0.2m, 0.21m, 0.26m, 0.36m, 0.51m, 0.96m, 1.41m, and h, respectively, where h is the height of the composite steel plate beam. The obtained temperature time course is shown in the figure below: Figure 5 As shown, denoted as T te (y,t).

[0149] 3. Analysis of the effect of vertical temperature gradient fatigue load model

[0150] The stress time history at different heights is obtained according to the simplified temperature stress formula, where the stress calculation formula is:

[0151]

[0152] Where: c (y) is the concrete composite layer stress, σ s (y) is the cross-sectional stress of the steel box girder, E s is the elastic modulus of steel, α s is the thermal expansion coefficient of steel, E c is the elastic modulus of the pavement material, α c is the thermal expansion coefficient of the pavement material, A s is the cross-sectional area of ​​the flat steel box girder, A c is the cross-sectional area of ​​the pavement layer, I s is the moment of inertia of the flat steel box girder section, I c is the moment of inertia of the pavement section, γ is the structural system coefficient, in this case E s 2.06×10 5 MPa, α s 1.1×10 -5 ℃ -1 , E c 3.45×10 4 MPa, α c 1.1×10 -5 ℃ -1 .

[0153] Taking the bottom plate butt weld of the composite plate-beam bridge as an example, the temperature stress history is plotted based on the temperature measurement monitoring data, and the temperature stress amplitude cycle counting is performed using the rain flow counting method to obtain the temperature stress amplitude histogram. The results are as follows: Figure 6 Then the equivalent fatigue stress of the measured results was calculated, and the result was 9.2MPa.

[0154] Based on the vertical temperature gradient fatigue load model of composite plate-girder bridge, the temperature stress history is constructed and the temperature stress amplitude analysis is performed. The beam height h is 1.41m, the concrete bridge deck width B is 4.5m, the bottom plate width b is 0.5m, the number of I-beams k is 2, y=0, and the design service life N is d is 100 years, and the equivalent fatigue stress is obtained by substituting it into the following formula;

[0155] σ eq (y) = α s ×E s ×T(y)

[0156] The equivalent temperature fatigue stress is 9.4 MPa, which is basically consistent with the equivalent fatigue stress calculated from the measured results. This shows that the vertical temperature gradient fatigue load model of the composite plate girder bridge has good applicability and can be used for the vertical temperature and temperature-vehicle coupled fatigue damage calculation of the composite plate girder bridge.

Claims

1. A vertical temperature gradient fatigue load model for long-life composite plate girder bridges, characterized in that The model is composed of formula (1): In formula (1), y is the vertical distance from the bottom surface of the bottom plate, h is the height of the composite plate beam, and h s is the height of the steel plate beam, h c is the height of the combined layer, T1 is the representative value of the temperature at the top plate of the combined plate-beam section, in °C, and T2 is the distance from the top plate to the combined plate-beam section. c The representative value of the temperature at the position, in °C, T3 is the temperature at which the composite plate beam section is 0.7h from the top plate s The representative value of the temperature at the position, in °C, T4 is the representative value of the temperature at the bottom plate of the composite plate beam section, in °C, β T (h), β T (b), β T (B), β T (k) is the intermediate variable, b is the width of the lower flange of the composite plate beam, in m, B is the width of the concrete deck of the composite plate beam bridge, in m, k is the number of I-beams in the composite plate beam, T1 0 is the initial temperature value at the top plate of the composite plate-beam section, in °C. h is the distance between the composite plate beam section and the top plate c Initial temperature value at position, in °C, T3 0 The distance between the composite plate beam section and the top plate is 0.7h s The initial temperature value at position, in °C, is the initial temperature of the bottom plate of the composite plate-beam section, in °C, N d It is the design service life, and its value is 100, 150 or 200, in years.

2. The vertical temperature gradient fatigue load model for long-life composite plate girder bridge according to claim 1 is characterized by: In formula (1), the T1 0 The value range is [4.2,5.4], the unit is ℃, the value range of h is [1.2,4.0], the unit is m, the value of b is [0.3,0.6], the unit is m, the value of B is [1.5,30], the unit is m, the value of k is [2,6], and k is an integer.

3. A vertical temperature gradient fatigue load model for composite plate girder bridge according to claim 1, characterized in that T1 described in formula (1) 0 The values ​​of h, b, B, k are: T1 0 is 4.8℃, h is 1.41m, b is 0.5m, B is 6m, and k is 4.

4. A method for constructing a vertical temperature gradient fatigue load model for a long-life composite plate girder bridge according to claim 1, comprising the following steps: (1) Temperature monitoring and preliminary data processing Temperature measuring points are arranged inside the concrete composite slab and on the top, bottom and web of the steel plate beam of the composite slab beam bridge to collect temperature, denoted as T A (y), the collection time interval is 1 minute to 20 minutes, and the temperature gradient time history T is constructed A (y, t), where t is any time in the measured temperature history; (2) Standardization of raw data For the temperature gradient time history T A (y, t) is processed as follows to obtain the processing standard data T sta (y,t): Where: E(·) is the mean of the measured temperature data, and D(·) is the variance of the measured temperature data; (3) Constructing a long short-term memory neural network The long short-term memory neural network includes a forget gate, an input gate, and an output gate. The forget gate is connected in parallel with the input gate and then connected in series with the output gate. The construction method of the long short-term memory neural network is as follows: f t =σ(W f ×(h t-1 ,x t )+b f ) I t =σ(W I ×(h t-1 ,x t )+b I ) c t =f t oc t-1 +i t otanh(W c ×(h t-1 ,x t )+b c ) the t =σ(W o ×(h t-1 ,x t )+b o ) h t =o t otanh(c t ) In the formula, f t represents the forget gate, I t represents the input gate, c t Represents the cell state, the initial cell state matrix is ​​set to an empty matrix, o t represents the output gate, h t represents the final output value, the o symbol represents the prime multiplication of the unit; σ(·) is the sigmoid function expression, tanh(·) is the hyperbolic tangent function expression, W f is the weight matrix of the forget gate, which is initially set to the identity matrix, b f is the bias term of the forget gate, initially set to a zero matrix, W I is the weight matrix of the input gate, b I is the bias term of the input gate, W c is the weight matrix of the unit state, b c is the bias term of the unit state, W o is the weight matrix of the output gate, b o is the bias term of the output gate; (4) Training long short-term memory neural network The standard data T sta (y, t), weight matrix, and bias term are input into the long short-term memory neural network for training, with an initial learning rate of α, α∈[0.001, 0.01], and the number of learning times num2∈[1000, 12000], until the error between the weight matrix and bias term of the output gate is less than 0.01, and the training is completed; The error L of the weight matrix of the output gate W According to the following formula: In the formula, is the output gate weight matrix obtained from the u-th training; The error L of the output gate bias term bo According to the following formula: In the formula, is the bias term obtained from the u-th training; (5) Prediction of temperature gradient time course during design service life Based on the constructed long short-term memory neural network, the last temperature data collected is used as the initial point of prediction and input into the long short-term memory neural network to calculate the design service life N. d All data within are predicted to obtain the design service life N d Corresponding data standard value The design service life N is obtained by the following formula: d Corresponding temperature gradient time history (6) Determine the equivalent fatigue temperature at typical locations The temperature stress history is obtained by the following formula: In the formula, σ c (y) is the concrete composite layer stress, σ s (y) is the cross-sectional stress of the steel box girder, E s is the elastic modulus of steel, α s is the thermal expansion coefficient of steel, E c is the elastic modulus of the combined layer material, α c is the thermal expansion coefficient of the combined layer material, b s (y) is the thickness of the steel beam at position y, b c (y) is the thickness of the combined layer at position y, I s is the moment of inertia of the steel beam section, I c is the moment of inertia of the combined layer section, ρ is the structural system coefficient, M so 、N so 、M co 、M so is an intermediate variable; The temperature stress time history is processed based on the rain flow counting method to obtain the temperature stress amplitude vector corresponding to the design service life: 1) For the stress amplitude vector A within the steel beam range ys : In the formula, is the m1th stress amplitude at the y position, where m1 represents the number of elements in the vector and is a positive integer; 2) For the stress amplitude vector A within the combined layer yc : In the formula, is the m2th stress amplitude at the y position, where m2 represents the number of elements in the vector and is a positive integer; The equivalent temperature stress is obtained by using the above stress amplitude vector as follows: 1) For the equivalent temperature stress within the steel beam range In the formula, is the vector A ys The i-th element in ; 2) For the equivalent temperature stress within the combined layer In the formula, is the vector A yc The jth element in ; The equivalent temperature at a typical location is constructed as follows 1) For the equivalent temperature within the steel beam range 2) For the equivalent temperature within the combined layer (7) Constructing vertical temperature gradient fatigue load model Using Matlab software, we can analyze the different positions y and different design service life N. d The equivalent temperature is fitted by the least square method, and the vertical temperature gradient fatigue load model of the long-life composite plate girder bridge is established as shown in formula (1).

Images