Temperature gradient fatigue load spectrum of concrete box girder bridge and construction method

Abstract

The invention discloses a temperature gradient fatigue load spectrum of a concrete box girder bridge and a construction method, and the temperature gradient fatigue load spectrum comprises a vertical temperature gradient fatigue load spectrum and a transverse temperature gradient fatigue load spectrum which are respectively composed of four temperature sub-gradients and occurrence probability rates thereof. Long-term monitoring data of a temperature field of the concrete box girder bridge are utilized, spatial and temporal distribution characteristics of the temperature field of the concrete box girder bridge are analyzed, and the construction method of the temperature gradient fatigue load spectrum of the concrete box girder bridge is provided with a K-median clustering method as the core. A temperature gradient fatigue load spectrum suitable for the concrete box girder bridge with the design service life of 100 years, 150 years and 200 years is constructed, the stress history of the concrete box girder bridge can be calculated, and the method can be used for long-life anti-fatigue design of the concrete box girder bridge.

Description

Technical Field

[0001] The invention belongs to the technical field of bridge engineering, and in particular relates to a temperature gradient fatigue load spectrum of a concrete box girder bridge and a construction method thereof. Background Art

[0002] During operation, concrete box girder bridges not only have to withstand vehicle loads, but also variable temperature loads. By analyzing the temperature field monitoring data of concrete box girder bridges, it can be seen that the temperature effects on concrete box girder bridges include the cyclic effects of daily temperature differences and seasonal temperature differences. Existing studies have shown that the fatigue damage caused by the coupling of temperature loads and vehicle loads to details is 7% to 15% higher than the fatigue damage caused by vehicle loads, and the fatigue damage caused by temperature and vehicle alone cannot be simply superimposed. Therefore, it is necessary to use the temperature field monitoring data of concrete box girder bridges to conduct fatigue damage analysis of concrete box girder bridges.

[0003] Concrete box girder bridges are typical closed section bridges with thick walls and a thermal conductivity coefficient that is quite different from that of steel structure bridges. Under the action of solar radiation, there are not only vertical or lateral temperature gradients along the height or width of the section, but also temperature gradients along the thickness of the box girder wall. Since the temperature effects on concrete box girder bridges include cyclic effects such as daily temperature difference and seasonal temperature difference, it is necessary to construct a temperature gradient fatigue load spectrum for concrete box girder bridges in order to calculate the temperature fatigue stress history of concrete box girder bridges and meet the needs of fatigue damage analysis. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a temperature gradient fatigue load spectrum of a concrete box girder bridge and a construction method thereof.

[0005] The technical solution adopted to solve the above technical problems is: a temperature gradient fatigue load spectrum of a concrete box girder bridge, which load spectrum includes a vertical temperature gradient fatigue load spectrum and a transverse temperature gradient fatigue load spectrum; the vertical temperature gradient fatigue load spectrum is composed of 4 vertical temperature sub-gradients and the probability of occurrence of the sub-gradients within the design service life, specifically:

[0006]

[0007] Where, T Gi (y) is the temperature of the ith vertical temperature sub-gradient at position y, H is the cross-sectional height of the concrete box girder bridge, Gi is the ith vertical temperature sub-gradient, T Gi,1 , T Gi,2 , T Gi,3 , T Gi,4 are the representative values ​​of the temperature at the top plate, at a height of 0.9H from the bottom plate, at a height of 1 / 6H from the bottom plate, and at the bottom plate in the i-th vertical temperature sub-gradient, respectively. G1,1 , TG1,2 , T G1,3 , T G1,4 are the representative values ​​of the temperature at the top plate, 0.9H height, 1 / 6H height, and bottom plate in the first vertical temperature sub-gradient, respectively. G1,1 The value range is [18.0,23.0], the unit is ℃, T G1,2 The value range is [14.0,18.0], the unit is ℃, T Gi,j is the temperature representative value at the jth typical height in the ith vertical temperature sub-gradient, N d is the design service life, T G1,j is the temperature representative value at the jth typical height in the first vertical temperature sub-gradient, P(T G1 (y))、P(T G2 (y))、P(T G3 (y))、P(T G4 (y)) are the occurrence probabilities of the four vertical temperature sub-gradients within the design service life, T G1,1 , T G2,1 , T G3,1 , T G4,1 They are the representative values ​​of temperature at the top plate of the four vertical temperature sub-gradients;

[0008] The transverse temperature gradient fatigue load spectrum is composed of four transverse temperature sub-gradients and the probability of occurrence of the sub-gradients within the design service life, specifically:

[0009]

[0010] Where Hq is the qth transverse temperature sub-gradient, T Hq (x) is the temperature of the qth transverse temperature sub-gradient at the transverse position x, B is the top plate width of the concrete box girder bridge, T Hq,w is the temperature representative value at the wth typical transverse position in the qth transverse temperature sub-gradient, T Hq,1 , T Hq,2 , T Hq,3 , T Hq,4 are the representative values ​​of the temperature at the edge of the sunny side, 1 / 2B away from the edge of the sunny side, 3 / 4B away from the edge of the sunny side, and the edge of the shady side in the qth transverse temperature sub-gradient, respectively. H1,1 The value range is [8.0,10.0], the unit is ℃, T H1,2 The value range is [3.4,5.4], the unit is ℃, T H1,w is the temperature representative value at the wth typical transverse position in the first transverse temperature sub-gradient, P(T H1 (y))、P(T H2 (y))、P(T H3 (y))、P(TH4 (y)) are the occurrence probabilities of the four transverse temperature sub-gradients within the design service life, T H1,1 , T H2,1 , T H3,1 , T H4,1 They are the representative values ​​of the temperature at the edge of the positive side in the four lateral temperature sub-gradients.

[0011] Preferably, the designed service life N d For 100 years or 150 years or 200 years.

[0012] The present invention also provides a method for constructing a temperature gradient fatigue load spectrum of a concrete box girder bridge, comprising the following steps:

[0013] Step 1: Collect temperature data of concrete box girder bridge

[0014] A total of m measuring points are arranged on the concrete box girder bridge. The temperature values ​​of each measuring point are collected at a certain interval to obtain the measured temperature gradient time history curve at each position. The measured temperature gradient time history curve is extended to the design service life N using the long short-term memory recursive neural network algorithm. d The temperature history at typical locations is used to obtain the maximum daily temperature at each measuring point every day within the design service life, and the design service life daily temperature extreme value matrix Q = [M1, M2, …, M t ,…,M Nd×365 ],M t is the daily temperature maximum vector of each measuring point on the tth day;

[0015] Step 2: Use K-median clustering algorithm to define the initial cluster center and cluster family

[0016] Step 2.1: Define the number of clusters c l , randomly select c from the daily temperature extreme value matrix Q of the design service life l vectors as initial cluster centers, and construct a cluster family with each initial cluster center, constructing a total of c l Cluster families;

[0017] Step 2.2, calculate the Euclidean distance between other vectors except the initial cluster center in the daily temperature extreme value matrix Q of the design service life and each initial cluster center, and classify the vector into the cluster group corresponding to the initial cluster center with the smallest Euclidean distance;

[0018] Step 3: Update cluster centers and cluster families

[0019] Step 3.1, calculate the Euclidean distance between the virtual new cluster center of the cluster family and other vectors in the cluster family and sum them up, select the virtual new cluster center with the smallest sum of Euclidean distances, if the virtual new cluster center is inconsistent with the initial cluster center, then the virtual new cluster center is determined as the new cluster center, otherwise the cluster center is not updated until c l The cluster centers are determined and c l The final cluster centers;

[0020] Step 3.2, from c l The updated final cluster centers are selected from the final cluster centers, the Euclidean distance between each vector in the daily temperature extreme value matrix Q of the design service life and each updated final cluster center is calculated, and the vector is classified into the cluster family corresponding to the updated cluster center with the smallest Euclidean distance;

[0021] Step 4: Determine the optimal number of clusters c l :

[0022] Let the number of clusters c l are integers from 2 to 10 respectively. Follow steps 2 and 3 to get the corresponding number of clusters c. l The 9 final clustering families are obtained. The cluster centers of each final clustering family constitute the cluster center matrix. The residual square sum RMSE of the maximum residual daily temperature is obtained according to the following formula for each cluster center matrix. l -RMSE two-dimensional line chart, based on the elbow rule, determine the optimal number of clusters c l The value is 4;

[0023]

[0024] In the formula, is the kth row vector in the cluster center matrix;

[0025] Step 5: Determine the temperature gradient fatigue load spectrum of the concrete box girder bridge

[0026] Use the number of clusters c determined in step 4 l And the corresponding cluster center matrix C end , each cluster family is a sub-gradient of the vertical temperature gradient fatigue load spectrum. The row vector of the cluster center matrix is ​​the typical representative value of each temperature sub-gradient. The number of sample points in each cluster family divided by the total number is the occurrence probability of each sub-gradient, thus obtaining the corresponding concrete box girder bridge temperature gradient fatigue load spectrum.

[0027] Preferably, the m measuring points in step 1 are used to construct a vertical temperature gradient fatigue load spectrum, and the arrangement method is: arrange a measuring point on the bottom plate of the concrete box girder bridge, and use the measuring point as the coordinate origin, bury measuring points on the web and the top plate along the vertical height direction, and arrange a total of m=8 measuring points. The positions of the measuring points are expressed as 0.00m, H-3.71m, H-2.40m, H-1.10m, H-0.70m, H-0.40m, H-0.20m, H in terms of the vertical distance from the lower surface of the bottom plate of the concrete box girder bridge.

[0028] Preferably, the m measuring points in step 1 are used to construct a transverse temperature gradient fatigue load spectrum, and the arrangement method is: a measuring point is arranged on the top plate of the concrete box girder bridge, and the measuring point is used as the coordinate origin, and measuring points are buried on the top plate along the transverse direction, and a total of m=5 measuring points are arranged. The positions of the measuring points are expressed as the horizontal distance from the concrete box girder bridge to the coordinate origin as 1.50m, L-8.75m, L-6.00m, L-3.25m, and L-1.50m.

[0029] Preferably, the temperature value collection interval of each measuring point in step 1 is 60 to 1200 seconds.

[0030] The beneficial effects of the present invention are as follows:

[0031] 1. The temperature gradient fatigue load spectrum of the concrete box girder bridge proposed in the present invention includes transverse and vertical temperature fatigue load spectra, both of which are composed of multiple temperature sub-gradients and the occurrence probability of sub-gradients, and can simulate the temperature stress history of the concrete box girder bridge within the design service life.

[0032] 2. The method for constructing the temperature gradient fatigue load spectrum of a concrete box girder bridge proposed in the present invention is based on the K-median clustering algorithm, which can efficiently perform cluster analysis on long-term temperature monitoring data of the temperature field to obtain different temperature sub-gradient models.

[0033] 3. The present invention uses long-term monitoring data of the temperature field of concrete box girder bridges to analyze the temporal and spatial distribution characteristics of the temperature field of concrete box girder bridges, ensures the accuracy and practicality of the load spectrum, and adopts a long short-term memory recursive neural network algorithm to expand the measured temperature gradient time history curve, thereby improving the accuracy and reliability of temperature prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is an elevation view of a concrete box girder bridge implemented in the present invention.

[0035] Figure 2 It is the vertical temperature sub-gradient model of concrete box girder bridge.

[0036] Figure 3 is the frequency of each vertical temperature sub-gradient of the concrete box girder bridge.

[0037] Figure 4 Construct a flow chart for the transverse and vertical temperature load spectra for a concrete box girder bridge.

[0038] Figure 5 This is a schematic diagram of the arrangement of transverse and vertical temperature measurement points for concrete box girder bridges.

[0039] Figure 6 This is the measured vertical temperature gradient history curve of the concrete box girder bridge.

[0040] Figure 7 c l -RMSE 2D line chart.

[0041] Figure 8 Cluster analysis results of vertical temperature data of concrete box girder bridge.

[0042] Fig. 9 This is the transverse temperature sub-gradient model of the concrete box girder bridge.

[0043] Fig.10 is the frequency of each temperature sub-gradient in the transverse direction of the concrete box girder bridge.

[0044] Fig.11 Cluster analysis results of transverse temperature data of concrete box girder bridge. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments, but the present invention is not limited to the following embodiments.

[0046] Example 1

[0047] This embodiment takes a concrete box continuous beam bridge with a beam height H of 4.40m in the northwest region as an example. The three-dimensional structure is shown in FIG. Figure 1 As shown in the figure, the temperature gradient fatigue load spectrum of the concrete box continuous beam bridge includes the vertical temperature gradient fatigue load spectrum and the transverse temperature gradient fatigue load spectrum; the vertical temperature gradient fatigue load spectrum is composed of four vertical temperature sub-gradients and the probability of occurrence of the sub-gradients within the design service life, specifically:

[0048]

[0049] Where, T Gi (y) is the temperature of the ith vertical temperature sub-gradient at position y, H is the cross-sectional height of the concrete box girder bridge, Gi is the ith vertical temperature sub-gradient, T Gi,1 , T Gi,2 , T Gi,3 , T Gi,4 are the representative values ​​of the temperature at the top plate, at a height of 0.9H from the bottom plate, at a height of 1 / 6H from the bottom plate, and at the bottom plate in the i-th vertical temperature sub-gradient, respectively. G1,1 , TG1,2 , T G1,3 , T G1,4 are the representative values ​​of the temperature at the top plate, 0.9H height, 1 / 6H height, and bottom plate in the first vertical temperature sub-gradient, respectively. G1,1 The value range is [18.0,23.0], the unit is ℃, T G1,2 The value range is [14.0,18.0], the unit is ℃, T Gi,j is the temperature representative value at the jth typical height in the ith vertical temperature sub-gradient, N d is the design service life, T G1,j is the temperature representative value at the jth typical height in the first vertical temperature sub-gradient, P(T G1 (y))、P(T G2 (y))、P(T G3 (y))、P(T G4 (y)) are the occurrence probabilities of the four vertical temperature sub-gradients within the design service life, T G1,1 , T G2,1 , T G3,1 , T G4,1 They are the representative values ​​of temperature at the top plate of the four vertical temperature sub-gradients;

[0050] In this embodiment, T G1,1 =21.1℃, T G1,2 =16.5℃, the temperature representative values ​​of each vertical temperature sub-gradient of the vertical temperature gradient fatigue load spectrum with a design service life of 100 years, 150 years, and 200 years are obtained from the above model, as shown in Table 1. The sub-gradient mode is as follows Figure 2 As shown in the figure, 150 years and 200 years are long-life design service life, and the vertical temperature gradient fatigue load spectrum gradient T of the concrete box continuous beam bridge is G1 ~T G4 The probability of occurrence is 3.6%, 12.5%, 33.9%, 50%, such as Figure 3 .

[0051] Table 1 Vertical temperature gradient fatigue load spectrum of concrete box girder Gradient temperature representative values

[0052]

[0053] exist Figure 4 In the embodiment, the method for constructing the vertical temperature gradient fatigue load spectrum of the concrete box girder bridge comprises the following steps:

[0054] Step 1: Figure 5As shown, a measuring point is arranged on the bottom plate of the concrete box girder bridge, and the measuring point is used as the coordinate origin. Measuring points are buried on the web and the top plate along the vertical height direction. A total of m=8 measuring points are arranged, and the vertical distance from the lower surface of the bottom plate of the concrete box girder is expressed as y. The positions of different measuring points can be recorded as 0.00m, 0.50m, 1.81m, 3.12m, 3.52m, 3.82m, 4.02m, and 4.22m, respectively. Long-term temperature collection is carried out on these measuring points to obtain valid temperature data for the most recent year, and the collection interval is 1 second. In this embodiment, the collection interval can also be 20 seconds, or 1800 seconds.

[0055] The measured temperature gradient time history curve is obtained based on the collected temperature data, such as Figure 6 The measured temperature gradient time history curve is extended to the design service life N using the long short-term memory recursive neural network algorithm. d The temperature history at typical locations is used to obtain the maximum daily temperature at each measuring point every day within the design service life, and the design service life daily temperature extreme value matrix Q = [M1, M2, …, M t ,…,M Nd×365 ],M t is the daily temperature maximum vector of each measuring point on the tth day;

[0056] Step 2: Use K-median clustering algorithm to define the initial cluster center and cluster family

[0057] Step 2.1: Define the number of clusters c l , randomly select c from the daily temperature extreme value matrix Q of the design service life l vectors as initial cluster centers, and construct a cluster family with each initial cluster center, constructing a total of c l Cluster families;

[0058] Step 2.2, calculate the Euclidean distance between other vectors except the initial cluster center in the daily temperature extreme value matrix Q of the design service life and each initial cluster center, and classify the vector into the cluster group corresponding to the initial cluster center with the smallest Euclidean distance;

[0059] Step 3: Update cluster centers and cluster families

[0060] Step 3.1: All vectors in the same cluster family are taken as the virtual new cluster center of the cluster family in turn, and the Euclidean distances between the virtual new cluster center of the cluster family and other vectors in the cluster family are calculated and summed up, and the virtual new cluster center with the smallest sum of Euclidean distances is selected. If the virtual new cluster center is inconsistent with the initial cluster center, the virtual new cluster center is determined as the new cluster center, otherwise the cluster center is not updated until c l The cluster centers are determined and c l The final cluster centers;

[0061] Step 3.2, from c l The updated final cluster centers are selected from the final cluster centers, the Euclidean distance between each vector in the daily temperature extreme value matrix Q of the design service life and each updated final cluster center is calculated, and the vector is classified into the cluster family corresponding to the updated cluster center with the smallest Euclidean distance;

[0062] Step 4: Determine the optimal number of clusters c l :

[0063] Let the number of clusters c l are integers from 2 to 10 respectively. Follow steps 2 and 3 to get the corresponding number of clusters c. l The 9 final clustering families are obtained. The cluster centers of each final clustering family constitute the cluster center matrix. The residual square sum RMSE of the maximum residual daily temperature is obtained according to the following formula for each cluster center matrix. l -RMSE two-dimensional line chart, such as Figure 7 As shown, according to the elbow rule, the optimal number of clusters cl is determined to be 4, as shown in Figure 8 As shown;

[0064]

[0065] In the formula, is the kth row vector in the cluster center matrix;

[0066] Step 5: Determine the temperature gradient fatigue load spectrum of the concrete box girder bridge

[0067] Use the number of clusters cl determined in step 4 and the corresponding cluster center matrix C end , each cluster family is a sub-gradient of the vertical temperature gradient fatigue load spectrum. The row vector of the cluster center matrix is ​​the typical representative value of each temperature sub-gradient. The number of sample points in each cluster family divided by the total number is the occurrence probability of each sub-gradient, thus obtaining the corresponding concrete box girder bridge temperature gradient fatigue load spectrum.

[0068] The transverse temperature gradient fatigue load spectrum is composed of four transverse temperature sub-gradients and the probability of occurrence of the sub-gradients within the design service life, specifically:

[0069]

[0070] Where Hq is the qth transverse temperature sub-gradient, T Hq (x) is the temperature of the qth transverse temperature sub-gradient at the transverse position x, B is the top plate width of the concrete box girder bridge, T Hq,w is the temperature representative value at the wth typical transverse position in the qth transverse temperature sub-gradient, T Hq,1 , THq,2 , T Hq,3 , T Hq,4 are the representative values ​​of the temperature at the edge of the sunny side, 1 / 2B away from the edge of the sunny side, 3 / 4B away from the edge of the sunny side, and the edge of the shady side in the qth transverse temperature sub-gradient, respectively. H1,1 The value range is [8.0,10.0], the unit is ℃, T H1,2 The value range is [3.4,5.4], the unit is ℃, T H1,w is the temperature representative value at the wth typical transverse position in the first transverse temperature sub-gradient, P(T H1 (y))、P(T H2 (y))、P(T H3 (y))、P(T H4 (y)) are the occurrence probabilities of the four transverse temperature sub-gradients within the design service life, T H1,1 , T H2,1 , T H3,1 , T H4,1 They are the representative values ​​of the temperature at the edge of the positive side in the four lateral temperature sub-gradients.

[0071] In this embodiment, T H1,1 =8.9℃, T H1,2 =4.3℃, the temperature gradient fatigue load spectra constructed according to the design service life of 100 years, 150 years and 200 years are shown in Table 2. The lateral gradient mode is as follows Fig. 9 As shown in Figure 1, 150 and 200 years are long-life design service life, and the sub-gradient T H1 ~T H4 The probability of occurrence is 12.3%, 40.7%, 28.4%, and 18.5%. Fig.10 .

[0072] Table 2 Representative values ​​of gradient temperature of fatigue load spectrum of transverse temperature gradient of concrete box girder

[0073]

[0074] The method for constructing a transverse temperature gradient fatigue load spectrum of a concrete box girder bridge in this embodiment includes the following steps:

[0075] Step 1: Arrange a measuring point at the edge of the positive side of the roof, and use this measuring point as the coordinate origin. Arrange m = 5 measuring points along the width of the roof. The measuring point positions are expressed as 1.50m, 3.25m, 6.00m, 8.75m, and 10.50m from the coordinate origin. Figure 5 .

[0076] Steps 2 to 5 are the same as the method for constructing the vertical temperature gradient fatigue load spectrum of the concrete box girder bridge. The clustering results are as follows: Fig.11 shown.

Claims

1. A temperature gradient fatigue load spectrum for a concrete box girder bridge, characterized in that: The load spectrum includes a vertical temperature gradient fatigue load spectrum and a transverse temperature gradient fatigue load spectrum; the vertical temperature gradient fatigue load spectrum is composed of four vertical temperature sub-gradients and the probability of occurrence of the sub-gradients within the design service life, specifically: Where, T Gi (y) is the temperature of the ith vertical temperature sub-gradient at position y, H is the cross-sectional height of the concrete box girder bridge, Gi is the ith vertical temperature sub-gradient, T Gi,1 、T Gi,2 、T Gi,3 、T Gi,4 are the representative values ​​of the temperature at the top plate, at a height of 0.9H from the bottom plate, at a height of 1 / 6H from the bottom plate, and at the bottom plate in the i-th vertical temperature sub-gradient, respectively. G1,1 、T G1,2 、T G1,3 、T G1,4 are the representative values ​​of the temperature at the top plate, 0.9H height, 1 / 6H height, and bottom plate in the first vertical temperature sub-gradient, respectively. G1,1 The value range is [18.0,23.0], the unit is ℃, T G1,2 The value range is [14.0,18.0], the unit is ℃, T Gi,j is the temperature representative value at the jth typical height in the ith vertical temperature sub-gradient, N d is the design service life, T G1,j is the temperature representative value at the jth typical height in the first vertical temperature sub-gradient, P(T G1 (y))、P(T G2 (y))、P(T G3 (y))、P(T G4 (y)) are the occurrence probabilities of the four vertical temperature sub-gradients within the design service life, T G1,1 、T G2,1 、T G3,1 、T G4,1 They are the representative values ​​of temperature at the top plate of the four vertical temperature sub-gradients; The transverse temperature gradient fatigue load spectrum is composed of four transverse temperature sub-gradients and the probability of occurrence of the sub-gradients within the design service life, specifically: Where Hq is the qth transverse temperature sub-gradient, T Hq (x) is the temperature of the qth transverse temperature sub-gradient at the transverse position x, B is the top plate width of the concrete box girder bridge, T Hq,w is the temperature representative value at the wth typical transverse position in the qth transverse temperature sub-gradient, T Hq,1 、T Hq,2 、T Hq,3 、T Hq,4 are the representative values ​​of the temperature at the edge of the sunny side, 1 / 2B away from the edge of the sunny side, 3 / 4B away from the edge of the sunny side, and the edge of the shady side in the qth transverse temperature sub-gradient, respectively. H1,1 The value range is [8.0,10.0], the unit is ℃, T H1,2 The value range is [3.4,5.4], the unit is ℃, T H1,w is the temperature representative value at the wth typical transverse position in the first transverse temperature sub-gradient, P(T H1 (y))、P(T H2 (y))、P(T H3 (y))、P(T H4 (y)) are the occurrence probabilities of the four transverse temperature sub-gradients within the design service life, T H1,1 、T H2,1 、T H3,1 、T H4,1 They are the representative values ​​of the temperature at the edge of the positive side in the four lateral temperature sub-gradients.

2. The temperature gradient fatigue load spectrum of the concrete box girder bridge according to claim 1 is characterized in that: The design service life N d For 100 years or 150 years or 200 years.

3. The method for constructing the temperature gradient fatigue load spectrum of a concrete box girder bridge according to claim 1 is characterized in that: The following steps are involved: Step 1: Collect temperature data of concrete box girder bridge A total of m measuring points are arranged on the concrete box girder bridge. The temperature values ​​of each measuring point are collected at a certain interval to obtain the measured temperature gradient time history curve at each position. The measured temperature gradient time history curve is extended to the design service life N using the long short-term memory recursive neural network algorithm. d The temperature history at typical locations is used to obtain the maximum daily temperature at each measuring point every day within the design service life, and the design service life daily temperature extreme value matrix Q = [M1, M2, …, M t ,…,M Nd×365 ],M t is the daily temperature maximum vector of each measuring point on the tth day; Step 2: Use K-median clustering algorithm to define the initial cluster center and cluster family Step 2.1: Define the number of clusters c l , randomly select c from the daily temperature extreme value matrix Q of the design service life l vectors as initial cluster centers, and construct a cluster family with each initial cluster center, constructing a total of c l Cluster families; Step 2.2, calculate the Euclidean distance between other vectors except the initial cluster center in the daily temperature extreme value matrix Q of the design service life and each initial cluster center, and classify the vector into the cluster group corresponding to the initial cluster center with the smallest Euclidean distance; Step 3: Update cluster centers and cluster families Step 3.1, calculate the Euclidean distance between the virtual new cluster center of the cluster family and other vectors in the cluster family and sum them up, select the virtual new cluster center with the smallest sum of Euclidean distances, if the virtual new cluster center is inconsistent with the initial cluster center, then the virtual new cluster center is determined as the new cluster center, otherwise the cluster center is not updated until c l The cluster centers are determined and c l The final cluster centers; Step 3.2, from c l The updated final cluster centers are selected from the final cluster centers, the Euclidean distance between each vector in the daily temperature extreme value matrix Q of the design service life and each updated final cluster center is calculated, and the vector is classified into the cluster family corresponding to the updated cluster center with the smallest Euclidean distance; Step 4: Determine the optimal number of clusters c l : Let the number of clusters c l are integers from 2 to 10 respectively. Follow steps 2 and 3 to get the corresponding number of clusters c. l The 9 final clustering families are obtained. The cluster centers of each final clustering family constitute the cluster center matrix. The residual square sum RMSE of the maximum residual daily temperature is obtained according to the following formula for each cluster center matrix. l -RMSE two-dimensional line chart, based on the elbow rule, determine the optimal number of clusters c l The value is 4; In the formula, is the kth row vector in the cluster center matrix; Step 5: Determine the temperature gradient fatigue load spectrum of the concrete box girder bridge Use the number of clusters c determined in step 4 l And the corresponding cluster center matrix C end , each cluster family is a sub-gradient of the vertical temperature gradient fatigue load spectrum. The row vector of the cluster center matrix is ​​the typical representative value of each temperature sub-gradient. The number of sample points in each cluster family divided by the total number is the occurrence probability of each sub-gradient, thus obtaining the corresponding concrete box girder bridge temperature gradient fatigue load spectrum.

4. The method for constructing the temperature gradient fatigue load spectrum of a concrete box girder bridge according to claim 3 is characterized in that: The m measuring points described in step 1 are used to construct the vertical temperature gradient fatigue load spectrum. The arrangement method is: arrange a measuring point on the bottom plate of the concrete box girder bridge, and take the measuring point as the coordinate origin, bury measuring points on the web and the top plate along the vertical height direction, and arrange m=8 measuring points in total. The positions of the measuring points are expressed as 0.00m, H-3.71m, H-2.40m, H-1.10m, H-0.70m, H-0.40m, H-0.20m, H in terms of the vertical distance from the lower surface of the bottom plate of the concrete box girder bridge.

5. The method for constructing the transverse temperature gradient fatigue load spectrum of a concrete box girder bridge according to claim 3 is characterized in that: The m measuring points described in step 1 are used to construct the transverse temperature gradient fatigue load spectrum. The arrangement method is: arrange a measuring point on the top plate of the concrete box girder bridge, and use the measuring point as the coordinate origin, bury measuring points on the top plate along the transverse direction, and arrange m=5 measuring points in total. The positions of the measuring points are expressed as 1.50m, L-8.75m, L-6.00m, L-3.25m, and L-1.50m in terms of the horizontal distance from the concrete box girder bridge to the coordinate origin.

6. The method for constructing the temperature gradient fatigue load spectrum of a concrete box girder bridge according to claim 3 is characterized in that: The temperature value collection interval of each measuring point in step 1 is 60 to 1200 seconds.

Images