Trestle stability analysis method based on deep neural network

Abstract

The invention provides a trestle stability analysis method based on a deep neural network. The method comprises the following steps: collecting settlement data of trestle steel pipe columns in real time; J, calculating a differential settlement index data set of the steel pipe columns at each moment and differential settlement change variance of all the steel pipe columns in unit collection time; collecting stress data samples of the steel pipe columns in real time, and calculating a stress characteristic index data set of all the steel pipe columns and an overall stress distribution characteristic index M of the steel pipe columns; and calculating a stability coefficient U of the construction temporary bridge steel pipe stand column by combining the indexes P and M, collecting bridge floor vertical vibration displacement time history data in real time to calculate a bridge floor vibration fluctuation index C, fitting the indexes U and C through a binary higher-degree polynomial, and establishing a parameter-containing model of the overall stability coefficient of the construction temporary bridge. Influences of factors such as stress of the steel pipe columns, differential settlement of the steel pipe columns and bridge floor vibration on stability are comprehensively considered, and the stability of the construction temporary bridge under the long-term use condition can be evaluated.

Description

Technical Field

[0001] The present invention relates to the field of service performance analysis of construction temporary bridges, and in particular to a trestle stability analysis method based on a deep neural network. Background Art

[0002] Construction bridges are temporary structures commonly used in engineering projects. They are usually composed of steel structures such as steel pipe columns, load-bearing beams, Bailey beams, and distribution beams. Distribution beams and steel plates are then laid on top to serve as a passage for the transportation and movement of construction vehicles, workers, and equipment. In previous projects, due to the short service life of construction bridges, low frequency of use, and small loads, their stability and service life were rarely considered.

[0003] With the full launch of a large number of major projects in my country, construction bridges are facing a series of problems such as long service life, extremely harsh surrounding environment, huge project scale and complex load bearing. The stability, safety and service life of construction bridge structures are closely related. Therefore, stability monitoring and analysis of construction bridges is one of the problems that need to be solved urgently.

[0004] The service performance of the construction bridge depends not only on the force of the steel pipe column, but also on factors such as uneven settlement of the steel pipe column and change in the overall stiffness of the construction bridge. Among them, uneven settlement is one of the key indicators affecting the stability of the construction bridge. After the construction bridge is built, the stratum of the trestle foundation will produce uneven settlement under the load of construction machinery and equipment, the deadweight of the trestle, and then the various steel pipe supports will also settle synchronously. As the use time increases, the accumulated uneven settlement will gradually increase, thus affecting the stability of the steel trestle. At the same time, the stress of the steel pipe column is also an important indicator for evaluating the stability of the construction bridge. When the steel structure trestle is completed and enters the use stage, the random stacking position of the construction machinery and equipment, the walking route of the transport vehicle, etc. may cause the construction bridge to produce eccentric loads, resulting in increased stress in the local steel pipe columns. In addition, during long-term use, some steel pipe columns may also suffer fatigue damage due to the repeated action of moving loads such as construction vehicles, resulting in further increase in the stress of local steel pipe columns and even strength damage. In addition, under the action of moving loads such as vehicles and pedestrians, the deck of the construction bridge will produce vertical vibrations, which can easily cause problems such as loose connections and bridge deck fatigue, resulting in a decrease in the overall stiffness of the construction bridge, thereby further increasing the amplitude of the vertical vibration of the bridge deck. Therefore, the vertical displacement of the deck of the construction bridge is also one of the key indicators for evaluating the stability of the construction bridge.

[0005] At present, the analysis and evaluation method of the service performance of construction bridges is to install stress sensors at key locations of the bridges to monitor the stress in real time, and to stop work for maintenance when the monitored stress exceeds the finite element calculation value or the allowable value of the specification. However, this traditional evaluation method does not take into account the influence of factors such as uneven settlement of steel pipe columns, changes in the overall stiffness of the construction bridge, and vertical displacement of the bridge deck. Therefore, it cannot accurately reflect the service performance of the construction bridge and its deterioration over time, resulting in evaluation failure and safety risks. Therefore, it is very necessary to carry out service performance analysis and evaluation of construction bridges for major projects. Summary of the invention

[0006] In response to the above problems, the present invention provides a trestle stability analysis method based on deep neural network. This method can conveniently and quickly evaluate the stability performance of the construction trestles within unit acquisition time and the stability performance degradation trend during long-term use by analyzing the overall stability coefficient of the construction trestles.

[0007] In order to achieve the above technical objectives, the present invention provides a trestle stability analysis method based on a deep neural network, which specifically includes the following steps:

[0008] S1. Stress sensors and settlement monitoring sensors are installed on each steel pipe column of the temporary bridge under construction, and displacement sensors are installed under the bridge deck steel plate in the middle of each span; the steel pipe columns and settlement monitoring sensors are numbered as 1, 2, 3...k, the displacement sensors are numbered as 1, 2, 3...m, and the stress sensors are numbered as 1, 2, 3...s;

[0009] S2. By collecting the settlement data J of all steel pipe columns in real time, and calculating the uneven settlement index data set D of the steel pipe columns at each moment, and the variance P of the uneven settlement change of all steel pipe columns within the unit collection time, the calculated uneven settlement change variance P is finally used as the uneven settlement index of the steel pipe columns of the temporary bridge; the calculation process of the uneven settlement change variance P is as follows:

[0010]

[0011] Where: n is the total number of time steps;

[0012] D i is the settlement variation variance of all steel tube columns at any i-th time step within n time steps;

[0013] It is the average value of the uneven settlement index of the steel pipe columns of the temporary bridge under construction within the unit collection time;

[0014] i is the time step number; k is the total number of settlement sensors;

[0015] j i,sis the settlement data of the sth settlement sensor at the ith time step;

[0016] is the average settlement data of all k settlement sensors at the i-th time step;

[0017] S3. By collecting the stress data sample F of the steel pipe column in real time, the stress characteristic index data set E of all the steel pipe columns is calculated, and then the overall force distribution characteristic index M of the steel pipe column of the temporary bridge is calculated by the stress characteristic index data set E. The calculation formula of the overall force distribution characteristic index M of the steel pipe column of the temporary bridge is as follows:

[0018]

[0019] Where: M is the overall force distribution characteristic index of the steel pipe column of the temporary bridge under construction;

[0020] E max is the maximum column stress characteristic index in data set E;

[0021] E min is the smallest column stress characteristic index in data set E;

[0022] S4. According to the uneven settlement index P of the steel pipe column of the temporary bridge calculated in step S2, combined with the stress distribution characteristic index M of the steel pipe column of the temporary bridge calculated in step S3, the stability coefficient U of the steel pipe column of the temporary bridge is calculated:

[0023] U = P × M;

[0024] S5. Collect the time history data of vertical vibration displacement of the bridge deck in real time, calculate the vertical vibration displacement change index set of each position of the bridge deck within the unit collection time, and calculate the bridge deck vibration fluctuation index C by collecting the vertical vibration displacement change index set T of each position of the bridge deck; the calculation formula is as follows:

[0025]

[0026] Where: T x is the vertical displacement variance of the xth displacement sensor position of the temporary bridge under construction,

[0027] is the average value of the variance of vertical displacement changes at all displacement sensor locations of the temporary bridge under construction;

[0028] m is the displacement sensor serial number;

[0029] x is the serial number of any displacement sensor among the m displacement sensors;

[0030] T xis the variance of the vertical displacement change at the xth displacement sensor position of the temporary bridge under construction;

[0031] S6. According to the instability index U of the steel pipe column obtained in step S4 and the vibration displacement fluctuation index C of the temporary bridge deck obtained in step S5, a polynomial fitting model of the overall stability coefficient Q of the temporary bridge is established:

[0032]

[0033] Where: Q is the fitting value of the overall stability coefficient of the construction temporary bridge, which is calculated from the instability index U of the steel pipe column and the vibration displacement fluctuation index C of the bridge deck;

[0034] r is the highest power of variable U in the fitting formula, and v is the highest power of variable C in the fitting formula. To obtain sufficient fitting accuracy, it is recommended that r ≥ 6 and v ≥ 6 in the formula;

[0035] a ij are the polynomial coefficients;

[0036] S7. Determine a based on the overall stability coefficient sample of the construction temporary bridge obtained by simulation ij , and substitute it into the formula of step S6 to finally obtain the calculation formula of the overall stability coefficient Q of the construction temporary bridge; substitute the real-time monitored steel tube column instability index U and bridge deck vibration displacement fluctuation index C into the final calculation formula of the overall stability coefficient Q of the construction temporary bridge to analyze the overall stability of the construction temporary bridge; when the steel tube column instability index U and the bridge deck vibration displacement fluctuation index C are larger, the overall stability coefficient of the construction temporary bridge is smaller, and the structure is more likely to become unstable and damaged.

[0037] A further technical solution of the present invention also includes step S8, which uses a deep learning method to analyze the laws from a large amount of monitored time-varying data based on real-time measurement of uneven settlement, uneven stress and bridge deck vibration of the construction temporary bridge columns, and predicts the deterioration trend of the stability of the construction temporary bridge within a certain period of time, so as to carry out emergency treatment in a timely manner.

[0038] A better technical solution of the present invention: in the step S1, two stress sensors and one settlement monitoring sensor are arranged on each steel pipe column of the temporary bridge under construction, the three sensors are arranged along the circumference of the steel pipe column, and the two stress sensors are symmetrically arranged; and a displacement sensor is arranged at the lower part of the bridge deck steel plate in the middle of each span; if it is assumed that the temporary bridge under construction has a total of m spans and k steel pipe columns, then 2*k stress sensors need to be arranged, k settlement monitoring sensors need to be arranged, and m displacement sensors need to be arranged; multiple sensors are connected to the matching settlement monitor for data collection.

[0039] The preferred technical solution of the present invention: The specific process of step S2 is as follows:

[0040] S2(1). The settlement of all steel pipe columns during the construction process is monitored by buried settlement monitoring sensors, and settlement data is collected in real time; the settlement monitoring sensors transmit the settlement data signals collected in real time to the settlement monitoring instrument, and the settlement monitoring instrument collects the settlement data collected by each settlement monitoring sensor, and obtains the settlement data samples J of all steel pipe columns:

[0041] J=[J1 J2...J i ...J n ]

[0042] J i =[j i，1 j i，2 … i，s … i，k ]

[0043] Where: i is the time step number, n is the total number of time steps;

[0044] k is the serial number of the sedimentation sensor;

[0045] J i is the settlement data set of all settlement sensors at any i time step within n time steps; J n is the settlement data set of all settlement sensors at the nth (i.e. last) time step;

[0046] j i,s is the settlement data of the sth settlement sensor at the i-th time step;

[0047] j i,k is the settlement data of the kth (i.e. the last) settlement sensor at the i-th time step; assuming that the sampling frequency of the settlement monitor is 10 s each time, and 30 minutes is a sampling time unit, the total number of time steps n = 180, i = 1...n;

[0048] S2 (2) Calculate the uneven settlement index data set D of the steel pipe column at each moment; extract the settlement data collected by the settlement monitor at each moment, and calculate the uneven settlement index data set D of the steel pipe column at each moment by the following formula;

[0049] D=[D1 D2...D i ...D n ]

[0050]

[0051]

[0052] Where: i is the time step number; k is the total number of settlement sensors;

[0053] j i,s is the settlement data of the sth settlement sensor at the ith time step;

[0054] is the average settlement data of all k settlement sensors at the i-th time step;

[0055] D i is the settlement variation variance of all steel tube columns at any i-th time step within n time steps, which is used to evaluate the uneven settlement degree of all steel tube columns of the temporary bridge under construction at the i-th time step;

[0056] D n is the settlement variation variance of all steel tube columns at the nth (i.e. the last) time step, which is used to evaluate the uneven settlement degree of all steel tube columns of the temporary bridge under construction at the last moment;

[0057] S2(3). The settlement variance D of all steel pipe columns calculated in step S2(2) at the i-th time step i Calculate the variance P of uneven settlement changes of all steel pipe columns within a unit collection time; the variance P of uneven settlement changes of steel pipe columns within a unit collection time calculated represents the settlement data changes of all steel pipe columns of the temporary bridge under construction within a unit collection time. The larger the index value, the greater the uneven settlement of the steel pipe columns of the temporary bridge under construction, that is, an uneven settlement index of the steel pipe columns of the temporary bridge under construction is obtained;

[0058] The preferred technical solution of the present invention: The specific process of step S3 is as follows:

[0059] S3(1). Two stress sensors are arranged on each steel pipe column to collect stress data of each steel pipe column in real time, and the stress data sample F of all steel pipe columns is obtained:

[0060] F=[F1 F2...F i ...F n ]

[0061] F i =[f i,1 f i,2 ...f i,s ...f i,k ]

[0062]

[0063] Where: F i is the axial stress data set of all steel pipe columns at any i-th time step within n time steps;

[0064] F nis the axial stress data set of all steel tube columns at the nth (i.e. the last) time step;

[0065] f i,s is the axial stress data of the s-th steel pipe column at any i-th time step within n time steps;

[0066] f i,k is the axial stress data of the kth (i.e. the last) steel pipe column at any i-th time step within n time steps;

[0067] and They are the stress data monitored by two stress sensors arranged on the s-th steel pipe column;

[0068] The data sampling frequency of the stress sensor is 10 seconds each time, with 30 minutes as a time unit, to collect the stress values ​​of each steel pipe column of the temporary bridge under construction;

[0069] S3(2). According to the obtained stress data sample F, determine the maximum stress f in each steel pipe column stress data sample per unit time. max With the minimum stress f min The difference between them is combined with the stress variance of each steel tube column to calculate the stress characteristic index data set E of all steel tube columns;

[0070] E=[E1...E s ...E k ]

[0071]

[0072] Where: f i,s is the axial stress data of the s-th steel pipe column at the i-th time step;

[0073] f max is the maximum stress value within the unit acquisition time;

[0074] f min is the minimum stress value per unit acquisition time;

[0075] E s is the stress characteristic index of the s-th steel pipe column;

[0076] E k It is the stress characteristic index of the last steel pipe column;

[0077] The calculated stress characteristic index E of the s-th steel pipe column sIt is used to evaluate the stress change of the sth steel pipe column of the temporary bridge under construction within the unit collection time. By comparing the extreme values, it reflects the unevenness of the stress distribution of different steel pipe columns. Combined with the mean stress value of each steel pipe column, it reflects the load size borne by each bracket.

[0078] S3(3). Calculate the maximum stress characteristic index E among all stress characteristic indexes of steel tube columns through the stress characteristic index data set of steel tube columns. max and the minimum stress characteristic index E min The difference between them is combined with the variance of the stress characteristic indexes of all the supports to obtain the overall force distribution characteristic index M of the steel pipe columns of the temporary bridge. M is used to evaluate the force distribution of the overall steel pipe columns of the temporary bridge. The larger the range and mean, the more uneven the force distribution of the steel pipe columns of the temporary bridge, the greater the load, and the worse the stability of the steel pipe columns of the temporary bridge. Therefore, the force distribution characteristic index M of the steel pipe columns of the temporary bridge is obtained.

[0079] The preferred technical solution of the present invention: The specific process of step S5 is as follows:

[0080] S5(1). Use the displacement sensor to read the vertical vibration displacement time history data of the bridge deck; the displacement sensor sampling frequency is 10s each time, with 30 minutes as a time unit, and the vertical vibration displacement data in each unit time is collected, which can be obtained from the construction temporary bridge deck vibration amplitude sample set Z:

[0081] Z=[Z1...Z i ...Z n ]

[0082] Z i =[z i,1 ...z i,x ...z i,m ]

[0083] Where: n is the total number of time steps; i is the time step sequence number, i = 1...n;

[0084] m is the displacement sensor serial number; x is any displacement sensor among the m displacement sensors;

[0085] Z i is the displacement data set of all displacement sensors at any i-th time step in n time steps;

[0086] Z n is the displacement data set of all displacement sensors at the nth time step;

[0087] z i、x is the displacement data of the x-th displacement sensor at any i-th time step in n time steps;

[0088] z i、m is the displacement data of the mth (i.e. the last) displacement sensor at any i-th time step in n time steps;

[0089] S5(2). Calculate the vertical displacement variation variance of each displacement sensor position within a unit acquisition time based on the displacement data collected in S5(1), that is, the vertical vibration displacement variation index set T of each position on the bridge deck within a unit acquisition time:

[0090] T=[T1...T x ...T m ]

[0091]

[0092]

[0093] Where: n is the total number of time steps; x is the serial number of any displacement sensor among the m displacement sensors;

[0094] is the average displacement data of the xth displacement sensor within the unit acquisition time;

[0095] T x is the variance of the vertical displacement change at the xth displacement sensor position of the temporary bridge under construction;

[0096] T m is the vertical displacement variance of the mth (last) displacement sensor position of the temporary bridge under construction; T x It is used to evaluate the change of vertical displacement of the xth bridge deck position within the acquisition unit time;

[0097] S5(3). Calculate the bridge deck vibration fluctuation index C through the vertical vibration displacement change index set T of each bridge deck position within the unit collection time calculated in step S5(2); with the increase of service time, the overall stiffness of the construction temporary bridge decreases, and the vertical vibration displacement under the action of the moving load also increases accordingly. C can evaluate the overall vibration displacement change of the construction temporary bridge. When the bridge deck vibration displacement increases, the construction temporary bridge vibration fluctuation index C also increases, indicating that the stability of the construction temporary bridge is reduced. Thus, the bridge deck vibration displacement fluctuation index C of the trestle structure is obtained.

[0098] The preferred technical solution of the present invention is: in the step S6, the overall stability coefficient sample of the construction temporary bridge obtained by simulation is used to determine a ij , the specific calculation method is as follows:

[0099] a. Based on a given set of steel pipe column settlement data J * and stress data F *, a refined model of the temporary bridge under construction considering the initial uneven settlement and uneven stress distribution is established in the finite element software; then, according to another set of given bridge deck vertical displacement change indicators T * , the corresponding dynamic load is applied to the construction bridge model (based on the principle of the same maximum vertical displacement), and finally the design load is applied to the bridge model in multiples until the overall structure becomes unstable, thereby obtaining the overall stability coefficient Q of the construction bridge * ;

[0100] b. Based on the data J given in step a * and data F * , calculate the corresponding steel pipe column instability index U according to step S4 * ; According to the bridge deck vertical displacement change index T given in step a * , calculate the bridge deck vibration displacement fluctuation index C according to step S5 * ;

[0101] c. Repeat steps a and b at least (r+1)*(v+1) times to obtain a series of data and

[0102] d. Q * , U * and C * Substitute into the parameter-containing model in step S6(1), and determine the coefficients of each polynomial in the model by solving the overdetermined equations The calculation formula for the overall stability coefficient Q of the construction temporary bridge is finally obtained. When the instability index U of the steel tube column and the vibration displacement fluctuation index C of the bridge deck are larger, the overall stability coefficient of the construction temporary bridge is smaller, and the structure is more likely to become unstable and damaged.

[0103] The preferred technical solution of the present invention: The specific process of step S8 is as follows:

[0104] S8(1). The failure threshold of the overall stability coefficient Q of the temporary bridge is set. Generally, the elastic-plastic stability bearing capacity of the structure is considered to be twice the design load. Therefore, it is recommended that the failure threshold of the overall stability coefficient Q of the temporary bridge be 2;

[0105] S8(2). Based on the prediction of stability degradation trend of neural network, repeat the above steps S2 to S5, and use the formula in step S6 to calculate the stability coefficient of the construction bridge for multiple acquisition units, and obtain the time-based change sequence of the stability coefficient of the construction bridge Q = [Q1, Q2, ..., Q i};

[0106] S8(3). Use the time convolution neural network (TCN) to predict the stability of the temporary bridge under construction, and collect the stability change sequence of the temporary bridge under construction {Q1, Q2, …, Q i-1} as samples, and use the overall structural stability coefficient of the temporary bridge at time i as output data for model training until the structural stability coefficient at time i is accurately predicted;

[0107] S8(4). Evaluation of the stability degradation trend of the construction bridge. When the stability coefficient of the construction bridge predicted by the time convolution (TCN) neural network is less than the threshold, the construction bridge has suffered major fatigue damage, uneven settlement, loose connection and other problems. The construction bridge needs to be repaired and inspected immediately, or the construction should be stopped.

[0108] The present invention comprehensively considers three factors of uneven settlement of steel pipe columns of a temporary bridge under construction, uneven stress on steel pipe columns, and vibration displacement of the bridge deck, designs an uneven settlement index P of steel pipe columns, an overall stress distribution characteristic index M of steel pipe columns of the temporary bridge under construction, and a vibration fluctuation index C of the bridge deck, and proposes an overall stability coefficient Q of the temporary bridge under construction. By analyzing the overall stability coefficient of the temporary bridge under construction, the stability coefficient of the temporary bridge under construction within a unit acquisition time and the stability degradation trend during long-term use can be conveniently and quickly evaluated. At the same time, the stability of the temporary bridge under construction can be predicted in combination with technologies such as neural networks, thereby solving the problem that the existing evaluation method cannot comprehensively consider the influence of factors such as stress on steel pipe columns, uneven settlement of steel pipe columns, and vibration of the bridge deck on stability, and cannot evaluate the stability degradation trend of the temporary bridge under long-term use conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0109] Figure 1 is a schematic diagram of the installation of sensors on a temporary bridge in an embodiment;

[0110] In the figure: 1—displacement sensor, 2—bridge deck, 3—settlement monitoring sensor, 4—stress sensor, 5—steel pipe column. DETAILED DESCRIPTION

[0111] The present invention will be further described below in conjunction with the accompanying drawings and embodiments. Figure 1 The accompanying drawings of the embodiments are drawn in a simplified manner and are only used to clearly and concisely illustrate the embodiments of the present invention. The technical solutions shown in the following drawings are specific solutions of the embodiments of the present invention and are not intended to limit the scope of the invention claimed for protection. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0112] The embodiment provides a method for analyzing the stability of a trestle bridge based on a deep neural network, and the specific steps are as follows:

[0113] S1. The layout of monitoring points for construction temporary bridges Figure 1 As shown, two stress sensors 4 and one settlement monitoring sensor 3 need to be arranged on each steel pipe column 5 of the temporary bridge under construction. The three sensors are arranged along the circumference of the steel pipe column, and the two stress sensors 4 are symmetrically arranged; and a displacement sensor 1 is arranged at the lower part of the bridge deck steel plate in the middle of each span; if it is assumed that the temporary bridge under construction has a total of m spans and k steel pipe columns, then 2*k stress sensors need to be arranged, k settlement monitoring sensors need to be arranged, and m displacement sensors need to be arranged, and the steel pipe columns and settlement monitoring sensors are numbered according to 1, 2, 3...k, the displacement sensors are numbered according to 1, 2, 3...m, and the stress sensors are numbered according to 1, 2, 3...s; finally, multiple sensors are connected to the matching automatic acquisition and transportation system (settlement monitors can be used) for data acquisition;

[0114] S2. By collecting the settlement data J of all steel pipe columns in real time, and calculating the uneven settlement index data set D of the steel pipe columns at each moment, and the variance P of the uneven settlement change of all steel pipe columns within the unit collection time, the calculated variance P of the uneven settlement change is finally used as the uneven settlement index of the steel pipe columns of the temporary bridge under construction; the specific process is as follows:

[0115] S2(1). The settlement of all steel pipe columns during the construction process is monitored by buried settlement monitoring sensors, and settlement data is collected in real time; the settlement monitoring sensors transmit the settlement data signals collected in real time to the settlement monitoring instrument, and the settlement monitoring instrument collects the settlement data collected by each settlement monitoring sensor, and obtains the settlement data samples J of all steel pipe columns:

[0116] J=[J1 J2...J i ...J n ]

[0117] J i =[j i,1 j i,2 ... i,s ... i,k ]

[0118] Where: i is the time step number, n is the total number of time steps;

[0119] k is the serial number of the sedimentation sensor;

[0120] J i is the settlement data set of all settlement sensors at any i time step within n time steps; J n is the settlement data set of all settlement sensors at the nth (i.e. last) time step;

[0121] j i,sis the settlement data of the sth settlement sensor at the i-th time step;

[0122] j i,k is the settlement data of the kth (i.e. the last) settlement sensor at the i-th time step; assuming that the sampling frequency of the settlement monitor is 10 s each time, and 30 minutes is a sampling time unit, the total number of time steps n = 180, i = 1...n;

[0123] S2 (2) Calculate the uneven settlement index data set D of the steel pipe column at each moment; extract the settlement data collected by the settlement monitor at each moment, and calculate the uneven settlement index data set D of the steel pipe column at each moment by the following formula;

[0124] D=[D1 D2...D i ...D n ]

[0125]

[0126]

[0127] Where: i is the time step number; k is the total number of settlement sensors;

[0128] j i,s is the settlement data of the sth settlement sensor at the ith time step;

[0129] is the average settlement data of all k settlement sensors at the i-th time step;

[0130] D i is the settlement variation variance of all steel tube columns at any i-th time step within n time steps, which is used to evaluate the uneven settlement degree of all steel tube columns of the temporary bridge under construction at the i-th time step;

[0131] D n is the settlement variation variance of all steel tube columns at the nth (i.e. the last) time step, which is used to evaluate the uneven settlement degree of all steel tube columns of the temporary bridge under construction at the last moment;

[0132] S2(3). The settlement variance D of all steel pipe columns calculated in step S2(2) at the i-th time step i Calculate the variance P of the uneven settlement of all steel pipe columns within the unit collection time;

[0133]

[0134] Where: n is the total number of time steps;

[0135] D iis the settlement variation variance of all steel tube columns at the i-th time step;

[0136] It is the average value of the uneven settlement index of the steel pipe columns of the temporary bridge under construction within the unit collection time;

[0137] The variance P of the uneven settlement change of the steel pipe column within the unit collection time is calculated to represent the settlement data change of all the steel pipe columns of the temporary bridge within the unit collection time. The larger the index value, the greater the uneven settlement of the steel pipe columns of the temporary bridge, that is, an uneven settlement index of the steel pipe columns of the temporary bridge is obtained;

[0138] S3. By collecting the stress data sample F of the steel pipe column in real time, the stress characteristic index data set E of all the steel pipe columns is calculated, and then the overall force distribution characteristic index M of the steel pipe column of the temporary bridge under construction is calculated by the stress characteristic index data set E. The specific steps are as follows:

[0139] S3(1). Two stress sensors are arranged on each steel pipe column to collect stress data of each steel pipe column in real time, and the stress data sample F of all steel pipe columns is obtained:

[0140] F=[F1 F2...F i ...F n ]

[0141] F i =[f i,1 f i,2 ...f i,s ...f i,k ]

[0142]

[0143] Where: F i is the axial stress data set of all steel pipe columns at any i-th time step within n time steps;

[0144] F n is the axial stress data set of all steel tube columns at the nth (i.e. the last) time step;

[0145] f i,s is the axial stress data of the s-th steel pipe column at any i-th time step within n time steps;

[0146] f i,k is the axial stress data of the kth (i.e. the last) steel pipe column at any i-th time step within n time steps;

[0147] and They are the stress data monitored by two stress sensors arranged on the s-th steel pipe column;

[0148] The data sampling frequency of the stress sensor is 10 seconds each time, with 30 minutes as a time unit, to collect the stress values ​​of each steel pipe column of the temporary bridge under construction;

[0149] S3(2). According to the obtained stress data sample F, determine the maximum stress f in each steel pipe column stress data sample per unit time. max With the minimum stress f min The difference between them is combined with the stress variance of each steel tube column to calculate the stress characteristic index data set E of all steel tube columns;

[0150] E=[E1...E s ...E k ]

[0151]

[0152] Where: f i,s is the axial stress data of the s-th steel pipe column at the i-th time step;

[0153] f max is the maximum stress value within the unit acquisition time;

[0154] f min is the minimum stress value per unit acquisition time;

[0155] E s is the stress characteristic index of the s-th steel pipe column;

[0156] E k It is the stress characteristic index of the last steel pipe column;

[0157] The calculated stress characteristic index E of the s-th steel pipe column s It is used to evaluate the stress change of the sth steel pipe column of the temporary bridge under construction within the unit collection time. By comparing the extreme values, it reflects the unevenness of the stress distribution of different steel pipe columns. Combined with the mean stress value of each steel pipe column, it reflects the load size borne by each bracket.

[0158] S3(3). Calculate the maximum stress characteristic index E among all stress characteristic indexes of steel tube columns through the stress characteristic index data set of steel tube columns. max and the minimum stress characteristic index E min The difference between them is combined with the stress characteristic index variance of all supports to obtain the overall force distribution characteristic index M of the steel pipe column of the temporary bridge under construction;

[0159]

[0160] Where: M is the overall force distribution characteristic index of the steel pipe column of the temporary bridge under construction;

[0161] E max is the largest column stress characteristic index in data set E,

[0162] E min is the smallest column stress characteristic index in data set E,

[0163] M is used to evaluate the stress distribution of the overall steel pipe column of the temporary bridge. The larger the range and mean, the more uneven the stress distribution of the steel pipe column of the temporary bridge. The greater the load, the worse the stability of the steel pipe column of the temporary bridge. Therefore, the stress distribution characteristic index M of the steel pipe column of the temporary bridge is obtained.

[0164] S4. The larger the uneven settlement index P and the stress distribution characteristic index M of the temporary bridge steel pipe column, the worse the stability of the temporary bridge steel pipe column. Therefore, the stability coefficient U of the temporary bridge steel pipe column can be defined. According to the uneven settlement index P of the temporary bridge steel pipe column calculated in step S2, combined with the stress distribution characteristic index M of the temporary bridge steel pipe column calculated in step S3, the stability coefficient U of the temporary bridge steel pipe column is calculated:

[0165] U=P×M

[0166] In the above formula, U is the instability index of the steel pipe column of the temporary bridge. U changes with the uneven settlement of the steel pipe column of the temporary bridge and the change of the characteristic index of the force distribution. When the uneven settlement of the steel pipe column of the temporary bridge is greater and the force distribution is more uneven, the instability index of the steel pipe column of the temporary bridge is greater, and the risk of instability is greater. Thus, the instability index U of the steel pipe column of the temporary bridge is obtained;

[0167] S5. Real-time collection of vertical vibration displacement time history data of the bridge deck, calculation of the vertical vibration displacement change index set of each position of the bridge deck within the unit collection time, and calculation of the bridge deck vibration fluctuation index by collecting the vertical vibration displacement change index set T of each position of the bridge deck; the specific process is as follows:

[0168] S5(1). Use the displacement sensor to read the vertical vibration displacement time history data of the bridge deck; the displacement sensor sampling frequency is 10s each time, with 30 minutes as a time unit, and the vertical vibration displacement data in each unit time is collected, which can be obtained from the construction temporary bridge deck vibration amplitude sample set Z:

[0169] Z=[Z1...Z i ...Z n ]

[0170] Z i =[z i,1 ...zi,x ...z i,m ]

[0171] Where: n is the total number of time steps; i is the time step sequence number, i = 1...n;

[0172] m is the displacement sensor serial number; x is any displacement sensor among the m displacement sensors;

[0173] Z i is the displacement data set of all displacement sensors at any i-th time step in n time steps;

[0174] Z n is the displacement data set of all displacement sensors at the nth time step;

[0175] z i、x is the displacement data of the x-th displacement sensor at any i-th time step in n time steps;

[0176] z i、m is the displacement data of the mth (i.e. the last) displacement sensor at any i-th time step in n time steps;

[0177] S5(2). Calculate the vertical displacement variation variance of each displacement sensor position within a unit acquisition time based on the displacement data collected in S5(1), that is, the vertical vibration displacement variation index set T of each position on the bridge deck within a unit acquisition time:

[0178] T=[T1...T x ...T m ]

[0179]

[0180]

[0181] Where: n is the total number of time steps; x is the serial number of any displacement sensor among the m displacement sensors;

[0182] is the average displacement data of the xth displacement sensor within the unit acquisition time;

[0183] T x is the variance of the vertical displacement change at the xth displacement sensor position of the temporary bridge under construction;

[0184] T m is the vertical displacement variance of the mth (last) displacement sensor position of the temporary bridge under construction; T x It is used to evaluate the change of vertical displacement of the xth bridge deck position within the acquisition unit time;

[0185] S5(3). Calculate the bridge deck vibration fluctuation index C by using the vertical vibration displacement change index set T of each bridge deck position within the unit acquisition time calculated in step S5(2):

[0186]

[0187]

[0188] Where: T x is the vertical displacement variance of the xth displacement sensor position of the temporary bridge under construction,

[0189] is the average value of the variance of vertical displacement changes at all displacement sensor locations of the temporary bridge under construction;

[0190] C is the bridge deck vibration fluctuation index, which is used to evaluate the vibration displacement fluctuation of the construction temporary bridge deck;

[0191] As the service time increases, the overall stiffness of the construction bridge decreases, and the vertical vibration displacement under the action of the moving load also increases. C can evaluate the overall vibration displacement change of the construction bridge. When the vibration displacement of the bridge deck increases, the vibration fluctuation index C of the construction bridge also increases, indicating that the stability of the construction bridge decreases. Thus, the vibration displacement fluctuation index C of the bridge deck of the trestle structure is obtained.

[0192] S6. Overall stability coefficient model of construction temporary bridge; the overall structure of construction temporary bridge is mainly composed of steel tube columns and bridge deck system. Under the action of external disturbance, the uneven settlement of columns and the excessive vibration amplitude of bridge deck may lead to the weakening of the rigidity of trestle structure, and even induce the overall instability and collapse of the structure. Therefore, the instability index U of steel tube columns and the vibration displacement fluctuation index C of bridge deck can be used to fit the binary high-order polynomial to establish the parameter-containing model of the overall stability coefficient of construction temporary bridge; the specific process is as follows:

[0193] S6(1). According to the instability index U of the steel pipe column obtained in step S4 and the vibration displacement fluctuation index C of the temporary bridge deck obtained in step S5, a polynomial fitting model of the overall stability coefficient Q of the temporary bridge is established:

[0194] Where: Q is the fitting value of the overall stability coefficient of the construction temporary bridge, which is calculated from the instability index U of the steel pipe column and the vibration displacement fluctuation index C of the bridge deck;

[0195] r is the highest power of variable U in the fitting formula, and v is the highest power of variable C in the fitting formula. To obtain sufficient fitting accuracy, it is recommended that r ≥ 6 and v ≥ 6 in the formula;

[0196] a ij are the polynomial coefficients;

[0197] S6(2). Determine a based on the overall stability coefficient sample of the construction temporary bridge obtained by simulation ij , the specific calculation method is as follows:

[0198] a. Based on a given set of steel pipe column settlement data J * and stress data F * , a refined model of the temporary bridge under construction considering the initial uneven settlement and uneven stress distribution is established in the finite element software; then, according to another set of given bridge deck vertical displacement change indicators T * , the corresponding dynamic load is applied to the construction bridge model (based on the principle of the same maximum vertical displacement), and finally the design load is applied to the bridge model in multiples until the overall structure becomes unstable, thereby obtaining the overall stability coefficient Q of the construction bridge * ;

[0199] b. Based on the data J given in step a * and data F * , calculate the corresponding steel pipe column instability index U according to step S4 * ; According to the bridge deck vertical displacement change index T given in step a * , calculate the bridge deck vibration displacement fluctuation index C according to step S5 * ;

[0200] c. Repeat steps a and b at least (r+1)*(v+1) times to obtain a series of data and

[0201] d. Q * , U * and C * Substitute into the parameter-containing model in step S6(1), and determine the coefficients of each polynomial in the model by solving the overdetermined equations The calculation formula of the overall stability coefficient Q of the temporary bridge is finally obtained. When the instability index U of the steel pipe column and the vibration displacement fluctuation index C of the bridge deck are larger, the overall stability coefficient of the temporary bridge is smaller, and the structure is more likely to become unstable and damaged.

[0202] S7: Stability assessment of temporary bridges under construction. Since the instability of temporary bridge structures is usually sudden, it is urgent to predict in advance whether the structure has the risk of overall instability during construction. To this end, based on the real-time measurement of uneven settlement, uneven stress and bridge deck vibration of temporary bridge columns under construction, deep learning methods can be used to analyze the laws from a large amount of monitored time-varying data to predict the deterioration trend of the stability of temporary bridges under construction within a certain period of time, so as to carry out emergency treatment in a timely manner.

[0203] S7(1). The failure threshold of the overall stability coefficient Q of the temporary bridge is set. Generally, the elastic-plastic stability bearing capacity of the structure is considered to be twice the design load. Therefore, it is recommended that the failure threshold of the overall stability coefficient Q of the temporary bridge be 2;

[0204] S7(2). Based on the prediction of the stability degradation trend of the neural network, repeat the above steps S2 to S5, and use the formula in step S6 to calculate the stability coefficient of the construction bridge for multiple acquisition units, and obtain the time-based change sequence of the stability coefficient of the construction bridge Q = {Q1, Q2, ..., Q i};

[0205] S7(3). Use the time convolution neural network (TCN) to predict the stability of the temporary bridge under construction, and collect the stability change sequence of the temporary bridge under construction {Q1, Q2, …, Q i-1} as samples, and use the overall structural stability coefficient of the temporary bridge at time i as output data for model training until the structural stability coefficient at time i is accurately predicted;

[0206] S7(4). Evaluation of the stability degradation trend of the construction bridge. When the stability coefficient of the construction bridge predicted by the time convolution (TCN) neural network is less than the threshold, the construction bridge has suffered from major fatigue damage, uneven settlement, loose connection and other problems. The construction bridge needs to be repaired and inspected immediately, or the construction should be stopped.

[0207] The present invention comprehensively considers three factors of uneven settlement of steel pipe columns of a temporary bridge under construction, uneven stress on steel pipe columns, and vibration displacement of the bridge deck, designs an uneven settlement index P of steel pipe columns, an overall stress distribution characteristic index M of steel pipe columns of the temporary bridge under construction, and a vibration fluctuation index C of the bridge deck, and proposes an overall stability coefficient Q of the temporary bridge under construction. By analyzing the overall stability coefficient of the temporary bridge under construction, the stability coefficient of the temporary bridge under construction within a unit acquisition time and the stability degradation trend during long-term use can be conveniently and quickly evaluated. At the same time, the stability of the temporary bridge under construction can be predicted in combination with technologies such as neural networks, thereby solving the problem that the existing evaluation method cannot comprehensively consider the influence of factors such as stress on steel pipe columns, uneven settlement of steel pipe columns, and vibration of the bridge deck on stability, and cannot evaluate the stability degradation trend of the temporary bridge under long-term use conditions.

[0208] The above is only one embodiment of the present invention, and its description is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the attached claims.

Claims

1. A trestle stability analysis method based on deep neural network, characterized in that: The specific steps include: S1. Stress sensors and settlement monitoring sensors are installed on each steel pipe column of the temporary bridge under construction, and displacement sensors are installed under the bridge deck steel plate in the middle of each span; the steel pipe columns and settlement monitoring sensors are numbered as 1, 2, 3...k, the displacement sensors are numbered as 1, 2, 3...m, and the stress sensors are numbered as 1, 2, 3...s; S2. By collecting the settlement data J of all steel pipe columns in real time, and calculating the uneven settlement index data set D of the steel pipe columns at each moment, and the variance P of the uneven settlement change of all steel pipe columns within the unit collection time, the calculated uneven settlement change variance P is finally used as the uneven settlement index of the steel pipe columns of the temporary bridge; the calculation process of the uneven settlement change variance P is as follows: Where: n is the total number of time steps; D i is the settlement variation variance of all steel tube columns at any i-th time step within n time steps; It is the average value of the uneven settlement index of the steel pipe columns of the temporary bridge under construction within the unit collection time; i is the time step number; k is the total number of settlement sensors; j i,s is the settlement data of the sth settlement sensor at the ith time step; is the average settlement data of all k settlement sensors at the i-th time step; S3. By collecting the stress data sample F of the steel pipe column in real time, the stress characteristic index data set E of all the steel pipe columns is calculated, and then the overall force distribution characteristic index M of the steel pipe column of the temporary bridge is calculated by the stress characteristic index data set E. The calculation formula of the overall force distribution characteristic index M of the steel pipe column of the temporary bridge is as follows: Where: M is the overall force distribution characteristic index of the steel pipe column of the temporary bridge under construction; E max is the maximum column stress characteristic index in data set E; E min is the smallest column stress characteristic index in data set E; S4. According to the uneven settlement index P of the steel pipe column of the temporary bridge calculated in step S2, combined with the stress distribution characteristic index M of the steel pipe column of the temporary bridge calculated in step S3, the stability coefficient U of the steel pipe column of the temporary bridge is calculated: U = P × M; S5. Collect the time history data of vertical vibration displacement of the bridge deck in real time, calculate the vertical vibration displacement change index set of each position of the bridge deck within the unit collection time, and calculate the bridge deck vibration fluctuation index C by collecting the vertical vibration displacement change index set T of each position of the bridge deck; the calculation formula is as follows: Where: T x is the vertical displacement variance of the xth displacement sensor position of the temporary bridge under construction, is the average value of the variance of vertical displacement changes at all displacement sensor locations of the temporary bridge under construction; m is the displacement sensor serial number; x is the serial number of any displacement sensor among the m displacement sensors; T x is the variance of the vertical displacement change at the xth displacement sensor position of the temporary bridge under construction; S6. According to the instability index U of the steel pipe column obtained in step S4 and the vibration displacement fluctuation index C of the temporary bridge deck obtained in step S5, a polynomial fitting model of the overall stability coefficient Q of the temporary bridge is established: Where: Q is the fitting value of the overall stability coefficient of the construction temporary bridge, which is calculated from the instability index U of the steel pipe column and the vibration displacement fluctuation index C of the bridge deck; r is the highest power of variable U in the fitting formula, and v is the highest power of variable C in the fitting formula. To obtain sufficient fitting accuracy, it is recommended that r ≥ 6 and v ≥ 6 in the formula; a ij are the polynomial coefficients; S7. Determine a based on the overall stability coefficient sample of the construction temporary bridge obtained by simulation ij , and substitute it into the formula of step S6 to finally obtain the calculation formula of the overall stability coefficient Q of the construction temporary bridge; substitute the real-time monitored steel tube column instability index U and bridge deck vibration displacement fluctuation index C into the final calculation formula of the overall stability coefficient Q of the construction temporary bridge to analyze the overall stability of the construction temporary bridge; when the steel tube column instability index U and the bridge deck vibration displacement fluctuation index C are larger, the overall stability coefficient of the construction temporary bridge is smaller, and the structure is more likely to become unstable and damaged.

2. The method for analyzing trestles stability based on deep neural networks according to claim 1, characterized in that: It also includes step S8, which uses deep learning methods to analyze the patterns from a large amount of monitored time-varying data based on real-time measurement of uneven settlement of temporary bridge columns, uneven stress, and bridge deck vibration, and predicts the deterioration trend of the stability of the temporary bridge within a certain period of time, so as to carry out emergency treatment in a timely manner.

3. A trestle stability analysis method based on deep neural network according to claim 1 or 2, characterized in that: In the step S1, two stress sensors and one settlement monitoring sensor are arranged on each steel pipe column of the temporary bridge under construction. The three sensors are arranged along the circumference of the steel pipe column, and the two stress sensors are symmetrically arranged; and a displacement sensor is arranged at the lower part of the bridge deck steel plate in the middle of each span; if it is assumed that the temporary bridge under construction has a total of m spans and k steel pipe columns, then 2*k stress sensors need to be arranged, k settlement monitoring sensors need to be arranged, and m displacement sensors need to be arranged; multiple sensors are connected to the matching settlement monitor for data collection.

4. A trestle stability analysis method based on deep neural network according to claim 1 or 2, characterized in that: The specific process of the S2 step is as follows: S2(1). The settlement of all steel pipe columns during the construction process is monitored by buried settlement monitoring sensors, and settlement data is collected in real time; the settlement monitoring sensors transmit the settlement data signals collected in real time to the settlement monitoring instrument, and the settlement monitoring instrument collects the settlement data collected by each settlement monitoring sensor, and obtains the settlement data samples J of all steel pipe columns: J=[J1 J2...J i ...J n ] I i =[j i,1 I i,2 ...I i,s ...I i,k ] Where: i is the time step number, n is the total number of time steps; k is the serial number of the sedimentation sensor; J i is the set of settlement data of all settlement sensors at any i time step within n time steps; J n is the settlement data set of all settlement sensors at the nth time step; j i,s is the settlement data of the sth settlement sensor at the i-th time step; j i,k is the settlement data of the kth (i.e. the last) settlement sensor at the i-th time step; Assuming that the sampling frequency of the settlement monitor is 10 seconds each time, and 30 minutes is used as a sampling time unit, the total number of time steps n = 180, i = 1...n; S2 (2) Calculate the uneven settlement index data set D of the steel pipe column at each moment; extract the settlement data collected by the settlement monitor at each moment, and calculate the uneven settlement index data set D of the steel pipe column at each moment by the following formula; D=[D1 D2...D i ...D n ] Where: i is the time step number; k is the total number of settlement sensors; j i,s is the settlement data of the sth settlement sensor at the ith time step; is the average settlement data of all k settlement sensors at the i-th time step; D i is the settlement variation variance of all steel tube columns at any i-th time step within n time steps, which is used to evaluate the uneven settlement degree of all steel tube columns of the temporary bridge under construction at the i-th time step; D n is the settlement variation variance of all steel tube columns at the nth (i.e. the last) time step, which is used to evaluate the uneven settlement degree of all steel tube columns of the temporary bridge under construction at the last moment; S2(3). The settlement variance D of all steel pipe columns calculated in step S2(2) at the i-th time step i Calculate the variance P of uneven settlement changes of all steel pipe columns within a unit collection time; the variance P of uneven settlement changes of steel pipe columns within a unit collection time calculated represents the settlement data changes of all steel pipe columns of the temporary bridge under construction within a unit collection time. The larger the index value, the greater the uneven settlement of the steel pipe columns of the temporary bridge under construction, that is, an uneven settlement index of the steel pipe columns of the temporary bridge under construction is obtained; 5. A trestle stability analysis method based on deep neural network according to claim 1 or 2, characterized in that: The specific process of the S3 step is as follows: S3(1). Two stress sensors are arranged on each steel pipe column to collect stress data of each steel pipe column in real time, and the stress data sample F of all steel pipe columns is obtained: F=[F1 F2...F i ...F n ] F i =[f i,1 f i,2 ...f i,s ...f i,k ] Where: F i is the axial stress data set of all steel pipe columns at any i-th time step within n time steps; F n is the axial stress data set of all steel tube columns at the nth (i.e. the last) time step; f i,s is the axial stress data of the s-th steel pipe column at any i-th time step within n time steps; f i,k is the axial stress data of the kth steel pipe column at any i-th time step within n time steps; and They are the stress data monitored by two stress sensors arranged on the s-th steel pipe column; The data sampling frequency of the stress sensor is 10 seconds each time, with 30 minutes as a time unit, to collect the stress values ​​of each steel pipe column of the temporary bridge under construction; S3(2). According to the obtained stress data sample F, determine the maximum stress f in each steel pipe column stress data sample per unit time. max With the minimum stress f min The difference between them is combined with the stress variance of each steel tube column to calculate the stress characteristic index data set E of all steel tube columns; E=[E1...E s ...HAVE BEEN k ] Where: f i,s is the axial stress data of the s-th steel pipe column at the i-th time step; f max is the maximum stress value within the unit acquisition time; f min is the minimum stress value per unit acquisition time; E s is the stress characteristic index of the s-th steel pipe column; E k It is the stress characteristic index of the last steel pipe column; The calculated stress characteristic index E of the s-th steel pipe column s It is used to evaluate the stress change of the sth steel pipe column of the temporary bridge under construction within the unit collection time. By comparing the extreme values, it reflects the unevenness of the stress distribution of different steel pipe columns. Combined with the mean stress value of each steel pipe column, it reflects the load size borne by each bracket. S3(3). Calculate the maximum stress characteristic index E among all stress characteristic indexes of steel tube columns through the stress characteristic index data set of steel tube columns. max and the minimum stress characteristic index E min The difference between them is combined with the stress characteristic index variance of all supports to obtain the overall force distribution characteristic index M of the steel pipe column of the temporary bridge under construction; M is used to evaluate the stress distribution of the overall steel tube column of the temporary bridge. The larger the range and mean, the more uneven the stress distribution of the steel tube column of the temporary bridge. The greater the load, the worse the stability of the steel tube column of the temporary bridge. Therefore, the stress distribution characteristic index M of the steel tube column of the temporary bridge is obtained.

6. A trestle stability analysis method based on deep neural network according to claim 1 or 2, characterized in that: The specific process of step S5 is as follows: S5(1). Use the displacement sensor to read the vertical vibration displacement time history data of the bridge deck; the displacement sensor sampling frequency is 10s each time, with 30 minutes as a time unit, and the vertical vibration displacement data in each unit time is collected, which can be obtained from the construction temporary bridge deck vibration amplitude sample set Z: Z=[Z1...Z i ...WITH n ] WITH i =[of i,1 ...With i,x ...With i,m ] Where: n is the total number of time steps; i is the time step sequence number, i = 1...n; m is the displacement sensor serial number; x is any displacement sensor among the m displacement sensors; Z i is the displacement data set of all displacement sensors at any i-th time step in n time steps; Z n is the displacement data set of all displacement sensors at the nth time step; z i、x is the displacement data of the x-th displacement sensor at any i-th time step in n time steps; z i、m is the displacement data of the mth displacement sensor at any i-th time step in n time steps; S5(2). Calculate the vertical displacement variation variance of each displacement sensor position within a unit acquisition time based on the displacement data collected in S5(1), that is, the vertical vibration displacement variation index set T of each position on the bridge deck within a unit acquisition time: T=[T1...T x ...T m ] Where: n is the total number of time steps; x is the serial number of any displacement sensor among the m displacement sensors; is the average displacement data of the xth displacement sensor within the unit acquisition time; T x is the variance of the vertical displacement change at the xth displacement sensor position of the temporary bridge under construction; T m is the variance of the vertical displacement change at the mth displacement sensor position of the temporary bridge under construction; T x It is used to evaluate the change of vertical displacement of the xth bridge deck position within the acquisition unit time; S5(3). Calculate the bridge deck vibration fluctuation index C through the vertical vibration displacement change index set T of each bridge deck position within the unit collection time calculated in step S5(2); with the increase of service time, the overall stiffness of the construction temporary bridge decreases, and the vertical vibration displacement under the action of the moving load also increases accordingly. C can evaluate the overall vibration displacement change of the construction temporary bridge. When the bridge deck vibration displacement increases, the construction temporary bridge vibration fluctuation index C also increases, indicating that the stability of the construction temporary bridge is reduced. Thus, the bridge deck vibration displacement fluctuation index C of the trestle structure is obtained.

7. A trestle stability analysis method based on deep neural network according to claim 1 or 2, characterized in that: In step S6, a is determined based on the overall stability coefficient sample of the temporary bridge under construction obtained by simulation. ij , the specific calculation method is as follows: a. Based on a given set of steel pipe column settlement data J * and stress data F * , a refined model of the temporary bridge under construction considering the initial uneven settlement and uneven stress distribution is established in the finite element software; then, according to another set of given bridge deck vertical displacement change indicators T * , the corresponding dynamic load is applied to the construction bridge model (based on the principle of the same maximum vertical displacement), and finally the design load is applied to the bridge model in multiples until the overall structure becomes unstable, thereby obtaining the overall stability coefficient Q of the construction bridge * ; b. Based on the data J given in step a * and data F * , calculate the corresponding steel pipe column instability index U according to step S4 * ; According to the bridge deck vertical displacement change index T given in step a * , calculate the bridge deck vibration displacement fluctuation index C according to step S5 * ; c. Repeat steps a and b at least (r+1)*(v+1) times to obtain a series of data and d. Q * , U * and C * Substitute into the parameter-containing model in step S6(1), and determine the coefficients of each polynomial in the model by solving the overdetermined equations The calculation formula for the overall stability coefficient Q of the construction temporary bridge is finally obtained. When the instability index U of the steel tube column and the vibration displacement fluctuation index C of the bridge deck are larger, the overall stability coefficient of the construction temporary bridge is smaller, and the structure is more likely to become unstable and damaged.

8. A trestle stability analysis method based on deep neural network according to claim 2, characterized in that: The specific process of step S8 is as follows: S8(1). The failure threshold of the overall stability coefficient Q of the temporary bridge is set. Generally, the elastic-plastic stability bearing capacity of the structure is considered to be twice the design load. Therefore, it is recommended that the failure threshold of the overall stability coefficient Q of the temporary bridge be 2; S8(2). Based on the prediction of stability degradation trend of neural network, repeat the above steps S2 to S5, and use the formula in step S6 to calculate the stability coefficient of the construction bridge for multiple acquisition units, and obtain the time-based change sequence of the stability coefficient of the construction bridge Q = {Q1, Q2, ..., Q i }; S8(3). Use the time convolution neural network (TCN) to predict the stability of the temporary bridge under construction, and collect the stability change sequence of the temporary bridge under construction {Q1, Q2, …, Q i-1 } as samples, and use the overall structural stability coefficient of the temporary bridge at time i as output data for model training until the structural stability coefficient at time i is accurately predicted; S8(4). Evaluation of the stability degradation trend of the construction bridge. When the stability coefficient of the construction bridge predicted by the time convolutional neural network is less than the threshold, the construction bridge has suffered major fatigue damage, uneven settlement, loose connection and other problems. The construction bridge needs to be repaired and inspected immediately, or construction should be stopped.

Images