Method for extracting long-period bridge load effect under large traffic flow

Abstract

The application discloses a long-period bridge load effect extraction method under large traffic flow, comprising the following steps: S1, constructing a neural network model; S2, inputting vehicle sample data into the neural network model for network model training to obtain a trained neural network model; S3, collecting real-time vehicle data and inputting the vehicle data into the trained neural network model to output vehicle data in a time period T; S4, determining the lateral distribution of each beam in the bridge; S5, calculating the mid-span bending moment influence line of the most unfavorable beam; S6, taking the beam length as a total step length L, taking a step distance X as a step moving distance, gradually loading the vehicle data in the time period T on the mid-span bending moment influence line of the most unfavorable beam, and calculating the mid-span bending moment effect generated at each step. The application can expand the time length of vehicle data, and provides a possibility for predicting whether the bridge has good safety, applicability and durability in the future.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of bridge load, in particular to a long period bridge load effect extraction method under large traffic flow. BACKGROUND

[0002] In recent years, with the rapid growth of China's economy, the transportation industry is booming, the total mileage of highways and the proportion of high-grade highways have increased significantly, at the same time, with the rapid development of the automobile industry and the transportation industry, the highway traffic flow has increased significantly, the vehicle driving speed has improved significantly, and the heavy vehicles have emerged in large numbers, which has changed the vehicle load greatly, which has seriously threatened the operation safety of highway bridges, in addition, nowadays, with the continuous development and application of lightweight and high-strength concrete and the replacement of the traditional allowable stress method by the new probability limit state method, the working stress of concrete bridges is getting higher and higher, and in such a situation, it is particularly important to obtain the highway concrete bridge load effect under large traffic flow.

[0003] However, large traffic flow is not maintained at a certain stable level, but changes with time, in order to better explore the load level affecting the bridge, the investigation of traffic flow cannot stop at a specific time point, therefore, the acquisition of large traffic flow under long period is more meaningful. SUMMARY

[0004] Therefore, the purpose of the present application is to overcome the defects in the prior art, provide a long period bridge load effect extraction method under large traffic flow, and expand the time length of vehicle data, which provides the possibility for predicting whether the bridge has good safety, applicability and durability in the future.

[0005] The long period bridge load effect extraction method under large traffic flow of the present application comprises the following steps:

[0006] S1. Constructing a neural network model;

[0007] S2. Inputting vehicle sample data into the neural network model for network model training to obtain a trained neural network model;

[0008] S3. Collecting real-time vehicle data and inputting the vehicle data into the trained neural network model to output vehicle data in a time period T;

[0009] S4. Determining the lateral distribution of each piece of beam in the bridge according to the lateral distribution influence line of the bridge;

[0010] S5. Calculating the mid-span moment influence line of the most unfavorable force beam according to the lateral distribution of each piece of beam and the longitudinal bending moment influence line of the bridge;

[0011] S6. Take the beam length as the total step length L, take the step distance X as the moving distance of the step length, load the vehicle data in the time period T on the mid-span bending moment influence line of the most unfavorable force bearing beam step by step, and calculate the mid-span bending moment effect generated by each step.

[0012] Further, in step S2, the correctness of the vehicle sample data is checked and the vehicle sample data is normalized before being input into the neural network model.

[0013] Further, the step S2 specifically comprises:

[0014] S21. Input the vehicle sample data into the neural network model, and output the actual output value;

[0015] S22. Determine whether the error between the actual output value and the expected output value is within the set range, if yes, the training is ended and step S23 is entered; if not, the parameter value in the neural network model is adjusted, and steps S21-S22 are repeated;

[0016] S23. The neural network model after training is taken as the trained neural network model.

[0017] Further, the error E between the actual output value and the expected output value is determined according to the following formula:

[0018]

[0019] Wherein, d k is the expected output value, y k is the actual output value, k is the total number of output layer neurons; i is the total number of input layer neurons.

[0020] Further, the vehicle data includes vehicle type, daily traffic volume, lane lateral distribution, vehicle spacing and vehicle weight.

[0021] Further, the transverse direction is the bridge width direction of the bridge, and the longitudinal direction is the bridge length direction of the bridge.

[0022] The beneficial effects of the present application are: the long period bridge load effect extraction method under large traffic flow disclosed in the present application expands the time length of the data, provides the possibility for predicting whether the bridge has good safety, applicability and durability in the future, and analyzes the effective information under different longitudinal and lateral distributions of vehicle load based on long period vehicle data, which is more real and reliable than the information only considering the longitudinal influence of the vehicle. BRIEF DESCRIPTION OF DRAWINGS

[0023] The present application will be further described below in conjunction with the drawings and examples:

[0024] Figure 1A schematic diagram of a long-period vehicle data acquisition process of the present application is shown in the figure.

[0025] Figure 2 A schematic diagram of a most unfavorable beam mid-span moment effect extraction process of the present application is shown in the figure. DETAILED DESCRIPTION

[0026] The present application is further described below in conjunction with the accompanying drawings of the specification, as shown in the figures:

[0027] The long-period bridge load effect extraction method under heavy traffic flow of the present application comprises the following steps:

[0028] S1. Construct a neural network model.

[0029] S2. Input vehicle sample data into the neural network model for network model training to obtain a trained neural network model.

[0030] S3. Collect real-time vehicle data and input the vehicle data into the trained neural network model to output vehicle data within a time period T; wherein the vehicle data includes vehicle type, daily traffic volume, lane transverse distribution, vehicle spacing, and vehicle weight; the time period T is determined by the error between the input value and the output value during the neural network model training.

[0031] S4. Determine the transverse distribution of each piece of beam in the bridge according to the transverse distribution influence line of the bridge.

[0032] S5. Calculate the mid-span moment influence line of the most stressed beam according to the transverse distribution of each piece of beam and the longitudinal bending moment influence line of the bridge; wherein the transverse direction is the bridge width direction, and the longitudinal direction is the bridge length direction.

[0033] S6. Take the beam length as the total step length L, take the step distance X as the step moving distance, gradually load the vehicle data within the time period T on the mid-span moment influence line of the most stressed beam, and calculate the mid-span moment effect generated at each step.

[0034] The long-period bridge load effect extraction method under heavy traffic flow of the present application needs to first acquire long-period vehicle data under heavy traffic flow, and acquiring long-period vehicle data is to understand the random vehicle situation in the present time and even a relatively long period of time in the future; the current traffic flow data can directly read the traffic flow observation data of the relevant site to obtain vehicle type, daily traffic volume, lane transverse distribution, vehicle spacing, and vehicle weight, etc., however, these data are seemingly disordered, but the internal rules need to be excavated, such as different vehicle types (statistical chart), daily traffic volume (utilizing time relationship diagram), lane transverse distribution (statistical chart), vehicle spacing (generally recommended to adopt logarithmic normal distribution in general running state and dense running state), and vehicle weight (statistical chart);

[0035] However, highway bridges are long-term service, and the load history is also long-term, and only using the real-time vehicle data provided by the observation site at the moment can only get the current bridge load effect information. For predicting whether the bridge has good safety, applicability and durability in the future, there is a lack of reliable vehicle load information, so it is necessary to extract large traffic flow vehicle data information in a long period in the extraction of load effect.

[0036] In this embodiment, in step S1, the current vehicle sample data can be collected or made in advance, and the size of the vehicle sample data can be determined according to the actual working condition, and the vehicle sample data includes vehicle type, daily traffic volume, lane transverse distribution, vehicle spacing and vehicle weight.

[0037] The neural network model is constructed, specifically including: determining the structure layer number and the number of neurons of each layer of the neural network model according to the actual working condition, and initially setting the parameter values of each layer, so as to obtain the overall structure of the neural network model. The neural network model adopts a BP neural network model.

[0038] In this embodiment, in step S2, before the vehicle sample data is input into the neural network model, the correctness of the vehicle sample data is checked and the vehicle sample data is normalized. By checking the correctness of the vehicle sample data, the effectiveness of the data input into the neural network model is ensured, and by normalizing the vehicle sample data, the complexity of the neural network model processing data is reduced.

[0039] In this embodiment, the step S2 specifically includes:

[0040] S21. Input the vehicle sample data into the neural network model to output an actual output value;

[0041] S22. Determine whether the error between the actual output value and the expected output value is within the set range, if yes, the training is ended, and step S23 is entered; if not, the parameter value in the neural network model is adjusted, and steps S21-S22 are repeated; wherein the expected output value can be obtained by statistically evaluating the historical vehicle data of the bridge under large traffic flow in a long period; and according to the specific structure and parameter setting of the neural network model, the important parameters affecting the output of the neural network model are adjusted, so that the error between the actual output value and the expected output value gradually reaches the set range;

[0042] Through the above steps, the BP neural network is repeatedly learned and optimized, and the training is stopped after the error is reduced to the expected error range;

[0043] S23. The neural network model after the training is ended is used as a trained neural network model.

[0044] The error E between the actual output value and the expected output value is determined according to the following formula:

[0045]

[0046] Wherein, d k is the expected output value, y k is the actual output value, k is the total number of output layer neurons in the BP neural network model; i is the total number of input layer neurons in the BP neural network model.

[0047] In this embodiment, as shown in Figure 2 , when extracting the bridge load effect considering the real vehicle information, the bridge is regarded as a spatial structure, the longitudinal (along the bridge length) and transverse (along the bridge width) of the traffic flow are combined, and the principle of influence line loading is used to obtain the real load effect (the maximum bending moment of the most unfavorable beam of the simply supported bridge is shown).

[0048] In step S4, it is assumed that the obtained large traffic flow long-period vehicle data information keeps unchanged in the transverse position during the driving process of the simply supported bridge, and the influence of the structural nonlinearity on the load effect is ignored; after the style of the target bridge is determined, each piece of beam in the bridge is numbered, the transverse distribution influence line of the bridge is obtained based on the principle of influence line loading, and then the transverse distribution of each piece of beam in the bridge can be obtained.

[0049] In step S5, based on the above assumption and the principle of influence line loading, the mid-span bending moment influence line of the most unfavorable beam is calculated according to the transverse distribution of each piece of beam and the longitudinal bending moment influence line of the bridge.

[0050] In step S6, the beam length is taken as the total step length L, the step distance X is taken as the moving distance of the step, and the vehicle data in the time period T is gradually loaded on the mid-span bending moment influence line of the most unfavorable beam to calculate the mid-span bending moment effect generated at each step. The step distance X is At each loading step, a mid-span bending moment effect is obtained, so that more comprehensive and detailed load effect information can be obtained by calculating the mid-span bending moment effect generated at each step, and the load level affecting the bridge can be better explored by using the long-period vehicle data under large traffic flow for the above research.

[0051] Finally, it should be pointed out that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the purpose and scope of the present application, and all of them should be covered in the scope of the claims of the present application.

Claims

1. A method for extracting long-term bridge load effects under heavy traffic, characterized in that: The method comprises the following steps: S1. Constructing a neural network model; S2. Inputting vehicle sample data into the neural network model for network model training to obtain a trained neural network model; S3. Collect real-time vehicle data and input the vehicle data into the trained neural network model to output vehicle data in a time period The vehicle data includes vehicle type, daily traffic volume, lane lateral distribution, vehicle spacing, and vehicle weight. S4. Determining the lateral distribution of each piece of beam in the bridge according to the lateral distribution influence line of the bridge, specifically comprising: It is assumed that the obtained large traffic flow long period vehicle data information keeps unchanged in the lateral position during the driving process of the simply supported bridge, and the influence of structural nonlinearity on load effect is ignored; after the style of the target bridge is determined, each piece of beam in the bridge is numbered, the lateral distribution influence line of the bridge is obtained based on the principle of influence line loading, and then the lateral distribution of each piece of beam in the bridge can be obtained; S5. Calculating the mid-span moment influence line of the most stressed beam according to the lateral distribution of each piece of beam and the longitudinal bending moment influence line of the bridge; S6. Take the beam length as the total step size With a step size of The vehicle data in the time period is loaded step by step on the influence line of the mid-span bending moment of the most unfavorable beam under force, and the mid-span bending moment effect produced by each step is calculated.

2. The method for extracting long-term bridge load effects under large traffic flow according to claim 1, characterized in that: In step S2, before the vehicle sample data is input into the neural network model, the correctness of the vehicle sample data is checked and the vehicle sample data is normalized.

3. The method of claim 1, wherein: The step S2 specifically comprises: S21. Inputting the vehicle sample data into the neural network model to output an actual output value; S22. Judging whether the error between the actual output value and an expected output value is within a set range, if yes, the training is ended and step S23 is entered, if not, adjusting the parameter value in the neural network model and repeating steps S21-S22; S23. Taking the neural network model after the training as the trained neural network model.

4. The method for extracting long-term bridge load effects under large traffic flow according to claim 3, characterized in that: The error between the actual output value and the desired output value is determined according to the following equation : ； wherein, is the desired output value, is the actual output value, is the total number of output layer neurons; is the total number of input layer neurons.

5. The method for extracting long-term bridge load effects under large traffic flow according to claim 1, characterized in that: The lateral direction is the bridge width direction of the bridge, and the longitudinal direction is the bridge length direction of the bridge.

Images