Bridge pier local scour depth prediction method and prediction system thereof

Abstract

According to the bridge pier local scour depth prediction method and the prediction system thereof provided by the invention, related parameters influencing the pier local scour depth are determined, the BP neural network is constructed, the corresponding parameters are determined as final input quantities, and the BP neural network is utilized to perform correlation analysis on the influence of each parameter, so that the prediction accuracy of the local scour depth of the bridge pier is improved. Compared with the prior art, independent constant value calculation is not needed, and due to the error back propagation characteristic of the BP neural network, the error amount of final prediction is small and is close to an actual value, so that the local punching depth of the pier can be accurately predicted, and accurate data support is provided for follow-up bridge maintenance.

Description

Technical Field

[0001] The present invention relates to the field of bridge monitoring, and particularly to a method and a system for predicting the local scour depth of bridge piers. Background Art

[0002] A pier is a support for a bridge, mainly used to bear the load of the bridge deck structure or ship body, and transfer the load and its own weight to the ground through the pier foundation.

[0003] When a pier is in a flowing water body, the pier changes the water flow structure and sediment condition around it, thus destroying the equilibrium state of the incoming water flow. As a result, there is a scour phenomenon in the riverbed around the pier. The riverbed scour phenomenon directly affects the embedment depth of the pier foundation seriously. Since the scour depth of the pier is one of the important data for determining the embedment depth of the pier foundation, after the embedment depth of the pier foundation is affected, it will affect the bearing capacity of the bridge, etc., and further affect the overall safety of the bridge.

[0004] In the prior art, the determination of the local scour depth of a pier is generally based on empirical formulas, such as empirical formulas established according to the scour stopping velocity or empirical formulas established according to sediment transport equilibrium. These empirical formulas are established based on current parameters, such as water depth, flow rate, and sediment diameter, etc. Although these formulas have been further optimized and improved in some studies, such as adding factors like river width and vertical flow velocity, there are still the following defects in the prior art: these formulas are calculated based on fixed values at a certain moment. However, the scour depth is formed by the superposition of multiple factors and gradually evolves. The influence between various factors is not considered in the existing methods, resulting in low accuracy of the final scour depth and poor guidance for actual bridge maintenance.

[0005] Therefore, in order to solve the above technical problems, it is urgent to propose a new technical means. Summary of the Invention

[0006] In view of this, the purpose of the present invention is to provide a method and a system for predicting the local scour depth of bridge piers. By determining the relevant parameters affecting the local scour depth of the pier, constructing a BP neural network, and determining the corresponding parameters as the final input quantities, the influence of each parameter is analyzed through the BP neural network in an associated manner rather than independent fixed-value calculation. And due to the error backpropagation characteristic of the BP neural network, the error of the final prediction is small and close to the actual value, so as to accurately predict the local scour depth of the pier and provide accurate data support for subsequent bridge maintenance.

[0007] A method for predicting the local scour depth of a bridge pier provided by the present invention is characterized by comprising the following steps:

[0008] S1. Obtain the historical characteristic parameters of the water area where the target bridge pier is located and the structural parameters of the target bridge pier to be measured itself;

[0009] S2. Obtain the local scour depth of the pier corresponding to the time when the historical characteristic parameters are collected;

[0010] S3. Construct a BP neural network, input the historical characteristic parameters and the structural parameters of the pier itself into the BP neural network, train the BP neural network, and judge whether the BP neural network is trained. If so, enter step S4; if not, repeat step S3;

[0011] S4. Obtain the characteristic parameters of the water area where the target bridge pier to be measured is located at the current moment, and input the dynamic parameter values into the trained BP neural network to obtain the local scour depth of the pier within a set future period.

[0012] Further, the historical characteristic parameters include fluid density, sediment incipient motion velocity, fluid kinematic viscosity, gravitational acceleration, sediment density, sediment particle size, approach flow depth, approach flow velocity, water flow direction, and Froude number.

[0013] Further, the structural parameters of the pier itself include pier shape coefficient, pier width, pier diameter, and pier axis angle.

[0014] Further, step S3 specifically includes: the BP neural network includes an input layer, a hidden layer, and an output layer;

[0015] Taking the sediment incipient motion velocity, approach flow velocity, pier diameter, and sediment particle size as the reference input quantities, input any one or any two or more of the historical characteristic parameters and the structural parameters of the pier itself combined with the reference input quantities into the input layer;

[0016] Judge the difference between the local scour depth of the pier output by the BP neural network and the actual local scour depth of the pier, and select the set of input quantities with the smallest difference as the final input quantity;

[0017] When the BP neural network is trained, select the characteristic parameters corresponding to the final input quantity from the characteristic parameters at the current moment as the current actual input, so as to predict the local scour depth of the pier within a set future period.

[0018] Further, the characteristic parameters and the structural parameters of the pier itself are normalized before being input into the BP neural network.

[0019] Correspondingly, the present invention also provides a prediction system for the local scour depth of a bridge pier, including a collection unit, a processing unit, and an output unit;

[0020] The acquisition unit is used to acquire the dynamic parameters of local scour of the bridge pier. The dynamic parameters include approach flow velocity, approach flow depth, sediment particle size, and water body viscosity.

[0021] The processing unit receives the dynamic parameters of local scour of the bridge pier output by the acquisition unit, determines the sediment incipient motion velocity and fluid specific weight according to the dynamic parameters, and the structural parameters of the bridge pier itself are preset in the processing unit. A BP neural network is established in the processing unit, and the processing unit determines the input parameters of the BP neural network and determines the local scour depth of the bridge pier in a future set time period according to the input parameters.

[0022] The output unit is connected to the processing unit and is used to output the local scour depth of the processing unit.

[0023] Furthermore, the acquisition unit includes a camera, a flow velocity sensor, a water depth sensor, and a viscosity sensor.

[0024] Furthermore, the processing unit includes a first processing chip, a second processing chip, and a positioning module.

[0025] The input end of the first processing chip is connected to the camera. The camera is used to acquire the sediment image in the water body and output it to the first processing chip. The first processing chip processes the sediment image and calculates the median particle size of the sediment and outputs it to the second processing chip.

[0026] The second processing chip has a BP neural network. The second processing chip is used to determine the dynamic parameters and the input parameters of the BP neural network, and predict the local scour depth of the bridge pier in a future set time period through the BP neural network.

[0027] The positioning module is used to acquire the current bridge pier position information and output it to the second processing chip. The second processing chip outputs the predicted local scour depth of the bridge pier and the current position information together.

[0028] Furthermore, the BP neural network includes an input layer, a hidden layer, and an output layer. The control chip determines the actual input of the BP neural network through the following process:

[0029] Taking the sediment incipient motion velocity, approach flow velocity, bridge pier diameter, and sediment particle size as the reference input quantities, combining any one or any two or more of the historical characteristic parameters and the structural parameters of the bridge pier itself with the reference input quantities to form input quantities and inputting them into the input layer;

[0030] Judging the difference between the local scour depth of the bridge pier output by the BP neural network and the actual local scour depth of the bridge pier, and selecting the set of input quantities with the smallest difference as the final input quantity;

[0031] After the BP neural network training is completed, select the characteristic parameters corresponding to the final input quantity from the characteristic parameters at the current moment as the current actual input, so as to predict the local scour depth of the bridge pier in the set future time period.

[0032] Furthermore, the output module is an Ethernet module or a wireless communication module.

[0033] The beneficial effects of the present invention: Through the present invention, by determining the relevant parameters affecting the local scour depth of the bridge pier, constructing a BP neural network, determining the corresponding parameters as the final input quantity, using the BP neural network, the influences of each parameter are analyzed in an associated manner rather than independent fixed-value calculations, and due to the error backpropagation characteristic of the BP neural network, the error of the final prediction is small and close to the actual value, so that the local scour depth of the bridge pier can be accurately predicted, providing accurate data support for the subsequent maintenance of the bridge. Description of the Drawings

[0034] The present invention will be further described below in conjunction with the drawings and embodiments:

[0035] Figure 1 It is a flow chart of the present invention.

[0036] Figure 2 It is a schematic diagram of the system structure of the present invention.

[0037] Figure 3 It is a schematic diagram of the specific embodiment structure of the system of the present invention. Detailed Embodiments

[0038] The following further elaborates on the present invention in detail:

[0039] A method for predicting the local scour depth of a bridge pier provided by the present invention is characterized in that it includes the following steps:

[0040] S1. Obtain the historical characteristic parameters of the water area where the target bridge pier is located and the structural parameters of the target bridge pier to be measured itself;

[0041] S2. Obtain the local scour depth of the bridge pier corresponding to the historical characteristic parameter acquisition time;

[0042] S3. Construct a BP neural network, input the historical characteristic parameters and the structural parameters of the bridge pier itself into the BP neural network, train the BP neural network, and judge whether the BP neural network training is completed. If so, enter step S4; if not, repeat step S3;

[0043] S4. Obtain the characteristic parameters of the water area where the bridge pier of the target bridge to be measured is located at the current moment, and input the dynamic parameter values into the trained BP neural network to obtain the local scour depth of the bridge pier within a set future period. Through the above method, by determining the relevant parameters affecting the local scour depth of the bridge pier, constructing a BP neural network, and determining the corresponding parameters as the final input quantities, the BP neural network is used to conduct a correlation analysis on the influences of each parameter, rather than independent fixed-value calculations. And due to the error backpropagation characteristic of the BP neural network, the error amount of the final prediction is small and close to the actual value, so as to accurately predict the local scour depth of the bridge pier and provide accurate data support for the subsequent maintenance of the bridge.

[0044] In this embodiment, the historical characteristic parameters include fluid specific weight, sediment incipient motion velocity, fluid kinematic viscosity, acceleration of gravity, sediment specific weight, sediment particle size, approach flow depth, approach flow velocity, water flow direction, and Froude number.

[0045] Among them, the bridge pier self-structure parameters include bridge pier shape coefficient, bridge pier width, bridge pier diameter, and pier axis included angle. Based on the selection of the above parameters, the subsequent prediction accuracy can be guaranteed.

[0046] In this embodiment, step S3 specifically includes: the BP neural network includes an input layer, a hidden layer, and an output layer; among them, the number of neurons set in the hidden layer is 15, and parameters such as the loss function of each layer can adopt existing technologies.

[0047] Taking the sediment incipient motion velocity, approach flow velocity, bridge pier diameter, and sediment particle size as the reference input quantities, combine any one or any two or more of the historical characteristic parameters and the bridge pier self-structure parameters with the reference input quantities to form input quantities and input them into the input layer.

[0048] Judge the difference between the local scour depth of the bridge pier output by the BP neural network and the actual local scour depth of the bridge pier, and select the set of input quantities with the smallest difference as the final input quantity.

[0049] When the BP neural network is trained, select the characteristic parameters corresponding to the final input quantity from the characteristic parameters at the current moment as the current actual input, so as to predict the local scour depth of the bridge pier in a set future period. For example: when the final input quantity during training is 6 parameters including sediment incipient motion velocity, approach flow velocity, bridge pier diameter, sediment particle size, bridge pier shape coefficient, and fluid kinematic viscosity, then among the characteristic parameters at the current moment, also select these 6 parameters as the final actual input quantities. Through the above method, the final prediction result can be made more accurate.

[0050] In this embodiment, the characteristic parameters and the structural parameters of the bridge pier itself are normalized before being input into the BP neural network. By the above method, the influence of different dimensions of each parameter on the BP neural network is avoided, thus ensuring the final prediction accuracy.

[0051] Correspondingly, the present invention also provides a prediction system for the local scour depth of a bridge pier, including a collection unit, a processing unit, and an output unit;

[0052] The collection unit is used to collect the dynamic parameters of the local scour of the bridge pier. The dynamic parameters include the approaching flow velocity, the approaching flow depth, the sediment particle size, and the water body viscosity;

[0053] The processing unit receives the dynamic parameters of the local scour of the bridge pier output by the collection unit, determines the sediment incipient motion velocity and the fluid specific weight according to the dynamic parameters, and the structural parameters of the bridge pier itself are preset in the processing unit; a BP neural network is established in the processing unit, and the processing unit determines the input parameters of the BP neural network and determines the local scour depth of the bridge pier in a future set time period according to the input parameters;

[0054] The output unit is connected to the processing unit and is used to output the local scour depth of the processing unit.

[0055] In this embodiment, the collection unit includes a camera, a flow velocity sensor, a water depth sensor, and a viscosity sensor. The flow velocity sensor is used to obtain the approaching flow velocity, the water depth sensor is used to obtain the approaching flow depth, the viscosity sensor is used to determine the actual viscosity of the water body, and for the specific weight, the density of the water body is determined by a density sensor, and then the specific weight is determined by the product of the density and the acceleration of gravity. The sediment incipient motion velocity is related to three parameters: the median particle size of the sediment, the density of the water body, and the water depth. Based on these three parameters and existing calculation formulas, the sediment incipient motion velocity can be determined. [[ID=]16]

[0056] In this embodiment, the processing unit includes a first processing chip, a second processing chip, and a positioning module;

[0057] The first processing chip has its input end connected to a camera. The camera is used to acquire sediment images in the water body and output them to the first processing chip. The first processing chip processes the sediment images, calculates the median particle size of the sediment, and outputs it to the second processing chip. The first processing chip is used for image processing. Among them, image processing includes image preprocessing, edge extraction, and particle size calculation. Image preprocessing is used to enhance the brightness of the image. Generally, the brightness of the images taken in the water body will be affected, so enhancement processing is required to avoid loss of edge features. Then edge feature extraction is performed, and finally particle size calculation is carried out. Finally, the median particle size of the sediment is obtained. Edge feature extraction can use Sobel operator, Roberts operator, Canny operator, etc. based on differential algorithms, or can also use methods such as relaxation method and wavelet transform method;

[0058] The second processing chip has a BP neural network. The second processing chip is used to determine the dynamic parameters and the input parameters of the BP neural network, and predict the local scour depth of the bridge pier within a set future time period through the BP neural network;

[0059] The positioning module is used to acquire the current bridge pier position information and output it to the second processing chip. The second processing chip outputs the predicted local scour depth of the bridge pier and the current position information together. Among them, the positioning module is used to acquire the current position information. For example, it can be achieved by using a GPS positioning module or a Beidou satellite positioning module, which is convenient for formulating subsequent bridge maintenance measures.

[0060] In this embodiment, the BP neural network includes an input layer, a hidden layer, and an output layer; the control chip determines the actual input of the BP neural network through the following process:

[0061] Taking the sediment incipient velocity, approach velocity, bridge pier diameter, and sediment particle size as the reference input quantities, combine any one or any two or more of the historical characteristic parameters and the bridge pier's own structural parameters with the reference input quantities to form input quantities and input them into the input layer;

[0062] Judge the difference between the local scour depth of the bridge pier output by the BP neural network and the actual local scour depth of the bridge pier, and select the set of input quantities with the smallest difference as the final input quantity;

[0063] When the BP neural network training is completed, select the characteristic parameters corresponding to the final input quantity from the characteristic parameters at the current moment as the current actual input, so as to predict the local scour depth of the bridge pier in the set future time period

[0064] In this embodiment, the output module is an Ethernet module or a wireless communication module. For example, the wireless communication module can be a 5G module, a UWB module, etc. Among them, the output module is used to transmit the information output by the first processing chip to the monitoring host or server of the monitoring center, and display it through the display interface connected to the monitoring host or server, so as to perform corresponding observation and early warning.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting the local scour depth of bridge piers, characterized in that: It includes the following steps: S1. Obtain the historical characteristic parameters of the water area where the target bridge pier is located and the structural parameters of the target bridge pier to be measured itself; S2. Obtain the local scour depth of the pier corresponding to the time when the historical characteristic parameters are collected; S3. Construct a BP neural network, input the historical characteristic parameters and the structural parameters of the pier itself into the BP neural network, train the BP neural network, and determine whether the BP neural network is trained. If so, enter step S4; if not, repeat step S3; S4. Obtain the characteristic parameters of the water area where the target bridge pier to be measured is located at the current moment, and input the dynamic parameter values into the trained BP neural network to obtain the local scour depth of the pier within a set future time period.

2. The method for predicting the local scour depth of a bridge pier according to claim 1, characterized in that: The historical characteristic parameters include fluid density, sediment incipient motion velocity, fluid kinematic viscosity, gravitational acceleration, sediment density, sediment particle size, approach flow depth, approach flow velocity, water flow direction, and Froude number.

3. The method for predicting the local scour depth of a bridge pier according to claim 1, characterized in that: The structural parameters of the pier itself include pier shape coefficient, pier width, pier diameter, and pier axis angle.

4. The method for predicting the local scour depth of a bridge pier according to claim 1, characterized in that: Specifically included in step S3: The BP neural network includes an input layer, a hidden layer, and an output layer; Taking the sediment incipient motion velocity, approach flow velocity, pier diameter, and sediment particle size as the reference input quantities, input any one or any two or more of the historical characteristic parameters and the structural parameters of the pier itself combined with the reference input quantities into the input layer; Judge the difference between the local scour depth of the pier output by the BP neural network and the actual local scour depth of the pier, and select the set of input quantities with the smallest difference as the final input quantity; When the BP neural network is trained, select the characteristic parameter corresponding to the final input quantity from the characteristic parameters at the current moment as the current actual input, so as to predict the local scour depth of the pier within a set future time period.

5. The method for predicting the local scour depth of a bridge pier according to claim 4, characterized in that: The characteristic parameters and the structural parameters of the pier itself are normalized before being input into the BP neural network.

6. A bridge pier local scour depth prediction system based on the method according to any one of claims 1-5, characterized in that: It includes a collection unit, a processing unit, and an output unit; The collection unit is used to collect the dynamic parameters of the local scour of the pier. The dynamic parameters include approach flow velocity, approach flow depth, sediment particle size, and water body viscosity; The processing unit receives the dynamic parameters of the local scour of the pier output by the collection unit, determines the sediment incipient motion velocity and fluid density according to the dynamic parameters, and the structural parameters of the pier itself are preset in the processing unit; a BP neural network is established in the processing unit, and the input parameters of the BP neural network are determined by the processing unit and the local scour depth of the pier within a set future time period is determined according to the input parameters; The output unit is connected to the processing unit and is used to output the local scour depth of the processing unit.

7. The bridge pier local scour depth prediction system according to claim 6, characterized in that: The collection unit includes a camera, a flow velocity sensor, a water depth sensor, and a viscosity sensor.

8. The bridge pier local scour depth prediction system according to claim 7, characterized in that: The processing unit includes a first processing chip, a second processing chip, and a positioning module; The input end of the first processing chip is connected to the camera. The camera is used to obtain the sediment image in the water body and output it to the first processing chip. The first processing chip processes the sediment image and calculates the median particle size of the sediment and outputs it to the second processing chip; The second processing chip has a BP neural network. The second processing chip is used to determine the dynamic parameters and the input parameters of the BP neural network, and predict the local scour depth of the bridge pier within a set future time period through the BP neural network; The positioning module is used to obtain the current position information of the bridge pier and output it to the second processing chip, and the second processing chip outputs the predicted local scour depth of the bridge pier and the current position information together.

9. The bridge pier local scour depth prediction system according to claim 8, wherein: The BP neural network includes an input layer, a hidden layer and an output layer; the control chip determines the actual input of the BP neural network through the following process: Taking the sediment incipient velocity, approach velocity, bridge pier diameter, and sediment particle size as the reference input quantities, combining any one or any two or more of the historical characteristic parameters and the bridge pier's own structural parameters with the reference input quantities to form input quantities and input them into the input layer; Judging the difference between the local scour depth of the bridge pier output by the BP neural network and the actual local scour depth of the bridge pier, and selecting the set of input quantities with the smallest difference as the final input quantity; After the BP neural network is trained, select the characteristic parameter corresponding to the final input quantity from the characteristic parameters at the current moment as the current actual input, so as to predict the local scour depth of the bridge pier in the set future time period.

10. The bridge pier local scour depth prediction system according to claim 6, characterized in that: The output module is an Ethernet module or a wireless communication module.

Images