Bridge crack development prediction method and device, equipment, storage medium and program product

Abstract

The invention relates to the technical field of bridge monitoring, provides a bridge crack development prediction method and device, equipment, a storage medium and a program product, and can predict the crack development condition more reasonably. According to the method, the cracking occurrence time of the target bridge is predicted according to the cracking data set of the bridge type to which the target bridge belongs, and the cracking time node of the target bridge is obtained; when the cracking time node arrives, acquiring state observation data of a crack on the target bridge at the cracking time node and disease observation data of a non-cracking disease on the target bridge at the cracking time node; and according to the state observation data of the crack at the cracking time node and the disease observation data of the non-cracking disease at the cracking time node, obtaining state prediction data of the crack at the next time node.

Description

Technical Field

[0001] This application relates to the field of bridge monitoring technology, and in particular to a method, device, computer equipment, storage medium and computer program product for predicting the development of bridge cracks. Background Technology

[0002] Bridge defects can be categorized by shape into cracked defects (hereinafter referred to as cracks) and non-cracked defects (defects other than cracks). Cracks have a relatively small area, making them more predictable at the physical level compared to other defects. However, because bridges are complex mechanical systems, predicting the development of cracks is quite difficult, making it challenging to accurately forecast their progression. Summary of the Invention

[0003] Therefore, it is necessary to provide a method, device, computer equipment, storage medium, and computer program product for predicting the development of bridge cracks, in order to address the above-mentioned technical problems.

[0004] This application provides a method for predicting the development of cracks in bridges, the method comprising:

[0005] Based on the cracking dataset of the bridge type to which the target bridge belongs, the timing of the cracking of the target bridge is predicted to obtain the cracking time node of the target bridge.

[0006] When the cracking time point arrives, acquire the status observation data of the cracks on the target bridge at the cracking time point and the disease observation data of non-cracking defects on the target bridge at the cracking time point.

[0007] Based on the state observation data of the crack at the cracking time point and the disease observation data of the non-cracking type at the cracking time point, the state prediction data of the crack at the next time point is obtained.

[0008] In one embodiment, based on the state observation data of the crack at the cracking time point and the disease observation data of the non-cracking type at the cracking time point, the state prediction data of the crack at the next time point is obtained, including:

[0009] Obtain a pre-built crack development prediction model;

[0010] The state observation data of the crack at the cracking time point and the disease observation data of the non-cracking disease at the cracking time point are input into the crack development prediction model.

[0011] Based on the output of the crack development prediction model, the state prediction data of the crack at the next time node is obtained.

[0012] In one embodiment, based on the output of the crack development prediction model, the predicted state data of the crack at the next time node is obtained, including:

[0013] Based on the output of the crack development prediction model, state change prediction data are obtained;

[0014] Based on the state observation data of the crack at the cracking time node and the state change prediction data, the state prediction data of the crack at the next time node is obtained.

[0015] In one embodiment, the method further includes:

[0016] The actual service time of the bridge training sample and the state observation data of the target crack on the bridge training sample at the target time node are obtained; the bridge training sample is a bridge of the same type as the target bridge that has completed service.

[0017] Based on the finite element model of the bridge with pure cracks corresponding to the bridge training sample, the theoretical service time of the bridge training sample affected by pure cracks and the theoretical state data of the target crack at the target time node are calculated; the theoretical service time is the service time of the bridge training sample under the influence of pure cracks; the theoretical state data is the state data of the target crack at the target time node that is not affected by non-cracking defects.

[0018] Based on the state observation data and state theory data of the target crack at the target time node, the state change data of the target crack are obtained;

[0019] The difference in service time of the bridge training sample is obtained based on the theoretical service time affected by pure cracks and the actual service time.

[0020] Training input data is obtained based on the theoretical data of the target crack at the target time node and the observation data of non-cracking defects on the bridge training sample at the target time node.

[0021] Training label data is obtained based on the state change data of the target crack and the service time difference;

[0022] Based on the training input data and the training label data, training data corresponding to the bridge training samples is generated and trained to obtain a crack development prediction model.

[0023] In one embodiment, training data corresponding to the bridge training samples is generated based on the training input data and the training label data, and then trained to obtain a crack development prediction model, including:

[0024] Based on the training input data and the training label data, training data corresponding to the bridge training samples is formed and trained to obtain the neural network to be verified.

[0025] The verification input data from the bridge verification sample is input into the neural network to be verified to obtain state change prediction data and service time prediction difference, so as to synthesize a finite element model of a bridge with defects and cracks.

[0026] Based on the finite element model of the bridge with defects, the theoretical service time of the bridge verification sample affected by cracks and non-cracking defects is obtained.

[0027] The scaling ratio is obtained based on the theoretical and actual service time of the bridge verification sample affected by cracks and non-cracking defects.

[0028] When the neural network to be verified is determined to pass the verification based on the scaling ratio, a crack development prediction model is obtained.

[0029] In one embodiment, the method provided in this application further includes:

[0030] After the target bridge cracks, the cracking data of the target bridge is added to the corresponding cracking dataset.

[0031] This application provides a device for predicting the development of cracks in bridges, the device comprising:

[0032] The cracking prediction module is used to predict the timing of cracking of the target bridge based on the cracking dataset of the bridge type to which the target bridge belongs, and to obtain the cracking time node of the target bridge.

[0033] The observation data acquisition module is used to acquire the state observation data of cracks on the target bridge at the cracking time node and the disease observation data of non-cracking defects on the target bridge at the cracking time node when the cracking time node arrives.

[0034] The crack development prediction module is used to obtain the crack's state prediction data at the next time node based on the crack's state observation data at the cracking time node and the non-cracking disease's disease observation data at the cracking time node.

[0035] This application provides a computer device, including a memory and a processor, wherein the memory stores a computer program and the processor executes the above-described method.

[0036] This application provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor using the methods described above.

[0037] This application provides a computer program product having a computer program stored thereon, the computer program being executed by a processor using the above-described method.

[0038] The aforementioned bridge crack development prediction method, device, computer equipment, storage medium, and computer program product predict the timing of cracking of the target bridge based on a cracking dataset of the bridge type to which the target bridge belongs, thus obtaining the cracking time node of the target bridge. When the cracking time node arrives, the system acquires state observation data of the cracks on the target bridge at the cracking time node, as well as disease observation data of non-cracking defects on the target bridge at the cracking time node. Based on the state observation data of the cracks at the cracking time node and the disease observation data of non-cracking defects at the cracking time node, the system obtains the state prediction data of the cracks at the next time node. The solution provided in this application takes the crack development of bridges as the main analysis object. Based on the crack mechanism and actual observation data, and on this basis, the influence of noise from non-cracking defects is introduced to obtain more reasonable state prediction data of the cracks at the next time node, thereby forming a more reasonable crack development prediction result. Furthermore, this application uses the time node of crack appearance (i.e., the cracking time node) as the prediction starting point for crack development of similar bridges, rather than the service start point of non-bridges as the actual data starting point, which can effectively remove the noise caused by crack development under normal service conditions and obtain more accurate crack development prediction results. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart illustrating a method for predicting the development of bridge cracks in one embodiment.

[0041] Figure 2(a) shows the cracking defects of the inclined section of the beam in one embodiment;

[0042] Figure 2(b) shows another cracking defect on the oblique section of the beam in one embodiment;

[0043] Figure 3(a) is a schematic diagram of the recursive relationship of multiple terms in one embodiment;

[0044] Figure 3(b) is a flowchart illustrating the cracking prediction process in one embodiment;

[0045] Figure 4 A structural block diagram of a bridge crack development prediction device in one embodiment;

[0046] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0048] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0049] The bridge crack development prediction method provided in this application can be used in a concrete bridge post-cracking service performance analysis system. This method includes... Figure 1 The steps shown can be performed by computer equipment.

[0050] Step S101: Based on the cracking dataset of the bridge type to which the target bridge belongs, predict the timing of the cracking of the target bridge to obtain the cracking time node of the target bridge.

[0051] Among bridges currently in service, some require crack development prediction, and these bridges can be used as target bridges. These bridges can be concrete bridges.

[0052] The pre-crack bridge data acquisition unit can record data about a cracked bridge before the cracking occurs, thus obtaining cracking data. Based on the cracking data of cracked bridges of the same type, a cracking dataset for that bridge type can be obtained. From the cracking datasets of multiple bridge types, the cracking dataset of the target bridge type can be determined. Based on this cracking dataset, the timing of the target bridge's cracking can be predicted, thus obtaining the cracking time node of the target bridge.

[0053] Step S102: When the cracking time node arrives, acquire the status observation data of the cracks on the target bridge at the cracking time node and the disease observation data of non-cracking defects on the target bridge at the cracking time node.

[0054] When the cracking time point arrives, the post-cracking bridge data acquisition unit can record the status observation data of the cracks on the target bridge at the cracking time point, such as the observed crack location and width. The post-cracking bridge data acquisition unit can also collect observation data on non-cracking defects on the target bridge at the cracking time point, such as the defect type, location, and quantitative values ​​of defect severity (e.g., carbonization depth, corrosion rate).

[0055] Step S103: Based on the state observation data of cracks at the cracking time node and the disease observation data of non-cracking diseases at the cracking time node, obtain the state prediction data of cracks at the next time node.

[0056] Once the status observation data of cracks at the cracking time point and the disease observation data of non-cracking diseases at the cracking time point are obtained, the service status analysis unit can be called. The service status analysis unit can predict crack development based on the status observation data of cracks at the cracking time point, the disease observation data of non-cracking diseases at the cracking time point, and the pre-built crack development prediction model, and obtain the status prediction data of cracks at the next time point, such as the predicted crack location and the predicted crack width.

[0057] In the aforementioned method for predicting the development of bridge cracks, the timing of cracking on the target bridge is predicted based on the cracking dataset of the bridge type to which the target bridge belongs, thus obtaining the cracking time node. When the cracking time node arrives, the state observation data of the crack on the target bridge at the cracking time node and the disease observation data of non-cracking defects on the target bridge at the cracking time node are acquired. Based on the state observation data of the crack at the cracking time node and the disease observation data of non-cracking defects at the cracking time node, the state prediction data of the crack at the next time node is obtained. The solution provided in this application takes the development of bridge cracks as the main analysis object. Based on the crack mechanism and actual observation data, the influence of noise from non-cracking defects is introduced to obtain more reasonable state prediction data of the crack at the next time node, thereby forming a more reasonable crack development prediction result. Furthermore, this application uses the time node of crack appearance (i.e., the cracking time node) as the prediction starting point for the development of cracks in similar bridges, rather than the service start point of non-bridges as the actual data starting point. This can effectively remove the noise caused by crack development under normal service conditions and obtain a more accurate crack development prediction result.

[0058] In one embodiment, based on the observation data of the crack's condition at the cracking time point and the observation data of non-cracking defects at the cracking time point, the predicted condition data of the crack at the next time point is obtained, including:

[0059] Obtain a pre-built crack development prediction model; input the crack status observation data at the cracking time node and the non-cracking disease observation data at the cracking time node into the crack development prediction model; based on the output of the crack development prediction model, obtain the crack status prediction data at the next time node.

[0060] Since there are many factors affecting noise in non-cracking diseases, the model can be trained based on the impact of past noise data on cracks. The influence of non-cracking diseases on crack development can be established through mathematical relationships, and a crack development prediction model can be constructed to avoid the problem of mechanistic level being difficult to simulate or directional omission caused by overly complex influencing factors.

[0061] The data on the state of cracks at the cracking time point and the data on the condition of non-cracking diseases at the cracking time point are used as input data and fed into the crack development prediction model.

[0062] The crack development prediction model analyzes and processes the input data to obtain the output results, thereby obtaining the crack state prediction data at the next time node. It can also obtain the difference in the service time prediction of the target bridge based on the output results.

[0063] In this embodiment, a pre-constructed crack development prediction model is used to predict crack development by combining the state observation data of cracks at the cracking time node and the disease observation data of non-cracking diseases at the cracking time node. The crack development prediction model can be trained based on the influence of past noise data on cracks. It describes the influence of non-cracking diseases on crack development through mathematical relationships, which can avoid the problem of the mechanism level being difficult to simulate or the problem of directional omission caused by the overly complex influencing factors, thus improving the accuracy of crack development prediction.

[0064] In one embodiment, based on the output of the crack development prediction model, the predicted state data of the crack at the next time node is obtained, including:

[0065] Based on the output of the crack development prediction model, the state change prediction data is obtained; based on the crack state observation data and state change prediction data at the crack initiation time node, the crack state prediction data at the next time node is obtained.

[0066] The crack development prediction model analyzes and processes the input data to obtain output results. These output results can include predicted crack state changes, such as the extent of changes in crack location and crack width. Based on the observed crack state data and the predicted state change data at the crack initiation time point, the predicted crack state data for the next time point is obtained.

[0067] In this embodiment, the crack development prediction model predicts the state changes of the crack, forming state change prediction data. Combined with the state observation data of the crack at the cracking time node, a more reasonable state prediction data of the crack at the next time node can be obtained, thus improving the accuracy of crack development prediction.

[0068] In one embodiment, the method provided in this application further includes:

[0069] The process involves acquiring the actual service time of bridge training samples and the state observation data of target cracks on the bridge training samples at the target time node. The bridge training samples are completed bridges of the same type as the target bridge. Based on the finite element model of the pure crack bridge corresponding to the bridge training samples, the theoretical service time of the bridge training samples affected by pure cracks and the theoretical state data of the target cracks at the target time node are calculated. The theoretical service time is the service time of the bridge training samples under the influence of pure cracks; the theoretical state data is the state data of the target cracks at the target time node without being affected by non-cracking defects. Based on the state observation data and theoretical state data of the target cracks at the target time node, the state change data of the target cracks is obtained. Based on the theoretical service time and actual service time of the bridge training samples affected by pure cracks, the service time difference of the bridge training samples is obtained. Based on the theoretical state data of the target cracks at the target time node and the defect observation data of non-cracking defects on the bridge training samples at the target time node, training input data is obtained. Based on the state change data and service time difference of the target cracks, training label data is obtained. Based on the training input data and training label data, training data corresponding to the bridge training samples is formed and trained to obtain a crack development prediction model.

[0070] The system can utilize a defect noise unit, which, based on historical data, assesses the impact of non-cracking defects on the development of cracking defects. Specifically, among completed bridges of the same type as the target bridge, bridges with multiple defects and cracks are selected based on historical data (this can be called bridge training samples). A dataset is formed based on defect types and locations. A basic crack distance unit is defined, radiating outwards at a distance A from the crack edge. If non-cracking defects are present within distance A and do not exist outside the crack radiation distance, the data is distilled into a new dataset, and the actual service time (denoted as T) of the corresponding bridge is obtained.

[0071] Using dt as the time interval, the total actual service time of the bridge training samples is divided into n segments, resulting in a time node sequence. The i-th time node can be represented as t. i Let i be a positive integer from 0 to n, and let t be the i-th time node. i+1 With the i-th time node t i The relationship is: t i+1 =ti +dt.

[0072] Using the i-th time node t i This will be illustrated using a target time point and an example of a reinforced concrete bridge.

[0073] Establish a solid finite element model of the load field of the reinforced concrete beam and obtain t i The distribution and width of cracks in the beam under load at specific time points are analyzed. Specifically, a crack analysis model is constructed, and the tensile constitutive equation of the concrete is calculated, as shown in the following formula:

[0074] ;

[0075] The formula for calculating crack width is as follows:

[0076] ;

[0077] Among them, E c ε is the elastic modulus of concrete. c For the tensile strain of concrete, ε cr ε represents the peak tensile strain of the concrete. e For the elastic tensile strain of concrete, For the tensile stress in concrete, f t w represents the tensile strength of concrete. cr L represents the width of the concrete crack. e The characteristic length of the concrete element is in the cracking direction.

[0078] The steel reinforcement section is divided into 4 elements, which share a common node at the center of the section. The axial direction is the length of the steel reinforcement and the normal direction is the outer surface of the steel reinforcement. The axial stiffness of the element is taken according to the elastic modulus of the steel reinforcement, and the normal stiffness is taken as 0.

[0079] Static analysis is performed by applying loads to the solid element model of the reinforced concrete beam, solving for the tensile strain of the elements, and outputting the crack width w of the concrete beam solid element through the crack width of the concrete solid element. cr The crack width field variable includes: crack location (which can be characterized by crack coordinates) and crack width.

[0080] The crack width field variable data, including crack location, crack width, and tensile stress, are imported into the pure crack bridge finite element model corresponding to the bridge training sample. Based on the pure crack bridge finite element model, the expected service time of the bridge training sample under the influence of pure cracks and without the influence of non-cracking defects is calculated. This time is called the theoretical service time of the bridge training sample under the influence of pure cracks.

[0081] Based on the crack width field variable data, including crack location and crack width, the target crack of the bridge training sample at time node t can be obtained. i The state theory data. The target crack is any crack on the bridge training sample, and the state theory data is the target crack at the i-th time node t. i Data on the expected state before being affected by non-cracking diseases.

[0082] Based on historical data from bridge training samples, the target crack in the bridge training sample at time node t can be obtained. i Condition observation data, such as crack location and crack width.

[0083] Based on the target crack at time node t i Based on state observation data and state theory data, the target crack at time node t is obtained. i State change data.

[0084] Based on the difference between the theoretical service time and the actual service time of the bridge training sample affected by pure cracks, the service time difference of the bridge training sample is obtained. The difference is the cumulative result of the effects of non-cracking defects.

[0085] The target crack can be located at time node t. i The theoretical data of the state and the non-cracking defects on the bridge training samples at the i-th time node t i The defect observation data is used as training input data. The service time difference of the bridge training samples and the target crack at the i-th time node t are used as training input data. i The state change data is used as training label data. Based on the training input data and training label data, bridge training samples can be formed at the i-th time node t. i The training data.

[0086] During training, the training input data is fed into the neural network to be trained. Based on the parameters within the neural network, its output is obtained, including the difference in service time prediction and state change prediction data. To minimize the difference between the output and the training label data, the parameters of the neural network are adjusted, resulting in the neural network to be validated. After the neural network to be validated passes validation, a crack development prediction model can be obtained.

[0087] In this embodiment, training is conducted using completed bridges of the same type as the target bridge as training samples to uncover the mapping relationship between various defects and bridge lifespan, which can then reasonably predict crack development.

[0088] In one embodiment, training data corresponding to bridge training samples is generated based on training input data and training label data, and then trained to obtain a crack development prediction model, including:

[0089] Based on the training input data and training label data, training data corresponding to the bridge training samples is generated and trained to obtain the neural network to be verified. The verification input data from the bridge verification samples is input into the neural network to be verified to obtain the state change prediction data and the service time prediction difference, so as to synthesize the finite element model of the bridge with defects and cracks. Based on the finite element model of the bridge with defects and cracks, the theoretical service time of the bridge verification samples affected by cracks and non-cracking defects is obtained. Based on the theoretical service time and actual service time of the bridge verification samples affected by cracks and non-cracking defects, the scaling ratio is obtained. When the neural network to be verified is determined to pass the verification based on the scaling ratio, the crack development prediction model is obtained.

[0090] During training, the training input data is fed into the neural network to be trained. Based on the parameters inside the neural network, the output of the neural network to be trained can be obtained. The output includes the difference in service time prediction and the state change prediction data. In order to minimize the difference between the output and the training label data, the parameters inside the neural network to be trained are adjusted, thereby obtaining the neural network to be verified.

[0091] Among bridges of the same type as the target bridge that have already completed service, those not used as training samples are selected to obtain bridge verification samples. Verification input data for these bridge verification samples is obtained using the same method as the training input data described above. This verification input data is then input into the neural network to be verified, yielding predicted crack state changes and the difference between the predicted service time and the predicted value of the bridge verification samples. This difference is used to synthesize a finite element model of the cracked bridge with defects. Based on this finite element model, the theoretical service time of the bridge verification samples under the influence of both non-cracking defects and cracks is obtained. The scaling ratio is then calculated based on the theoretical and actual service times of the bridge verification samples under the influence of cracks and non-cracking defects. A scaling ratio closer to 1 indicates a more accurate output from the neural network to be verified.

[0092] After obtaining the scaling ratios of several bridge validation samples, linear regression can be performed on all scaling ratios to obtain the target scaling ratio. The degree to which the target scaling ratio approaches 1 determines whether the neural network to be validated passes validation; the closer the target scaling ratio is to 1, the higher the probability that the neural network to be validated will pass validation. After the neural network to be validated passes validation, it can be used as a crack development prediction model.

[0093] In this embodiment, a finite element model of a bridge with defects is synthesized based on the output of the neural network to be verified. This model is used to obtain the theoretical service time of the bridge verification sample under the influence of non-cracking defects and cracks. The model is verified by scaling the theoretical service time with the actual service time. This can determine whether the model is developing in an accurate direction or an inaccurate direction. If the model is developing in an inaccurate direction (i.e., the target scaling ratio is not close to 1), denoising can be performed iteratively.

[0094] In one embodiment, the method provided in this application further includes:

[0095] After the target bridge cracks, the cracking data of the target bridge is added to the corresponding cracking dataset.

[0096] After a target bridge cracks, the data of the target bridge before the cracking occurred can be used as cracking data and added to the cracking dataset of the bridge type to which the target bridge belongs. This allows for updating of the cracking dataset and more accurate prediction of the timing of cracking in other bridges of the same type.

[0097] To better understand the above method, the following details an application example of the bridge crack development prediction method of this application.

[0098] Bridge defects can be categorized by shape into cracked defects (hereinafter referred to as cracks) and non-cracked defects (defects other than cracks). Cracks have a relatively small area, making them more predictable at the physical level compared to other defects. However, because bridges are complex mechanical systems, predicting crack development is challenging and often relies on mechanistic analysis. Since bridge fatigue is influenced by numerous factors, crack development can be affected by other defects. Therefore, a reliable method for predicting the post-cracking service performance of bridges, based on both mechanistic analysis and actual data, is needed.

[0099] Traditional research on the fatigue performance of simply supported concrete beams mainly focuses on the shear fatigue strength of beams with and without stirrups after cracking. Research on the crack resistance of stirrup-reinforced beams primarily concentrates on the calculation and analysis of cracking loads under different shear span ratios, prestressing forces, and cross-sectional dimensions. The understanding of the concrete fatigue failure mechanism under the principal stress of the inclined section remains unclear and in-depth. Systematic research on the fatigue crack resistance and post-cracking shear fatigue performance of concrete beams under ultra-high loads has not yet been conducted. Due to these issues, the overall study of bridges, as complex mechanical systems, currently lacks a mature understanding of the concrete fatigue failure mechanism, and the research results do not yet have industrial applications. Cracking defects in the inclined section of the beam are shown in Figures 2(a) and 2(b).

[0100] In actual service, bridges are large, complex mechanical systems composed of stress and strain components with single characteristics. This makes the factors influencing bridge defects quite complex inside the bridge, and there are many types of bridge defects. As a result, the development of cracks in bridges is affected by a variety of factors.

[0101] Some traditional techniques predict crack development based on overall simulation, but they fail to consider additional interfering factors, such as other defects, that may occur in bridges as complex mechanical systems beyond the given stress. While these traditional techniques are also dual-channel solutions, the data integration between the two channels is weak, and the final result is only a combination of the two predictions. When the actual results from the mechanism and the images are difficult to reconcile, this approach can lead to significant prediction errors. Due to these problems, it is currently difficult to gain a mature understanding of the concrete fatigue failure mechanism in bridge research, which deals with complex mechanical systems, and there are no industrially applicable products based on the research findings.

[0102] This application example uses the disease mechanism to predict the overall cracking of a bridge during its service life.

[0103] This application example includes a bridge data acquisition unit before cracking, a bridge data acquisition unit after cracking, a disease noise unit, a service status analysis unit, and a service prediction unit.

[0104] The pre-cracking bridge data acquisition unit is used to record the bridge stress and strain data before cracking.

[0105] The post-cracking bridge data acquisition unit is used to record the bridge stress-strain data and cracking data after the bridge cracks.

[0106] The defect noise unit is used to obtain the impact of non-cracking defects on the development of cracking defects based on past data. Specifically, among bridges that have been completed and put into service, similar to the target bridge, bridges with multiple defects and cracks are selected based on past data (this can be called bridge training samples). A dataset is formed according to the type and location of defects. A basic crack distance unit is set, radiating outward from the crack edge by a distance A. If there are non-cracking defects within distance A and these non-cracking defects are not located outside the crack radiation distance, the data is distilled into a new dataset, and the actual service time of the corresponding bridge (which can be denoted as T) is obtained.

[0107] The service status analysis unit collects historical information on similar bridges to obtain the actual service time of such bridges, removes bridges that have ended their service life except for cracks, and provides the relevant data of the remaining bridges as a basic dataset to the machine learning model for predicting service life.

[0108] Let t be the total actual service time of reinforced concrete beams of the same type. Divide the total actual service time into n segments with dt as the time interval to obtain a time node sequence. The i-th time node can be represented as t. i Let i be a positive integer from 0 to n, and let t be the i-th time node. i+1 With the i-th time node t i The relationship is: t i+1 =t i +dt.

[0109] Establish a solid finite element model of the load field of the reinforced concrete beam and obtain t i The distribution and width of cracks in the beam under load at specific time points are analyzed. Specifically, a crack analysis model is constructed, and the tensile constitutive equation of the concrete is calculated, as shown in the following formula:

[0110] ;

[0111] The formula for calculating crack width is as follows:

[0112] ;

[0113] Among them, E c ε is the elastic modulus of concrete. c For the tensile strain of concrete, ε cr ε represents the peak tensile strain of the concrete. e For the elastic tensile strain of concrete, For the tensile stress in concrete, f t w represents the tensile strength of concrete. cr L represents the width of the concrete crack. e The characteristic length of the concrete element is in the cracking direction.

[0114] The steel reinforcement section is divided into 4 elements, which share a common node at the center of the section. The axial direction is the length of the steel reinforcement and the normal direction is the outer surface of the steel reinforcement. The axial stiffness of the element is taken according to the elastic modulus of the steel reinforcement, and the normal stiffness is taken as 0.

[0115] Static analysis is performed by applying loads to the solid element model of the reinforced concrete beam, solving for the tensile strain of the elements, and outputting the crack width w of the concrete beam solid element through the crack width of the concrete solid element. cr The crack width field variable includes: crack location (which can be characterized by crack coordinates) and crack width.

[0116] The crack width field variable data, including crack location, crack width, and tensile stress, are imported into the pure crack bridge finite element model corresponding to the bridge training sample. Based on the pure crack bridge finite element model, the expected service time of the bridge training sample under the influence of pure cracks and without the influence of non-cracking defects is calculated. This time is called the theoretical service time of the bridge training sample under the influence of pure cracks. The difference between the actual service time and the theoretical service time is obtained, and the difference is used as the cumulative result of the defect influence to obtain the mapping relationship between the influence of various defects and service life.

[0117] The mapping relationship can be obtained using the Apriori association analysis algorithm. Since data confirmation based on detection results is required in practical applications, the principle of the Apriori association analysis algorithm is introduced here:

[0118] The Apriori association analysis algorithm is the most fundamental algorithm in association rule mining. Currently, most parallel data mining algorithms for association rules are based on the Apriori algorithm, giving it an irreplaceable and unique position. The Apriori association analysis algorithm essentially involves two main problems: first, identifying all frequent data item sets in a transactional database; and second, generating strong association rules.

[0119] The core idea of ​​the Apriori association analysis algorithm is a recursive method based on the theory of frequent itemsets. It uses a layer-by-layer search iterative method to mine all frequent itemsets in the target transaction database until the highest-order frequent itemset is found. Finally, strong association rules are obtained by calculating the obtained frequent itemsets.

[0120] For example, the recursive relationship of items 0, 1, 2 and 3 is shown in Figure 3(a). If 0, 1, 2 and 3 correspond to results of different physical properties, then Figure 3(a) shows all possible combinations of data (lattice structure).

[0121] In this application example, disease type, location information, and lifespan change value are all used as items in the basic set.

[0122] At this point, the above relationship should be evaluated based on three points:

[0123] 1. Support: The ratio of the number of times a data item appears to the total number of items in the dataset;

[0124] ;

[0125] 2. Confidence level: The probability that another data point will appear after a certain data point has been found;

[0126] The confidence level of x with respect to y can be expressed as:

[0127] ;

[0128] 3. Lift: The ratio of the probability of x occurring given y to the overall probability of x occurring;

[0129] ;

[0130] If the lift is greater than one, it means that y has a strong and effective association with x; if the lift is less than one, it means that y has an ineffective and strong association with x.

[0131] Based on the above three points, the relations existing in the above set items can take two forms:

[0132] Frequent item sets: A set of items that frequently appear in a group.

[0133] Association rules: imply a strong relationship between two items.

[0134] Support and confidence are methods used to quantify the success of association analysis. For example, given a set of only 4 items {0,1,2,3}, we want to obtain the support for each possible set. First, we need to list the number of possible combinations of the 4 items; there are a total of 15 combinations.

[0135] For example, to calculate the support of the itemset {0, 3}, we need to iterate through each record, checking if it contains both 0 and 3. If it does, we increment the count by 1. This will give us the support of the itemset {0, 3}. To obtain the support of every possible set, we need to repeat the above process.

[0136] For a dataset of N items, there are a total of n possible itemset combinations, resulting in a huge computational burden. To reduce the computation time, Apriori can be used to discover frequent itemsets.

[0137] If {0,1} is frequent, then {0} and {1} must also be frequent, because the support of {0} and {1} must be greater than or equal to that of {0,1}. Conversely, if an itemset is infrequent, then all its supersets are also infrequent.

[0138] If {2,3} is infrequent, then {0,2,3}, {1,2,3}, and {0,1,2,3} must also be infrequent, because the support of {2,3} must be greater than or equal to the support of its superset.

[0139] After determining the relationships between the disease status and the numerical changes of the data, it is possible to determine whether the data are strongly correlated or not (mutually exclusive, etc.). Based on the above correlations and the numerical changes of the data, a reasonable data change relationship matrix can be established.

[0140] Because the Apriori association analysis algorithm is slow on large datasets, it is necessary to pre-screen and preprocess the input data in practical applications. This will minimize the number of data with known associations entering the machine learning process as independent itemsets, thus reducing the number of itemsets processed by the algorithm. In this application example, static analysis models, dynamic analysis models, and reverse influence models are used to preprocess the data to avoid data complexity and mitigate the impact of data noise.

[0141] Based on the mapping relationship between lifespan and disease type, and the stress condition of isolated cracks, a finite element bridge is synthesized using the final disease source dataset. This data is then compared with the actual service time to obtain the scaling ratio. Linear regression is performed on all obtained scaling ratios to obtain the scaling relationship between the final theoretical and actual values ​​(i.e., the target scaling ratio).

[0142] The service prediction unit retrieves the corresponding cracking data from the service status analysis module based on the cracked bridge data obtained from the current bridge, and predicts the crack development status at the current crack location at the next time node.

[0143] This application example records data before cracking using a pre-cracking bridge data acquisition unit. This pre-cracking data is used to infer the timing of cracking. At the point when cracking occurs in the beam, the real-time data acquisition unit is switched from the pre-cracking bridge data acquisition unit to the post-cracking bridge data acquisition unit. This switching point is recorded, and the post-cracking bridge data acquisition unit begins collecting crack-related data. The collected crack data is sent to the service status analysis unit, which retrieves data from the defect noise unit, reconstructs the mechanism of the detected cracks, and predicts the cracking status at the next time point based on the service analysis results, as shown in Figure 3(b).

[0144] Since there are many factors affecting noise from disease, the noise in this application example is trained using the impact of past data on cracks. By establishing mathematical relationships and the influence of disease on crack development, we avoid the problem of mechanistic difficulty in simulation or the easy omission of direction caused by overly complex influencing factors.

[0145] The data collected by the pre-cracking bridge data acquisition unit and the post-cracking bridge data acquisition unit are mainly based on the following two aspects:

[0146] ①Bridge visual inspection:

[0147] The investigation focused on the cracking of the inclined sections of prestressed concrete simply supported T-beams, and statistically analyzed the beam types and cracking morphologies with the most severe inclined section cracking. Basic information such as mileage, bridge number, drawing number, and beam fabrication unit was recorded for typical cracked beams. The investigation also covered the condition of the bridge deck and ballast, as well as the condition of ancillary facilities such as sidewalks and sound barriers, and measured and recorded excessive ballast thickness and track eccentricity.

[0148] ② Concrete crack detection:

[0149] For beams with typical oblique cross-section cracking defects, concrete crack detection was performed. The depth of the oblique cracks was tested using ultrasonic methods, and verified using core sampling. Simultaneously, the carbonation depth of the core samples was tested. A rebound hammer was used to test the concrete strength at the web, and a rebar detector was used to test the location of the rebars and the thickness of the protective layer, thus assessing the beam construction quality.

[0150] For the two detection methods described above, the preprocessed image is divided into several small regions (cells). For each pixel within each cell, its grayscale value is compared with that of its neighboring pixels. If the grayscale value of a neighboring pixel is greater than or equal to that of the center pixel, the position of that neighboring pixel is marked as 1; otherwise, it is marked as 0. Each pixel then receives a binary number consisting of 0s and 1s. The binary number of each pixel is converted to a decimal number, and the resulting decimal number is the LBP (Local Binary Number) of that pixel. Pattern (Local Binary Pattern) values: In this step, it's important to note the application of uniform LBP patterns, i.e., only considering binary patterns with no more than two 0-to-1 or 1-to-0 transitions. This helps filter noise and focus on more representative texture features. The LBP values ​​for the entire cell are statistically analyzed to construct a histogram. Each bin in the histogram corresponds to an LBP value, and the bin value represents the frequency of that LBP value within the cell. The LBP histograms of all cells are concatenated to form the image's feature vector. The LBP feature vector can describe the local texture features of the image and has a significant effect on identifying and describing the texture features of bridge surface defects. By comparing the LBP features of different bridge defects, the type of bridge surface defect can be identified.

[0151] In this application example, crack coordinates, crack width, tensile stress, and defect noise are used as data dimensions to determine the neural network structure. The parameters of the neural network and the CPSO (Comprehensive Particle Swarm Optimization) algorithm are set, and the weights and thresholds of the BP (Back Propagation Neural Network) are optimized using the CPSO algorithm. To improve the fitting and prediction accuracy for highly volatile values, the Monte Carlo (MC) model is combined to enhance its ability to handle highly volatile stochastic processes and to correct the prediction results of the CPSO network prediction model. The interference factors affecting the development of various cracks in the bridge are calculated, and the development prediction data are calculated.

[0152] Furthermore, this application example could consider using deep learning to predict crack initiation points based on pre-crack stress-strain data features. Bridges already in service could be used as the training set, while bridges currently in operation could be used as the detection set.

[0153] In practical implementation, assuming that the neural network has N weights and thresholds to be optimized, then there are N particles in the population of the CPSO algorithm. The position vectors of the N particles represent the weights and thresholds of the neural network. Their dimensions are equal to the number of weights and thresholds. The error of the neural network is selected as the fitness value of the particle and the population optimal value. It can directly reflect the quality of the weights and thresholds searched by the particle during the movement, and thus find the optimal value of the particle and the population.

[0154] To improve the accuracy of fitting and predicting values ​​with large fluctuations, the MC model's ability to handle large-scale stochastic processes and the prediction results of the CPSO-BP network prediction model are combined and corrected.

[0155] Based on the previous year's bridge health status and the characteristic that the technical condition of each structure and component of the bridge will not decrease without maintenance, the range of deterioration changes of each structure and component of the bridge in the predicted year is derived, and the median value is taken as the specific predicted value. In addition, combined with the bridge defects in the previous year, the location, type and degree of bridge defects in the predicted year are analyzed.

[0156] This embodiment addresses the main defects of prestressed concrete simply supported T-beams, including: longitudinal cracking of the web, diagonal cracking of the web, surface cracking of the beam, separation, cracking, and whitening of the anchorage zone of the external prestressed reinforcement system at the bottom of the beam for heavy-duty vehicles, scraping and spalling of concrete on the bottom surface of the beam crossing the highway, drainage failure, beam contamination, and peeling of protective coating. The beams for light-duty vehicles have not yet undergone transverse reinforcement. The main defects of the piers include: loose, non-dense, cracked, vertical and transverse cracking of the concrete surface (or the reinforced surface), rust expansion and peeling of the protective layer, exposed reinforcement, peeling of the protective layer on the cap beam, exposed reinforcement, whitening, peeling of the protective layer at the bottom of the pier, exposed reinforcement, efflorescence, and whitening. The main defects of the abutments include: water seepage, whitening, and narrow beam joints. The main defects of the bridge deck ancillary facilities include: severe deterioration of the concrete sidewalks and pier top railings; whitening at the bottom joints of the prestressed concrete continuous beam segments; loose, non-dense, cracked, vertical and horizontal cracks on the pier surface; rust and peeling of the protective layer; exposed reinforcement; drainage failure; and pollution of the pier body.

[0157] Longitudinal cracking in the web of prestressed concrete simply supported T-beams has a significant impact on the durability of bridge structures and also poses certain adverse effects on structural safety.

[0158] This application example is based on cost considerations.

[0159] Bridge operating costs depend on the reliability of the bridge structure. Damage to the bridge structure affects the originally designed service life of the bridge, and the service life of the bridge determines its functional benefits. Several issues typically arise regarding operating costs:

[0160] (1) Bridges are often designed and built at the lowest possible initial cost while ensuring service requirements, thus ignoring the costs of repairing bridges in case of damage or failure during subsequent operation.

[0161] (2) Repairing bridges only after structural problems occur, rather than taking preventative measures, shortens their service life and increases resource and financial costs. Therefore, determining the optimal intervention point can not only control maintenance costs but also reduce significant changes made to the bridge by maintenance personnel.

[0162] Therefore, optionally, this application example may also include:

[0163] The cost and benefit acquisition unit is used to obtain various costs and benefits, and to construct cost formulas and benefit formulas respectively; the costs include: the installation and monitoring costs of the bridge structural health monitoring system, the maintenance and renovation costs during operation, and the bridge demolition costs; the benefits include bridge operation benefits and bridge recycling benefits;

[0164] The difference acquisition unit is used to calculate the difference between the total benefits and the total costs;

[0165] The results output unit is used to combine the difference with other factors of the bridge to obtain the analysis results, including the bridge's service life and structural reliability; the analysis results indicate whether to continue maintenance or demolition.

[0166] This study employs a degradation framework, combining cost-effectiveness assessments with the bridge's current condition and service life, to evaluate bridges and support decision-making. In situations where bridge performance significantly declines and maintenance costs increase in the later stages of operation, it provides theoretical support for deciding whether to continue repairing or demolishing bridges nearing the end of their life cycle that are still in service. This offers a reference for bridge operation and maintenance work, reducing maintenance costs, improving the utilization rate of bridge resources, and enabling the timely dismantling of bridges with severely compromised structural functions, thus preventing potential accidents.

[0167] This application example focuses on the development of bridge cracks. Based on the crack mechanism and actual data, it incorporates the influence of noise caused by defects to obtain reasonable predictive data for bridge crack development. This application example uses the time of crack appearance as the starting point for predicting crack development in similar bridges, rather than the bridge's service start point, to effectively remove noise caused by normal service conditions on crack development. This approach represents a technical solution for comprehensive, multi-location, multi-condition, and multi-characteristic monitoring and analysis of bridges based on data characteristics.

[0168] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0169] Based on the same inventive concept, this application also provides a bridge crack development prediction device for implementing the bridge crack development prediction method described above. The solution provided by this device is similar to the implementation scheme described in the above method; therefore, the specific limitations of one or more bridge crack development prediction device embodiments provided below can be found in the limitations of the bridge crack development prediction method described above, and will not be repeated here.

[0170] In one embodiment, such as Figure 4 As shown, a device for predicting the development of bridge cracks is provided, comprising:

[0171] The cracking prediction module 401 is used to predict the timing of cracking of the target bridge based on the cracking dataset of the bridge type to which the target bridge belongs, and to obtain the cracking time node of the target bridge.

[0172] The observation data acquisition module 402 is used to acquire the state observation data of the cracks on the target bridge at the cracking time node and the disease observation data of the non-cracking defects on the target bridge at the cracking time node when the cracking time node arrives.

[0173] The crack development prediction module 403 is used to obtain the crack's state prediction data at the next time node based on the crack's state observation data at the cracking time node and the non-cracking disease's disease observation data at the cracking time node.

[0174] In one embodiment, the crack development prediction module 403 is further configured to:

[0175] Obtain a pre-constructed crack development prediction model; input the state observation data of the crack at the cracking time node and the disease observation data of the non-cracking disease at the cracking time node into the crack development prediction model; based on the output of the crack development prediction model, obtain the state prediction data of the crack at the next time node.

[0176] In one embodiment, the crack development prediction module 403 is further configured to:

[0177] Based on the output of the crack development prediction model, state change prediction data is obtained; based on the state observation data of the crack at the cracking time node and the state change prediction data, state prediction data of the crack at the next time node is obtained.

[0178] In one embodiment, the apparatus further includes a crack development prediction model training module, used for:

[0179] The actual service time of the bridge training sample and the state observation data of the target crack on the bridge training sample at the target time node are obtained; the bridge training sample is a bridge of the same type as the target bridge that has completed service; based on the pure crack bridge finite element model corresponding to the bridge training sample, the theoretical service time of the bridge training sample affected by pure crack and the state theoretical data of the target crack at the target time node are calculated; the theoretical service time is the service time of the bridge training sample under the influence of pure crack; the state theoretical data is the state data of the target crack at the target time node that is not affected by non-cracking defects; based on the state observation data of the target crack at the target time node... Based on measured data and theoretical state data, the state change data of the target crack is obtained; based on the theoretical service time of the bridge training sample affected by pure cracks and the actual service time, the service time difference of the bridge training sample is obtained; based on the theoretical state data of the target crack at the target time node and the observation data of non-cracking defects on the bridge training sample at the target time node, training input data is obtained; based on the state change data of the target crack and the service time difference, training label data is obtained; based on the training input data and the training label data, training data corresponding to the bridge training sample is formed and trained to obtain a crack development prediction model.

[0180] In one embodiment, the crack development prediction model training module is used for:

[0181] Based on the training input data and the training label data, training data corresponding to the bridge training samples is generated and trained to obtain a neural network to be verified. The verification input data from the bridge verification samples is input into the neural network to be verified to obtain state change prediction data and service time prediction differences, thereby synthesizing a finite element model of a bridge with defects. Based on the finite element model of the bridge with defects, the theoretical service time of the bridge verification samples affected by cracks and non-cracking defects is obtained. Based on the theoretical service time and actual service time of the bridge verification samples affected by cracks and non-cracking defects, a scaling ratio is obtained. When the neural network to be verified passes verification based on the scaling ratio, a crack development prediction model is obtained.

[0182] In one embodiment, the apparatus further includes a cracked dataset update module, configured to:

[0183] After the target bridge cracks, the cracking data of the target bridge is added to the corresponding cracking dataset.

[0184] The modules in the aforementioned bridge crack development prediction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0185] In one exemplary embodiment, a computer device is provided, the internal structure of which can be as shown in the figure. Figure 5 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores the data involved in the aforementioned methods. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for predicting the development of bridge cracks.

[0186] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0187] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in the various method embodiments described above.

[0188] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the various method embodiments described above.

[0189] In one embodiment, a computer program product is provided having a computer program stored thereon, the computer program being executed by a processor of the steps described in the various method embodiments above.

[0190] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0191] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0192] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0193] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for predicting the development of cracks in bridges, characterized in that, The method includes: Based on the cracking dataset of the bridge type to which the target bridge belongs, the timing of the cracking of the target bridge is predicted to obtain the cracking time node of the target bridge. When the cracking time point arrives, acquire the status observation data of the cracks on the target bridge at the cracking time point and the disease observation data of non-cracking defects on the target bridge at the cracking time point. Based on the state observation data of the crack at the cracking time point and the disease observation data of the non-cracking type at the cracking time point, the state prediction data of the crack at the next time point is obtained.

2. The method according to claim 1, characterized in that, Based on the observation data of the crack's condition at the cracking time point and the observation data of the non-cracking defects at the cracking time point, the predicted condition data of the crack at the next time point is obtained, including: Obtain a pre-built crack development prediction model; The state observation data of the crack at the cracking time point and the disease observation data of the non-cracking disease at the cracking time point are input into the crack development prediction model. Based on the output of the crack development prediction model, the state prediction data of the crack at the next time node is obtained.

3. The method according to claim 2, characterized in that, Based on the output of the crack development prediction model, the predicted state data of the crack at the next time node is obtained, including: Based on the output of the crack development prediction model, state change prediction data are obtained; Based on the state observation data of the crack at the cracking time node and the state change prediction data, the state prediction data of the crack at the next time node is obtained.

4. The method according to claim 3, characterized in that, The method further includes: The actual service time of the bridge training sample and the state observation data of the target crack on the bridge training sample at the target time node are obtained; the bridge training sample is a bridge of the same type as the target bridge that has completed service. Based on the finite element model of the bridge with pure cracks corresponding to the bridge training sample, the theoretical service time of the bridge training sample affected by pure cracks and the theoretical state data of the target crack at the target time node are calculated; the theoretical service time is the service time of the bridge training sample under the influence of pure cracks; the theoretical state data is the state data of the target crack at the target time node that is not affected by non-cracking defects. Based on the state observation data and state theory data of the target crack at the target time node, the state change data of the target crack are obtained; The difference in service time of the bridge training sample is obtained based on the theoretical service time affected by pure cracks and the actual service time. Training input data is obtained based on the theoretical data of the target crack at the target time node and the observation data of non-cracking defects on the bridge training sample at the target time node. Training label data is obtained based on the state change data of the target crack and the service time difference; Based on the training input data and the training label data, training data corresponding to the bridge training samples is generated and trained to obtain a crack development prediction model.

5. The method according to claim 4, characterized in that, Based on the training input data and the training label data, training data corresponding to the bridge training samples is generated and trained to obtain a crack development prediction model, including: Based on the training input data and the training label data, training data corresponding to the bridge training samples is formed and trained to obtain the neural network to be verified. The verification input data from the bridge verification sample is input into the neural network to be verified to obtain state change prediction data and service time prediction difference, so as to synthesize a finite element model of a bridge with defects and cracks. Based on the finite element model of the bridge with defects, the theoretical service time of the bridge verification sample affected by cracks and non-cracking defects is obtained. The scaling ratio is obtained based on the theoretical and actual service time of the bridge verification sample affected by cracks and non-cracking defects. When the neural network to be verified is determined to pass the verification based on the scaling ratio, a crack development prediction model is obtained.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: After the target bridge cracks, the cracking data of the target bridge is added to the corresponding cracking dataset.

7. A device for predicting the development of bridge cracks, characterized in that, The device includes: The cracking prediction module is used to predict the timing of cracking of the target bridge based on the cracking dataset of the bridge type to which the target bridge belongs, and to obtain the cracking time node of the target bridge. The observation data acquisition module is used to acquire the state observation data of cracks on the target bridge at the cracking time node and the disease observation data of non-cracking defects on the target bridge at the cracking time node when the cracking time node arrives. The crack development prediction module is used to obtain the crack's state prediction data at the next time node based on the crack's state observation data at the cracking time node and the non-cracking disease's disease observation data at the cracking time node.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images