Road fatigue damage prediction method, system, device, medium and program product

Abstract

The application discloses a road fatigue damage prediction method, system, device, medium and program product, and relates to the technical field of pavement damage prediction. The method comprises the following steps: according to obtained pavement traffic load data, converting each axle load of each vehicle into the equivalent axle times of standard axle load, summarizing the equivalent axle times of all vehicles in the current moment to obtain the multi-axle load cumulative value of the current moment, and obtaining the cumulative damage degree according to the multi-axle load cumulative value; predicting the evolution trend of the cumulative damage degree according to the multi-axle load cumulative value of the current moment, the proportion of each type of vehicle, the average vehicle speed of the current moment and the cumulative damage degree of the last moment, obtaining the damage degree prediction value of a plurality of future moments set, and pushing an early warning signal when the damage degree prediction value exceeds the set threshold. The high-frequency and multi-axle load data are accumulated according to time periods, a more real and quantitative road damage degree is obtained, and the potential damage risk of the pavement is reduced through the rolling prediction of the damage degree.

Description

Technical Field

[0001] This invention relates to the field of pavement damage prediction technology, and in particular to a method, system, device, medium, and program product for predicting road fatigue damage. Background Technology

[0002] Traditional road design and maintenance assessments typically use single axle loads or vehicle gross weight to estimate the degree of damage to the pavement. However, in today's environment of high traffic volume and heavy freight, a large number of heavy-duty trucks and multi-axle vehicles frequently travel on main roads, expressways, and urban expressways. The fatigue damage caused by these vehicles to the pavement far exceeds the scope that single axle loads or vehicle gross weights can reflect.

[0003] Existing methods calculate road fatigue damage based on the vehicle's total weight or a representative single-axle weight (such as the load passing through a fixed axle). However, in multi-axle vehicles, the impact frequency and amplitude of each axle on the road surface differ, and variations in wheelbase and the number of axles affect the way wheel loads act on the road surface and their cumulative effects. Furthermore, heavy-duty vehicles with a high number of axles often exert multiple, cumulative fatigue effects on the road surface within a short period. This leads to a significant lack of information regarding the specific details of multi-axle load characteristics (such as 5-axle, 6-axle, or even 8-axle loads) when calculating cumulative fatigue, if only a single axle or total weight is used as a benchmark. This results in errors in the assessment results and may even significantly underestimate the risk of road damage. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes a road fatigue damage prediction method, system, device, medium, and program product. This method accumulates high-frequency, multi-axle load data over time periods to obtain a more accurate quantification of road damage. By predicting the damage level over time, it reduces the risk of potential road surface damage.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] In a first aspect, the present invention provides a method for predicting road fatigue damage, comprising:

[0007] Based on the acquired road traffic load data, each axle load of each vehicle is converted into the equivalent axle load of the standard axle load. The equivalent axle loads of all vehicles at the current moment are summarized to obtain the cumulative value of the multi-axle load at the current moment. The cumulative damage degree is obtained based on the cumulative value of the multi-axle load.

[0008] Based on the current cumulative multi-axle load, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage level of the previous moment, the evolution trend of the cumulative damage level is predicted to obtain the damage level prediction values ​​for multiple future moments. A warning signal is pushed out when the damage level prediction value exceeds the set threshold.

[0009] As an alternative implementation, the process of converting to equivalent axle loads for standard axle loads includes: In the formula, For the equivalent number of shaft loads on a single shaft, To convert the index, For shaft group coefficients, This is the wheelset coefficient. For the single-axis axle load of the i-th axle type, This is the standard axle load.

[0010] As an alternative implementation method, the process of obtaining the cumulative damage degree based on the cumulative value of multi-axis load includes: In the formula, for Accumulated multi-axis load value at any given time; for Cumulative damage over time for Cumulative damage over time It is a constant that adjusts the relationship between road materials and load. It is the durable axle load capacity of the road under standard load.

[0011] As an alternative implementation method, the process of predicting the evolution trend of cumulative damage includes: based on the current cumulative multi-axle load value, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage value of the previous time, a long short-term memory network is used to map the hidden state, and then the predicted damage value for the next time moment is obtained. ;in,

[0012] ;

[0013] ;

[0014] ;

[0015] in, This represents the hidden state at time t-1; The cell state at time t-1; Let be the hidden state at time t; Let be the cell state at time t; Output for the forget gate; Use the Sigmoid activation function; , , and These are the weight matrices for the forget gate, input gate, cell update, and output gate, respectively. , , and These are the bias vectors for the forget gate, input gate, cell update, and output gate, respectively. This represents the current state of the candidate cells. For input gate output; Output gate output; The hyperbolic tangent activation function is used. This is the weight matrix of the output layer; This is the bias vector for the output layer.

[0016] As an alternative implementation, the network parameters of the Long Short-Term Memory network are updated based on the gradient descent algorithm using the mean squared error loss function, according to the difference between the predicted damage value at the next time step and the actual cumulative damage value.

[0017] As an alternative implementation, a damage prediction curve is constructed based on the damage prediction values ​​at multiple future times. In the damage prediction curve, if there exists a certain future time... If the set threshold is exceeded, an early warning signal will be sent to alert the system. Damage prediction value at time Road maintenance measures should be carried out before the problem occurs.

[0018] In a second aspect, the present invention provides a road fatigue damage prediction system, comprising:

[0019] The damage calculation module is configured to convert each axle load of each vehicle into the equivalent axle load of the standard axle load based on the acquired road traffic load data, summarize the equivalent axle loads of all vehicles at the current time, obtain the cumulative value of the multi-axle load at the current time, and obtain the cumulative damage degree based on the cumulative value of the multi-axle load.

[0020] The damage prediction module is configured to predict the evolution trend of cumulative damage based on the current cumulative multi-axle load, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage at the previous moment, to obtain the predicted damage values ​​for multiple future moments, and to push a warning signal when the predicted damage value exceeds a set threshold.

[0021] Thirdly, the present invention provides an electronic device including a memory and a processor, and computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the method described in the first aspect.

[0022] Fourthly, the present invention provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, perform the method described in the first aspect.

[0023] Fifthly, the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the method described in the first aspect.

[0024] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0025] This invention proposes a method, system, device, medium, and program product for predicting road fatigue damage. First, each axle load of each vehicle is converted into equivalent axle loads of standard axle loads, and the equivalent axle loads of all vehicles are summarized to obtain the cumulative multi-axle load value at the current moment. Based on the cumulative multi-axle load value, the cumulative damage degree is obtained. High-frequency, multi-axle load data is accumulated by time period to obtain a more realistic quantification of road damage degree. Then, based on the key traffic load characteristics at the current moment, the evolution trend of the cumulative damage degree is predicted to obtain predicted damage degrees for several future moments. An early warning signal is pushed when the predicted damage degree exceeds a set threshold. This achieves rolling prediction of damage degree, improves prediction accuracy and robustness, and reduces the risk of potential road surface damage.

[0026] Traditional methods rely solely on vehicle gross weight or a few axle load indicators to estimate road fatigue, easily overlooking the cumulative effect caused by multi-axle coupling. To address the problem of accurately mapping vehicle multi-axle load data to cumulative road fatigue, this invention introduces an equivalent axle load conversion formula and the Miner fatigue criterion. This allows for the accumulation of high-frequency, multi-axle load data over time periods, resulting in a more realistic quantification of road damage. Unlike traditional methods that primarily rely on estimates or recommended parameters, this invention uses measured data for each vehicle and axle, dynamically accumulating and converting data by vehicle type, axle, and time period. This maximizes the preservation of accurate traffic load information and effectively reduces systematic errors caused by simplification or empirical estimation. Therefore, the results calculated by this invention more closely reflect actual traffic load patterns, significantly improving assessment accuracy.

[0027] Traditional methods lack a refined and accurate prediction process and cannot systematically present the cumulative fatigue process of road surfaces from the perspective of single-vehicle multi-axle data. When faced with a surge in heavy-duty vehicles, the lag in traditional methods can easily lead to excessive road damage. Therefore, this invention uses in-depth mining of multi-axle load data, unlike traditional simplified assessments based on total weight or single axle, to meticulously depict the complex fatigue damage to the road surface caused by multi-axle vehicles. Then, it introduces time-series prediction technology based on long short-term memory networks, making road damage prediction more sensitive and accurate by taking into account the time-varying nature of traffic flow fluctuations and high-load vehicle characteristics. Finally, based on the damage prediction curve, timely repair is intervened before fatigue reaches its limit. Combined with incremental training and new data feedback, a new self-learning road maintenance management system is formed. In practical applications, this helps to carry out targeted maintenance in key sections or lanes, significantly extending the local or overall service life of the road surface and saving on later maintenance costs.

[0028] To address the significant fluctuations in traffic flow across daily, weekly, and seasonal cycles, as well as the variations in vehicle category ratios, this invention aims to achieve time-series prediction of road fatigue damage and effectively forecast short- and medium-term road damage trends. It utilizes Long Short-Term Memory (LSTM) networks to capture the nonlinear evolution mechanism of damage, enabling rolling predictions. Furthermore, it embeds multiple key features into the model, such as cumulative multi-axle load values, vehicle type ratios, and vehicle speeds, thereby improving the accuracy and robustness of the predictions.

[0029] This invention addresses how to establish a closed-loop maintenance decision-making mechanism based on damage prediction results, including early warning of maintenance timing, scheme selection, and dynamic verification of implementation effects. By setting a threshold on the damage prediction curve, if the cumulative fatigue level exceeds the allowable limit in a future period, an early warning signal is issued. This allows for the early deployment of optimized maintenance measures before heavy traffic causes serious damage, minimizing potential damage and economic losses.

[0030] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0032] Figure 1 A schematic diagram of the implementation environment of the solution provided in Embodiment 1 of the present invention;

[0033] Figure 2 This is a flowchart of the road fatigue damage prediction method provided in Embodiment 1 of the present invention;

[0034] Figure 3 This is a framework diagram of the road fatigue damage prediction system provided in Embodiment 2 of the present invention. Detailed Implementation

[0035] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0036] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0037] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form as well. Furthermore, it should be understood that the terms “comprising” and “including”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0038] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0039] Multi-axle load data can be refined to the actual weight borne by each axle of a vehicle. By combining information such as axle spacing and vehicle type, it provides a more accurate basis for identifying the contribution of different vehicles to road fatigue damage. Transforming and accumulating all axle loads according to a certain nonlinear relationship can comprehensively capture the real impact process of multi-axle vehicles. Therefore, when assessing road life or developing maintenance strategies, it can more closely reflect actual traffic conditions and reduce underestimations or overestimations caused by simplified models.

[0040] Therefore, this invention proposes a road fatigue damage prediction method. By combining multi-axle load data with a pavement cumulative damage model, and then using a Long Short-Term Memory (LSTM) network in deep learning for time series prediction, a dynamic management and early warning of road cumulative damage is formed, providing a more scientific and accurate basis for road maintenance decisions.

[0041] Example 1

[0042] Figure 1 This is an implementation environment diagram related to a road fatigue damage prediction method provided in an embodiment of the present invention. For example... Figure 1 As shown, the implementation environment includes a vehicle 101, an image acquisition device 102, a weighing system 103, and a server 104; wherein the vehicle 101 is equipped with an on-board GPS device.

[0043] The image acquisition device 102 can be deployed along the road. For example, it can be deployed at a highway checkpoint. Taking one checkpoint as an example, the image acquisition device 102 can capture photos of all vehicles passing through the checkpoint at any given time and transmit the captured photos of all vehicles passing through the checkpoint at any given time to the server 104. The server 104 can identify the license plate data and vehicle model data of all vehicles in the photos based on the received photos of all vehicles passing through the checkpoint at any given time.

[0044] The weighing system 103 is mainly deployed at the road entrance to weigh the vehicles entering the road and collect license plate data. The license plate data and weight of the vehicles are then stored.

[0045] Server 104 is used to obtain the vehicle's GPS data sent by the vehicle's onboard GPS device. The vehicle's GPS data includes the vehicle's license plate data, the vehicle's position coordinates on the road at each moment, and the time corresponding to when the vehicle passes the position coordinates.

[0046] Server 104, based on the acquired road traffic load data, such as vehicle type, axle load, total weight, and passage time, converts each axle load of each vehicle into the equivalent axle number of standard axle loads. It then summarizes the equivalent axle numbers of all vehicles at the current moment to obtain the cumulative multi-axle load value for the current moment, and calculates the cumulative damage degree based on the cumulative multi-axle load value. Furthermore, based on the cumulative multi-axle load value at the current moment, the proportion of various vehicle types, the average vehicle speed at the current moment, and the cumulative damage degree of the previous moment, it predicts the evolution trend of the cumulative damage degree to obtain the predicted damage degree values ​​for multiple future moments. When the predicted damage degree value exceeds the set threshold, it pushes an early warning signal.

[0047] The server 104 can be a single server, a server cluster, or a cloud platform; this embodiment of the invention does not limit the specific type of server.

[0048] The road fatigue damage prediction method provided by the embodiments of the present invention will be described next.

[0049] Figure 2 This invention provides a method for predicting road fatigue damage. The method can be applied to a target server, wherein the target server can be... Figure 1 The server shown. (As shown) Figure 2 As shown, the method includes the following steps:

[0050] S201: Based on the acquired road traffic load data, convert each axle load of each vehicle into the equivalent axle load of the standard axle load, summarize the equivalent axle loads of all vehicles at the current moment, obtain the cumulative value of the multi-axle load at the current moment, and obtain the cumulative damage degree based on the cumulative value of the multi-axle load.

[0051] S202: Based on the current cumulative multi-axle load, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage level of the previous moment, predict the evolution trend of the cumulative damage level to obtain the predicted damage level values ​​for multiple future moments, and push a warning signal when the predicted damage level value exceeds the set threshold.

[0052] In this embodiment, road traffic load data is first collected by the on-site dynamic weighing system, including: the total number of vehicles in each lane in each direction, vehicle type, vehicle speed, number of axles, axle composition, axle load, axle spacing, total weight, and time of passage (date, hour, minute, second), etc. Among them, the axle type can be determined based on the wheel set and axle set, and the vehicle type can be determined based on the axle type.

[0053] Then, calculate the equivalent axle load for each type of vehicle and axle type, which means converting each axle load of each vehicle into the equivalent axle load of the standard axle load.

[0054] ;

[0055] In the formula, Equivalent shaft number for single-axis axle load; For conversion index, if it is for fatigue and permanent deformation of asphalt mixture layer, n=4; if it is for permanent deformation of subgrade, n=5; if it is for fatigue of inorganic binder stabilized layer, n=13. For the axle group coefficient, when the distance between the front and rear axles is greater than 3m, it is calculated as a single axle; when the distance between the axles is less than or equal to 3m, the value is taken according to Table 1. This is the wheel set coefficient; 1 for a dual-wheel set and 4.5 for a single-wheel set. For the i-th type of shaft, is the single-shaft axle load; for double-shaft and triple-shaft, is the axle load evenly distributed to each single shaft. This is the standard axle load.

[0056] Table 1. Values ​​of shaft group coefficients;

[0057] .

[0058] Furthermore, the current moment The equivalent axle load of all vehicles is summed to obtain the cumulative multi-axle load value at the current time t. Compared with traditional methods that measure by total vehicle weight or single axle load, the method in this embodiment can accurately reflect the coupling effect of multi-axle load on road fatigue.

[0059] To quantify the contribution of multi-axle load data to the cumulative pavement damage, this embodiment employs a pavement cumulative damage model based on the Miner criterion. Each time step (e.g., hour or day) is used as the calculation unit, based on the cumulative value of multi-axle load... Obtain cumulative damage :

[0060] ;

[0061] In the formula, for Cumulative damage over time This represents the road's durable axle load capacity under standard load conditions. It is a constant that adjusts the relationship between road materials and loads. In practical engineering, The value usually needs to be determined comprehensively based on the specific road structure type, material properties, regional specifications and historical test data, and is generally selected between 0.8 and 1.2. However, it can be selected according to the actual engineering design or test situation, and no mandatory limit is imposed here.

[0062] In this embodiment, the cumulative damage degree is recursively calculated at each time step to characterize the cumulative effect of multi-axle load data on road fatigue at different time periods. When the value is close to or equal to 1, it indicates that the road structure is approaching or has reached its fatigue limit, requiring immediate assessment and maintenance or overhaul. If it continues to operate at a high level, it is highly susceptible to large-scale cracking, rutting, or slab breakage in a short period of time.

[0063] In this embodiment, the cumulative damage degree is obtained. Then, using long short-term memory networks to... The evolution trend is dynamically predicted by using key traffic load characteristics, namely the current cumulative value of multi-axle load, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage degree of the previous time, as input to the LSTM network, so that the network can "see" the changing patterns of multi-axle load data during training and prediction.

[0064] Specifically:

[0065] Each time step, i.e. The input feature vector of the LSTM network at time 1 Defined as:

[0066] ;

[0067] in, The percentage of each type of vehicle; The average vehicle speed at the current moment is calculated based on the speed of each vehicle and the total number of vehicles at the current moment. This represents the cumulative damage level at the previous moment.

[0068] Among them, As model input, it incorporates historical state information into the current prediction, enabling dynamic recursion of time series data, which aligns with common practices in time series modeling such as LSTM networks. As model input, it reflects the direct impact of the current load on the damage. Combining the two ensures that the model utilizes both historical damage conditions and reflects current load changes, improving the continuity and accuracy of predictions. In addition, factors such as temperature and rainfall can also be incorporated.

[0069] The input feature vector is fed into an LSTM network, passing through forget gates, input gates, and output gates, mapping the historical sequence and the current input together to a new hidden state. and cell state And finally output the predicted damage value for the next time step. .

[0070] The core process is represented as follows:

[0071] ;

[0072] ;

[0073] ;

[0074] in, This represents the hidden state at time t-1; The cell state at time t-1; Output for the forget gate; Use the Sigmoid activation function; , , and These are the weight matrices for the forget gate, input gate, cell update, and output gate, respectively. , , and These are the bias vectors for the forget gate, input gate, cell update, and output gate, respectively. This represents the current state of the candidate cells. For input gate output; Output gate output; It is the hyperbolic tangent activation function.

[0075] Finally, the predicted value is obtained through a linear mapping. ;in, This is the weight matrix of the output layer; This is the bias vector for the output layer.

[0076] During the model training phase, Cumulative damage at the actual time t+1 Calculate the difference (e.g., using the mean squared error loss function) and continuously update it using the gradient descent algorithm. , , Network parameters, such as these, allow the predictions to gradually approach the actual values.

[0077] Therefore, LSTM networks can capture nonlinear changes in traffic volume while learning the role of multi-axle load data in the accumulation of road damage into their own parameters, making predictions more sensitive to periods of high traffic volume and heavy-load vehicles, thus outputting fatigue trends that are more in line with reality.

[0078] By employing a multi-step rolling forecasting method, several time periods (hours, days, or weeks) can be projected forward from any given moment, thus obtaining a damage severity prediction curve. If, in the damage severity prediction curve, at a certain future moment... If the threshold is exceeded, the relevant road management department will be notified. Beforehand, make proper maintenance and construction plans.

[0079] In this embodiment, after predicting the damage level at multiple future time points, an early warning and maintenance management process is further established. When the prediction results show... When the road surface temperature consistently exceeds a set maximum threshold (e.g., 0.7 or 0.8), it is considered to face a high risk of fatigue damage. At this point, an early warning message can be sent to the management system, recommending relevant departments to arrange inspections, overlays, localized pothole repairs, or other necessary preventative maintenance before that time period. If there is a high probability of severe overloading (e.g., frequent occurrences of overloaded trucks in specific areas), dynamic scheduling measures are embedded in the maintenance management system, such as setting speed limits, load limits, or lane diversion emergency strategies to mitigate severe damage. After taking relevant measures, data for new time periods can continue to be monitored in real time and fed back to the LSTM network for incremental training, achieving closed-loop management of road surface conditions.

[0080] Example 2

[0081] Please refer to Figure 3 This is a block diagram of a road fatigue damage prediction system according to an embodiment of the present invention. The system is used to perform the steps of the road fatigue damage prediction method described above. The road fatigue damage prediction system includes: a damage degree calculation module 301 and a damage degree prediction module 302.

[0082] The damage calculation module 301 is configured to convert each axle load of each vehicle into the equivalent axle load of the standard axle load based on the acquired road traffic load data, summarize the equivalent axle loads of all vehicles at the current time, obtain the cumulative value of the multi-axle load at the current time, and obtain the cumulative damage degree based on the cumulative value of the multi-axle load.

[0083] The damage prediction module 302 is configured to predict the evolution trend of cumulative damage based on the current cumulative multi-axle load, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage at the previous moment, to obtain the predicted damage values ​​for multiple future moments, and to push a warning signal when the predicted damage value exceeds a set threshold.

[0084] It should be noted that the system provided in the above embodiments is only illustrated by the division of the above functional modules when implementing its functions. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0085] In an exemplary embodiment, an electronic device is also provided, including a memory and a processor, and computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the method described in Embodiment 1.

[0086] A processor may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor may be implemented using at least one of the following hardware forms: Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor may also include a main processor and coprocessors. The main processor, also known as the Central Processing Unit (CPU), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor may integrate a Graphics Processing Unit (GPU), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor may also include an AI processor, which handles computational operations related to machine learning.

[0087] The memory may include one or more computer-readable storage media, which may be non-transitory. The memory may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, the non-transitory computer-readable storage media in the memory are used to store computer instructions configured to be executed by one or more processors to implement the methods described above.

[0088] In an exemplary embodiment, a computer-readable storage medium is also provided, wherein computer instructions are stored therein, which, when executed by a processor, perform the method described in Embodiment 1. Optionally, the computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), solid-state drives (SSDs), or optical discs, etc. The random access memory may include resistive random access memory (ReRAM) and dynamic random access memory (DRAM).

[0089] In an exemplary embodiment, a computer program product is also provided, the computer program product including a computer program stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, causing the computer device to perform the method described in Embodiment 1.

[0090] It should be noted that the collection and processing of relevant data in this invention, when applied in practice, should strictly comply with the requirements of relevant laws and regulations, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing within the scope of laws and regulations and the authorization of the personal information subject.

[0091] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for predicting road fatigue damage, characterized in that, include: Based on the acquired road traffic load data, each axle load of each vehicle is converted into the equivalent axle load of the standard axle load. The equivalent axle loads of all vehicles at the current moment are summarized to obtain the cumulative value of the multi-axle load at the current moment. The cumulative damage degree is obtained based on the cumulative value of the multi-axle load. Based on the current cumulative multi-axle load, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage level of the previous moment, the evolution trend of the cumulative damage level is predicted to obtain the damage level prediction values ​​for multiple future moments. A warning signal is pushed out when the damage level prediction value exceeds the set threshold. The process of predicting the evolution trend of cumulative damage includes: based on the current cumulative multi-axle load, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage from the previous time step, a Long Short-Term Memory (LSTM) network is used to map the hidden states, and then the predicted damage value for the next time step is obtained from this. ;in, ； ； ； in, This represents the hidden state at time t-1; The cell state at time t-1; Let be the hidden state at time t; Let be the cell state at time t; Output for the forget gate; Use the Sigmoid activation function; , , and These are the weight matrices for the forget gate, input gate, cell update, and output gate, respectively. , , and These are the bias vectors for the forget gate, input gate, cell update, and output gate, respectively. This represents the current state of the candidate cells. For input gate output; Output gate output; The hyperbolic tangent activation function is used. This is the weight matrix of the output layer; This is the bias vector for the output layer; A damage prediction curve is constructed based on the predicted damage values ​​at multiple future times. If a certain future time exists within the damage prediction curve... If the set threshold is exceeded, an early warning signal will be sent to alert the system. Damage prediction value at time Road maintenance measures should be carried out before the problem occurs.

2. The road fatigue damage prediction method as described in claim 1, characterized in that, The process of converting to the equivalent number of axles for standard axle loads includes: In the formula, For the equivalent number of shaft loads on a single shaft, To convert the index, For shaft group coefficients, This is the wheelset coefficient. For the single-axis axle load of the i-th axle type, This is the standard axle load.

3. The road fatigue damage prediction method as described in claim 1, characterized in that, The process of obtaining the cumulative damage degree based on the cumulative value of multi-axis load includes: In the formula, for Accumulated multi-axis load value at any given time; for Cumulative damage over time for Cumulative damage over time It is a constant that adjusts the relationship between road materials and load. It is the durable axle load capacity of the road under standard load.

4. The road fatigue damage prediction method as described in claim 1, characterized in that, Based on the difference between the predicted damage value at the next time step and the actual cumulative damage value, the network parameters of the Long Short-Term Memory network are updated using the mean squared error loss function and the gradient descent algorithm.

5. A road fatigue damage prediction system, characterized in that, include: The damage calculation module is configured to convert each axle load of each vehicle into the equivalent axle load of the standard axle load based on the acquired road traffic load data, summarize the equivalent axle loads of all vehicles at the current time, obtain the cumulative value of the multi-axle load at the current time, and obtain the cumulative damage degree based on the cumulative value of the multi-axle load. The damage prediction module is configured to predict the evolution trend of cumulative damage based on the current cumulative multi-axle load, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage at the previous moment, to obtain the damage prediction values ​​for multiple future moments, and to push a warning signal when the damage prediction value exceeds a set threshold. The process of predicting the evolution trend of cumulative damage includes: based on the current cumulative multi-axle load, the proportion of various vehicle types, the current average vehicle speed, and the cumulative damage from the previous time step, a Long Short-Term Memory (LSTM) network is used to map the hidden states, and then the predicted damage value for the next time step is obtained from this. ;in, ； ； ； in, This represents the hidden state at time t-1; The cell state at time t-1; Let be the hidden state at time t; Let be the cell state at time t; Output for the forget gate; Use the Sigmoid activation function; , , and These are the weight matrices for the forget gate, input gate, cell update, and output gate, respectively. , , and These are the bias vectors for the forget gate, input gate, cell update, and output gate, respectively. This represents the current state of the candidate cells. For input gate output; Output gate output; The hyperbolic tangent activation function is used. This is the weight matrix of the output layer; This is the bias vector for the output layer; A damage prediction curve is constructed based on the predicted damage values ​​at multiple future times. If a certain future time exists within the damage prediction curve... If the set threshold is exceeded, an early warning signal will be sent to alert the system. Damage prediction value at time Road maintenance measures should be carried out before the problem occurs.

6. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the method according to any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, perform the method described in any one of claims 1-4.

8. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, implements the method described in any one of claims 1-4.

Images