A fast correction method for dynamic load bridge calibration coefficient based on monitoring data

Abstract

The present invention discloses a method for quickly correcting the dynamic load bridge calibration coefficient based on monitoring data, which belongs to the bridge structure health monitoring technology. The method first extracts dynamic weighing data and bridge measuring point deflection data based on the dynamic weighing system and deflection monitoring system deployed on the bridge, and analyzes the bridge single-vehicle crossing working condition information through the data; then, speed correction is performed based on the vehicle load information, and it is used as the input of the model to construct a neural network regression model with vehicle weight and speed as input and the bridge deflection change response of a single vehicle crossing the bridge as output; then, according to the regression model, the bridge deflection change response caused by different vehicle weights at low speed is predicted, and the dynamic load bridge calibration coefficient is calculated by using a function fitting method. The present invention constructs a set of calculation methods for bridge calibration coefficients through monitoring data, avoids the defects of closed roads and high costs for load tests, and realizes the rapid calculation of bridge calibration coefficients.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge structure health detection, and in particular relates to a method for quickly correcting a dynamic load bridge calibration coefficient based on monitoring data. Background Art

[0002] Bridges are the "throat" of transportation infrastructure. Their safety is closely related to the smooth operation of the transportation network. They are an important component of traffic safety and a vital part of social safety production. However, with the large number of bridges put into use, the age of existing bridges is increasing year by year, and traffic volume and load continue to grow. Bridge maintenance and management are facing increasing pressure, and the task of ensuring the health and safe operation of bridge structures has become increasingly urgent. The safety status of bridges is generally assessed through calibration coefficients. Traditional bridge calibration coefficients require load tests. However, although load tests are highly accurate, they are time-consuming and labor-intensive, and require long periods of traffic interruption.

[0003] There are currently gaps in the following research areas:

[0004] First, a large number of new bridge monitoring systems have been built, and the application of massive amounts of data is difficult, especially the integration of monitoring data with bridge assessments.

[0005] Second, there is currently a lack of a method that can avoid bridge load testing while obtaining the change in bridge deflection per unit load, which means that the calibration coefficient is difficult to calculate;

[0006] Third, bridges under dynamic loads will produce a dynamic impact effect, which will cause the deflection change at this time to be different from the deflection change of the bridge under static load. There is still a lack of a method to eliminate the dynamic load effect.

[0007] In response to the above problems, in order to form a method for obtaining test-free bridge calibration coefficients based on monitoring data, if traditional methods are used, there will be certain limitations and problems with large impacts such as closed traffic. Therefore, it is urgent to propose a fast correction method for dynamic load bridge calibration coefficients based on monitoring data, which can complete the calculation method of bridge calibration coefficients based on the elimination of dynamic load effects. Summary of the Invention

[0008] Purpose of the invention: In response to the defects of the prior art, the present invention aims to provide a method for quickly correcting the dynamic load bridge calibration coefficient based on monitoring data. The method can mine the single-vehicle bridge crossing condition information based on the dynamic weighing system and the deflection monitoring system and correct the vehicle speed through the deflection data. The regression model constructed based on a large amount of vehicle weight, vehicle speed, and bridge deflection change data can predict the bridge deflection change under low vehicle speed and different vehicle weights, and the unit load deflection change of the bridge during the period of time can be obtained by fitting. Finally, the bridge calibration coefficient is calculated based on the bridge test data.

[0009] Technical solution: A method for quickly correcting the dynamic load bridge calibration coefficient based on monitoring data. The bridge is equipped with a deflection monitoring system at the mid-span position and a dynamic weighing system is arranged at a distance of L0 meters from the bridge head to obtain the weight of passing vehicles. The method is characterized by comprising the following steps:

[0010] (1) Based on the dynamic weighing system and the deflection monitoring system, dynamic weighing data and bridge measuring point deflection data are extracted. Based on this data, the bridge single-vehicle crossing condition information is analyzed under the assumption that the vehicle is traveling at a constant speed. The condition information includes the vehicle's travel time and speed on the bridge;

[0011] (2) Speed ​​correction is performed based on vehicle load information. Specifically, the deflection data D of the i-th vehicle crossing the bridge obtained from the analysis in step (1) is used. i , and judge the abnormality of its data. If there is deflection data corresponding to the bicycle crossing the bridge, then trace the deflection starting point when the vehicle just enters the bridge and the deflection ending point when the vehicle crosses the bridge. Then, reversely infer the bicycle's travel time on the bridge through the deflection data points, calculate the corrected average speed, and obtain the filtered bicycle crossing the bridge deflection data D deflection , the change in bridge deflection caused by a single vehicle passing the bridge , Corrected vehicle speed V revise , vehicle weight W and vehicle lane Lane;

[0012] (3) Construct a neural network regression model of vehicle weight-vehicle speed-deflection change, combined with Adam algorithm optimization, the model is based on vehicle weight W and corrected vehicle speed V revise As the input of the model, the change of bridge deflection As output;

[0013] (4) Based on the vehicle weight-vehicle speed-deflection change neural network regression model constructed in step (3), the bridge unit load deflection change is obtained by linearly fitting the output of the model, and then the bridge calibration coefficient under dynamic load is calculated based on the bridge unit load deflection change.

[0014] In the above scheme, the bridge single vehicle crossing condition information analysis is as follows:

[0015] In T start ~T end During a time period, bridge deflection data during the time period is obtained based on the deflection monitoring system;

[0016] Combined with the speed measurement at the weighing station to obtain the speed V of the i-th vehicle passing i and weight W i , and based on the assumption that the lane and vehicle speed remain unchanged when the vehicle passes through the bridge, the time t when the i-th vehicle reaches the bridge head is calculated i，0=T i + L0 / V i， T i is the time when vehicle i passes the dynamic weighing system, and the time when it passes the bridge is t i，2 =T i + (L 0+ L) / V i ;

[0017] Recording the operating condition information of all vehicles passing through, and then integrating the operating condition information to form a data column of all single-vehicle bridge crossing conditions, the data column includes the lane data column Lane of N vehicles passing through, the speed data column V, the weight data column W, the time data column T0 of all vehicles passing through the bridge, the time data column T1 of the time of passing through the bridge, and the time data column T2 of the time of passing through the bridge end;

[0018] According to the travel time T of the i-th vehicle i，0 and T i，2 Calculate the time T when the vehicle reaches the mid-span position of the bridge i，1 , T i，1 The deflection data of the bridge mid-span measuring point at the corresponding time is listed in D i to match.

[0019] Furthermore, step (2) includes processing the data obtained by the deflection monitoring system as follows:

[0020] (21) Deflection data abnormality judgment, specifically:

[0021] Use the Hampel filter method, set the threshold to 3 times the standard deviation, and obtain the filtered data clean_deflection and the position index outlier_idx exceeding 3 times the standard deviation;

[0022] Calculate the mean of the filtered data clean_deflection as the baseline to clean the position index outlier_idx. If the position index outlier_idx is not a null value, record the index position outlier_idx. Otherwise, determine that the deflection data D is abnormal and discard it.

[0023] (22) Based on the bridge deflection data, the speed of a single vehicle crossing the bridge is corrected, specifically:

[0024] The deflection data D is smoothed using the smooth method to obtain D smooth Then calculate the standard deviation of the data that does not exceed three times the threshold as the data noise standard deviation, and take the absolute value of the difference between the smoothed data that exceeds 5 times the noise standard deviation and the baseline as the super-threshold data point, and record D over_thresholdThe indexes of the first and last data in the bridge are start_indx and end_indx; then the data is extended forward and backward to the point where the deviation condition is not satisfied for the first time, and the number of data between the indexes is calculated; and the average speed of vehicles passing on the bridge is:

[0025] V revise = L / T average

[0026] Where, T average T is the corrected time for a bicycle to cross the bridge. average =Num / f, where f is the sampling frequency of deflection data;

[0027] The deflection data column corresponding to a bicycle crossing the bridge is represented as D deflection =D, the change in bridge deflection caused by a single vehicle crossing the bridge =abs(baseline-min(D deflection )), abs is the absolute value function.

[0028] Furthermore, the neural network regression model described in step (3) includes a linear layer and an activation function layer. In order to make the model training converge quickly, the training data is standardized:

[0029]

[0030] in, Represents each original data in the dataset, is the mean of the data set, is the standard deviation of the data set, For each data after standardization.

[0031] Furthermore, the neural network regression model is optimized using the Adam algorithm, and the model training loss function uses the mean square error loss function. For the input sample x and the model prediction output , the true label is y, and the calculation formula of the mean square error loss function is .

[0032] Furthermore, the neural network regression model is based on the weight W of the vehicle in the same lane, the corrected speed V of the vehicle revise， As input, the model predicts the output of the bridge deflection change Then, all elements of the neural network regression model output are linearly fitted according to y=kx+b to calculate the optimal k value, which is the change in deflection per unit load of the bridge;

[0033] Based on the change of bridge unit load deflection , calculate the bridge calibration coefficient under dynamic load , .

[0034] On the other hand, the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the above-mentioned method for rapid correction of dynamic load bridge verification coefficients based on monitoring data.

[0035] A computer storage medium stores a computer program, which, when executed by a computer, implements the above-mentioned method for quickly correcting the dynamic load bridge calibration coefficient based on monitoring data.

[0036] Compared with the prior art, the substantial features and significant effects of the present invention include: the present invention is based on the existing dynamic weighing data of the bridge and the data of the bridge deflection monitoring system, combined with the neural network deep learning technology, which can achieve rapid acquisition of the bridge calibration coefficient and avoid the defects of closed traffic and high costs caused by bridge load tests. Furthermore, the method of the present invention corrects the average speed of vehicles crossing the bridge based on the deflection data, ensuring data accuracy. Finally, the present invention is simpler in terms of algorithm implementation and automated processing, and does not require huge computing power to realize the extraction of bridge single-vehicle crossing working condition information, realize the elimination of the dynamic effect of dynamic load bridge deflection, and realize the dynamic calculation of the bridge calibration coefficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flow chart of the method of the present invention;

[0038] Figure 2 This is a diagram of the neural network regression model architecture in the present invention. DETAILED DESCRIPTION

[0039] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below with reference to the accompanying drawings.

[0040] First, to address the structural health monitoring of bridges, existing technologies include deploying two monitoring systems: a dynamic weighing system and a deflection monitoring system. The combination of these two systems can analyze the impact of vehicle weight on bridge deflection. The following explanations are needed for these two systems from the perspectives of practical application and bridge health:

[0041] 1. Considering the impact of underground weighing equipment on bridge structures, dynamic weighing systems cannot be deployed on bridges in practice. Instead, they should be placed at a certain distance from the bridge to enable necessary control measures to be taken for overweight vehicles.

[0042] 2. The deflectometers in the deflection monitoring system should be deployed at the mid-span of the bridge, so as to achieve the most effective data collection and monitoring in the lowest cost optimized deployment plan.

[0043] In response to the above two practical application requirements, the following application scenarios are given:

[0044] Assume that the selected bridge is L meters long, has separate spans, and has a dynamic weighing system at the upper end of the bridge, located L0 meters from the bridgehead. The bridge also has three lanes per span. Because the bridge's midspan deflection is greatest, deflection monitoring systems are typically deployed at midspan. Assuming the deflection measurement point is located at the midspan section of the selected bridge, the deflection system's frequency of acquisition is f (Hz).

[0045] The present invention combines a dynamic weighing system and a deflection monitoring system to realize a fast correction method for a dynamic load bridge calibration coefficient based on monitoring data.

[0046] like Figure 1 As shown, a method for quickly correcting the dynamic load bridge calibration coefficient based on monitoring data includes the following steps:

[0047] (1) Extracting single-vehicle load information based on the dynamic weighing system and deflection monitoring system, including:

[0048] (11) Extraction of dynamic weighing data and bridge measurement point deflection data, as follows:

[0049] Assume that the dynamic weighing system is start ~T end There are N vehicles passing through in a certain time. The speed of the nth vehicle passing the dynamic weighing is V n , vehicle weight is W n The time when the dynamic weighing system passes is T n , the number of lanes for vehicles is one of 1, 2, and 3, where the value range of n is [1, 2, 3, ..., N]. start ~T end The number of deflection data collected during this time is (T end -T start ) f.

[0050] (12) Data-driven extraction of single-vehicle bridge crossing condition information, as follows:

[0051] By assuming that the lanes and speeds of vehicles passing through the bridge remain unchanged, we can estimate the time when the nth vehicle arrives at the bridge head as t n，0 =T n + L0 / V n, we know that the time when the nth vehicle reaches the middle of the bridge is t n，1 =T n +( L0+L / 2) / V n The time when the nth vehicle passes the end of the bridge is t n，2 =T n +(L0+L) / V n Therefore, the start time, mid-span time, and end time of N vehicles crossing the bridge can be calculated.

[0052] In order to obtain the weight, speed, lane and travel time of a single vehicle passing through the bridge, since it is assumed that the speed on the bridge remains unchanged, the time when the vehicle reaches the bridge head is pushed forward 30 seconds, and the time when the vehicle reaches the bridge end is pushed back 30 seconds (the specific forward and backward time can be considered as appropriate according to the actual situation). Based on the dynamic weighing system and deflection monitoring system, the following are obtained: the number of all passing vehicles N, the lane data column Lane of all passing vehicles, the speed data column V of all vehicles passing, the weight data column W of all vehicles, the time data column T0 of all vehicles passing through the bridge head, the time data column T1 of all vehicles passing through the bridge, and the time data column T2 of all vehicles passing through the bridge end. Then, the data of T1, T2, V, W, Lane are arranged in order according to T0 to obtain the newly sorted time data column T1, T2, V, W, Lane. If the time when the current vehicle passes the bridge end is less than the time when the next vehicle just arrives at the bridge head, the travel information T of the single vehicle is recorded. i，0 、T i，1 、T i，2 、V、W、Lane,T i，0、 T i，1、 T i，2 They represent the time when the i-th vehicle arrives at the bridge, the mid-span position, and leaves the bridge. Thus, T start ~T end The information of all vehicles passing through in the time. According to the obtained travel time T of the i-th vehicle i，0 and T i，2 Match the deflection data column D of the bridge's mid-span measurement point within the corresponding time i (The amount of data is (T i，2 - T i，0 ) f).

[0053] (2) Speed ​​correction based on the deflection data of a single vehicle crossing the bridge, specifically including:

[0054] In the above step (1), it is assumed that the speed remains constant. However, according to the actual situation, the speed of the vehicle will not remain constant. Therefore, the average speed of the vehicle crossing the bridge needs to be corrected.

[0055] (21) Deflection data abnormality judgment

[0056] First, according to the deflection data D of the i-th bicycle crossing the bridge obtained in step (1), i , it is necessary to judge the abnormality of the data to determine whether there is a downward deflection trend when a single-load vehicle passes the bridge. If not, the current deflection data D is discarded. i The specific method to determine whether the current data is abnormal is:

[0057] Based on the data acquired by the deflection monitoring system, the Hampel filtering method is used, with a threshold set at 3 times the standard deviation. This method generates the filtered data clean_deflection and the position index outlier_idx exceeding 3 times the standard deviation. The mean of the clean_deflection data is then used as the baseline for data cleaning. If the position index outlier_idx is not null, the deflection data D is retained and the index position outlier_idx is recorded. Otherwise, it is discarded.

[0058] (22) Correcting the speed of a single vehicle crossing the bridge based on bridge deflection data

[0059] Considering that in the above steps, the time when the i-th vehicle arrives at the bridge, the mid-span position and leaves the bridge is based on an assumption, and this time does not correspond to the actual situation (the assumption is based on the dynamic weighing system at the distance from the bridge), so the present invention corrects the time based on the changes in the bridge deflection data collected by the deflection monitoring system.

[0060] First, the deflection starting point when the vehicle just enters the bridge and the deflection ending point when the vehicle crosses the bridge are tracked. Then, the time is reversed through the data points to calculate the corrected average speed. For this process, the deflection data must be processed as follows:

[0061] 1) Use the smooth method to smooth the deflection data D to obtain D smooth ;

[0062] 2) Calculate the standard deviation of the data that does not exceed three times the threshold as the data noise standard deviation, and take the absolute value of the difference between the smoothed data that exceeds 5 times the noise standard deviation and the baseline as the over-threshold data point, and record D over_threshold The indexes of the first and last data in start_indx and end_indx;

[0063] 3) Expand forward and backward to the point where the deviation condition is not satisfied for the first time, and then calculate the number of data Num between the indexes;

[0064] 4) Single vehicle bridge crossing correction time T average =Num / f, where f is the sampling frequency of deflection data, and the average velocity correction V revise = L / T averageAfter data processing based on the deflection monitoring system, the corresponding deflection data of a single vehicle crossing the bridge is listed as D deflection =D(start_idx: end_idx) , the change in bridge deflection caused by a single vehicle crossing the bridge =abs(baseline-min(D deflection )), abs represents the absolute value function;

[0065] According to steps 1)-4), the filtered deflection data D of a single vehicle crossing the bridge can be obtained. deflection 、 , vehicle correction speed V revise , vehicle weight W and vehicle lane Lane.

[0066] (3) Construct a regression model of vehicle weight, vehicle speed and deflection change

[0067] The construction of a regression model with vehicle weight and speed as input and the change in bridge measuring point deflection as output requires data preprocessing to obtain the change in bridge measuring point deflection caused by a single vehicle passing the bridge.

[0068] (31) The data preprocessing process includes:

[0069] The weight and corrected speed of the vehicle in the same lane are used as input, namely W and V revise , taking the change of bridge deflection as Output. W, V revise 、 It has been obtained according to step (2). In order to make the model training converge quickly, the input W and V of the training data need to be revise and output For standardization, the standardization formula is as follows:

[0070]

[0071] in, Represents each original data in the dataset, is the mean of the data set, is the standard deviation of the data set, For each data after standardization.

[0072] (32) Neural network regression model construction

[0073] Because the model is simple, there is no need to use an overly complex deep neural network architecture. In order to ensure that the model training can converge quickly, it is composed of linear layers and activation function layers, combined with Figure 2 shown.

[0074] The model training uses the Adam optimization algorithm, and the model training loss function uses the mean square error loss function. For the input sample x and the model prediction output , the true label is y, and the calculation formula of the mean square error loss function is ;

[0075] Set a certain number of training batches and save the trained model.

[0076] (4) Correction of bridge calibration coefficients under dynamic loads based on big data, including predicting the change in bridge deflection for vehicles with equal speeds and different vehicle weights, calculating the slope through data fitting, and calculating the static calibration coefficient of the bridge under dynamic loads.

[0077] Based on the neural network regression model constructed in step (32), the speed of 10 km / h and different vehicle weights are used as model input to obtain the regression prediction output.

[0078] Specifically, the input data is Input= = , where the first column is the speed (unit: Km / h), the second column is the vehicle weight (unit: tons), and each row of data represents an input. The output is Output = Model (Input) = According to the linear fitting, y=kx+b is used to fit all the elements of the model output, and the optimal k value is calculated, which is the change in the deflection of the bridge per unit load.

[0079] Check the bridge completion report and bridge data to find out the change in bridge deflection per unit load , calculate the bridge verification coefficient under dynamic load as At this point, the bridge calibration coefficient is calculated based on the dynamic load.

Claims

1. A method for quickly correcting the dynamic load bridge calibration coefficient based on monitoring data, wherein the bridge is equipped with a deflection monitoring system at the mid-span position and a dynamic weighing system is arranged at a distance of L0 meters from the bridge head to obtain the weight of passing vehicles, characterized in that: The steps include: (1) Based on the dynamic weighing system and the deflection monitoring system, dynamic weighing data and bridge measuring point deflection data are extracted. Based on this data, the bridge single-vehicle crossing condition information is analyzed under the assumption that the vehicle is traveling at a constant speed. The condition information includes the vehicle's travel time and speed on the bridge; (2) Speed ​​correction is performed based on vehicle load information. Specifically, the deflection data D of the i-th vehicle crossing the bridge obtained from the analysis in step (1) is used. i , and judge the abnormality of its data. If there is deflection data corresponding to the bicycle crossing the bridge, then trace the deflection starting point when the vehicle just enters the bridge and the deflection ending point when the vehicle crosses the bridge. Then, reversely infer the bicycle's travel time on the bridge through the deflection data points, calculate the corrected average speed, and obtain the filtered bicycle crossing the bridge deflection data D deflection , the change in bridge deflection caused by a single vehicle passing the bridge , Corrected vehicle speed V revise , vehicle weight W and vehicle lane Lane; (3) Construct a neural network regression model of vehicle weight-vehicle speed-deflection change, combined with Adam algorithm optimization, the model is based on vehicle weight W and corrected vehicle speed V revise As the input of the model, the change of bridge deflection As output; (4) Based on the vehicle weight-vehicle speed-deflection change neural network regression model constructed in step (3), the bridge unit load deflection change is obtained by linearly fitting the output of the model, and then the bridge calibration coefficient under dynamic load is calculated based on the bridge unit load deflection change.

2. The method for quickly correcting the dynamic load bridge calibration coefficient based on monitoring data according to claim 1 is characterized in that: The analysis of the bridge single vehicle crossing condition information is as follows: In T start ~T end During a time period, bridge deflection data during the time period is obtained based on the deflection monitoring system; Combined with the speed measurement at the weighing station to obtain the speed V of the i-th vehicle passing i and weight W i , and based on the assumption that the lane and vehicle speed remain unchanged when the vehicle passes through the bridge, the time t when the i-th vehicle reaches the bridge head is calculated i，0 =T i + L0 / V i， T i is the time when vehicle i passes the dynamic weighing system, and the time when it passes the bridge is t i，2 =T i + (L 0+ L) / V i ; Recording the operating condition information of all vehicles passing through, and then integrating the operating condition information to form a data column of all single-vehicle bridge crossing conditions, the data column includes the lane data column Lane of N vehicles passing through, the speed data column V, the weight data column W, the time data column T0 of all vehicles passing through the bridge, the time data column T1 of the time of passing through the bridge, and the time data column T2 of the time of passing through the bridge end; According to the travel time T of the i-th vehicle i，0 and T i，2 Calculate the time T when the vehicle reaches the mid-span position of the bridge i，1 , T i，1 The deflection data of the bridge mid-span measuring point at the corresponding time is listed in D i to match.

3. The method for rapid correction of dynamic load bridge calibration coefficient based on monitoring data according to claim 1 is characterized in that: Step (2) includes processing the data obtained by the deflection monitoring system as follows: (21) Deflection data abnormality judgment, specifically: Use the Hampel filter method, set the threshold to 3 times the standard deviation, and obtain the filtered data clean_deflection and the position index outlier_idx exceeding 3 times the standard deviation; Calculate the mean of the filtered data clean_deflection as the baseline to clean the position index outlier_idx. If the position index outlier_idx is not a null value, record the index position outlier_idx. Otherwise, determine that the deflection data D is abnormal and discard it. (22) Based on the bridge deflection data, the speed of a single vehicle crossing the bridge is corrected, specifically: The deflection data D is smoothed using the smooth method to obtain D smooth Then calculate the standard deviation of the data that does not exceed three times the threshold as the data noise standard deviation, and take the absolute value of the difference between the smoothed data that exceeds 5 times the noise standard deviation and the baseline as the super-threshold data point, and record D over_threshold The indexes of the first and last data in the bridge are start_indx and end_indx; then the data is extended forward and backward to the point where the deviation condition is not satisfied for the first time, and the number of data between the indexes is calculated; and the average speed of vehicles passing on the bridge is: V revise = L / T average Where, T average T is the corrected time for a bicycle to cross the bridge. average =Num / f, where f is the sampling frequency of deflection data; The deflection data column corresponding to a bicycle crossing the bridge is represented as D deflection =D, the change in bridge deflection caused by a single vehicle crossing the bridge =abs(baseline-min(D deflection )), abs represents the absolute value function.

4. The method for rapid correction of dynamic load bridge calibration coefficient based on monitoring data according to claim 1 is characterized in that: The neural network regression model described in step (3) includes a linear layer and an activation function layer. In order to make the model training converge quickly, the training data is standardized: ， in, Represents each original data in the dataset, is the mean of the data set, is the standard deviation of the data set, For each data after standardization.

5. The method for rapid correction of dynamic load bridge calibration coefficient based on monitoring data according to claim 1 or 4, characterized in that: The neural network regression model is optimized using the Adam algorithm, and the model training loss function uses the mean square error loss function. For the input sample x and the model prediction output , the true label is y, and the calculation formula of the mean square error loss function is .

6. The method for quickly correcting the dynamic load bridge calibration coefficient based on monitoring data according to claim 1 is characterized in that: The neural network regression model is based on the vehicle weight W and the corrected vehicle speed V in the same lane. revise， As input, the model predicts the output of the bridge deflection change Then, all elements of the neural network regression model output are linearly fitted according to y=kx+b to calculate the optimal k value, which is the change in deflection per unit load of the bridge; Based on the change of bridge unit load deflection , calculate the bridge calibration coefficient under dynamic load , .

7. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the method for quickly correcting the dynamic load bridge verification coefficient based on monitoring data as described in any one of claims 1 to 6.

8. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the method for quickly correcting the dynamic load bridge calibration coefficient based on monitoring data as described in any one of claims 1 to 6 is implemented.

Images