Self-adaptive online intercepting and storing method for actuating response of bridge high-frequency vehicle

Abstract

The invention, which relates to the technical field of bridge health monitoring, discloses a self-adaptive online interception and storage method for a bridge high-frequency vehicle actuation response, comprising the following steps: firstly, preliminarily segmenting total strain data into a pseudo temperature strain data segment and a pseudo vehicle-induced strain data segment by using a smooth iteration thought; secondly, correctly classifying the vehicle-induced strain data segments in the pseudo temperature strain data segments, and finally removing the temperature strain data segments from the pseudo vehicle-induced strain data segments through a range-based screening method according to short-time range characteristics of the temperature strain data segments in the pseudo vehicle-induced strain data segments; according to the method, the vehicle-induced strain data volume can be remarkably reduced, the data storage and transmission burden can be effectively relieved, the strong anti-noise capability is shown in the high-frequency noise environment, the high segmentation precision can be kept, the accuracy of vehicle-induced strain extraction is remarkably improved, the storage and real-time processing problems in large data volume monitoring are solved, and the method is suitable for popularization and application. And effective support is provided for sustainable operation of the bridge health monitoring system.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge health monitoring, and particularly relates to a data processing method capable of adaptively and real-time intercepting vehicle-induced strain data segments. Background Art

[0002] In recent years, lightweight monitoring of medium and small-span bridges has become a development trend. Compared with long-span bridges, due to their unique structural characteristics, the stress changes of medium and small-span bridges are more frequent. Therefore, higher monitoring frequencies of strain are required, resulting in a significant increase in the amount of monitoring data. Vehicle-induced strain can be used to identify local damages such as cracks, fatigue, and corrosion in bridges. After eliminating the influence of temperature, it can more accurately reflect the stress state of the bridge. In addition, vehicle-induced strain can be used to evaluate the impact of different traffic loads on the bridge, providing a basis for bridge design and traffic management. In high-frequency monitoring, vehicle-induced strain is the most valuable data for analysis, while the strain at other times has limited effects. Moreover, temperature-induced strain is often much greater than vehicle-induced strain, even ten times higher, which may mask the true influence of vehicle loads. Therefore, reducing data redundancy, improving the processing and storage efficiency of high-frequency monitoring data, and accurately extracting vehicle-induced strain are the keys to lightweight monitoring of medium and small-span bridges.

[0003] In order to effectively eliminate temperature strain, over the past few decades, researchers have proposed various methods, mainly divided into two categories. One category is based on the mathematical relationship model between temperature and strain. This method analyzes the correlation between temperature and strain, uses the measured temperature data to calculate the strain caused by temperature changes, and eliminates the influence of temperature from the actually measured strain, thereby realizing the separation of temperature strain and strain caused by vehicle loads. For example, some researchers predict the thermal response through a regression model combined with distributed temperature data; some researchers efficiently separate the temperature effect in the response of cable-stayed bridges through discretization of thermal loads and multiple linear modeling, solving the problems of complex calculation and dependence on long-term data in traditional methods; some researchers have proposed to consider the autoregressive component on the basis of considering the trend component, seasonal component, and regression component in the traditional Bayesian dynamic linear model, and improve the Bayesian dynamic linear model. The improved Bayesian dynamic linear model can capture the relationship between temperature and temperature-induced strain, and the Bayesian dynamic linear model can adapt to both stationary and non-stationary time series data and describe the time-varying structural strain response by calling different hidden components. However, although statistical models are convenient in separating temperature effects, due to the need to collect a large amount of temperature and strain data and accurately construct the functional relationship between parameters, such methods often have a complicated process and sometimes even have accuracy deviations.

[0004] Another type of method is based on signal decomposition techniques, such as wavelet transform and modal decomposition. By decomposing the signal at multiple scales, the dynamic strain and static strain can be effectively separated. For example, some researchers have used wavelet transform to identify the beam deflection caused by temperature and achieve response separation; some researchers have also used wavelet transform to decompose displacement data into three sub-components caused by temperature fluctuations, quasi-static effects, and dynamic factors, and then separate the thermally induced displacement component; some researchers have even proposed an enhanced empirical mode decomposition method that combines empirical mode decomposition, Hilbert marginal spectrum, and Gaussian mixture model clustering to achieve real-time decoupling of temperature and vehicle strain, with high intelligence and separation accuracy; some researchers have even solved the problems of mode aliasing and end effects through signal reconstruction and proposed a method for separating temperature-induced deflection, significantly improving the separation accuracy. Currently, the existing separation methods based on empirical mode decomposition can already achieve automatic separation of vehicle-induced strain and temperature strain, but pseudo eigenmode functions will be generated during the empirical mode decomposition process, increasing the difficulty of dynamic-static response separation.

[0005] However, with the continuous development of bridge health monitoring technology, the amount of data generated by the monitoring system has increased sharply, especially in large-scale bridge monitoring systems. The huge amount of data poses severe challenges to the health monitoring system in terms of storage, processing, and analysis. Long-term monitoring will generate a large amount of raw strain data, and how to efficiently store, manage, and back up these data has become a problem to be solved. To address this challenge, the compressive sensing method has emerged. It reduces the amount of sampled data and retains the main features of the signal, thus effectively reducing the burden of data storage and transmission. Some researchers have explored the application of compressive sensing in vibration data compression for bridge structural health monitoring, simulated the compressive sensing sensor algorithm, and verified its data compression ability through actual data. Some researchers have even applied compressive sensing to long-term monitoring of wireless sensors, demonstrating that this technology can reduce power consumption and the amount of data transmission. Some researchers have even proposed a bridge health monitoring method based on compressive sensing, which reduces the amount of data acquisition by optimizing the observation matrix and improves the signal reconstruction accuracy. Although the compressive sensing method can effectively reduce the amount of data acquisition, its disadvantage is that it requires a complex signal reconstruction process, which may affect real-time performance, especially in applications with high computational requirements, which may lead to delays. In addition, compressive sensing depends on the sparsity of the signal. For some complex or high-noise signals, the original information may not be fully recovered, which may cause loss of some details in the signal and affect the reconstruction accuracy.

[0006] In view of the above, this paper proposes an adaptive online truncation method for high-frequency vehicle-induced dynamic responses of bridges to solve the above problems. Summary of the Invention

[0007] In view of the deficiencies of the above-mentioned existing technologies, this study proposes an adaptive online truncation method for the high-frequency vehicle-induced response of bridges, aiming to reduce the data volume and effectively extract vehicle-induced strain. This method is real-time, self-adaptive, and easy to be embedded in existing monitoring systems. Under the high-frequency monitoring data environment, by using adaptive iteration and data segment post-processing technologies, the efficient real-time extraction of vehicle-induced strain is realized.

[0008] An adaptive online truncation method for the high-frequency vehicle-induced response of bridges, characterized by comprising the following steps:

[0009] S1: First, based on the characteristic that the short-term fluctuation of vehicle-induced strain is large, the total strain data is initially segmented into a pseudo-temperature strain data segment and a pseudo-vehicle-induced strain data segment by using the idea of smooth iteration;

[0010] S2: Subsequently, according to the length characteristics of the trivial vehicle-induced strain data segment, the vehicle-induced strain data segment in the pseudo-temperature strain data segment in S1 is correctly classified by a length segmentation method based on a threshold;

[0011] S3: Finally, according to the short-term range characteristics of the temperature strain data segment in the pseudo-vehicle-induced strain data segment, the temperature strain data segment is removed from the pseudo-vehicle-induced strain data segment in S1 by a screening method based on the range.

[0012] The beneficial effects of this technical solution are as follows:

[0013] (1) This method uses the smooth mean to perform iterative segmentation processing on the strain data, initially separating the pseudo-temperature strain data segment and the pseudo-vehicle-induced strain data segment. Since the pseudo-temperature strain data segment contains trivial vehicle-induced strain data segments and the pseudo-vehicle-induced strain data segment contains temperature strain data segments with a small range, the vehicle-induced strain data in the pseudo-temperature strain data segment and the pseudo-vehicle-induced strain data segment are respectively extracted. Finally, the online real-time extraction of high-frequency sampled vehicle-induced strain data is successfully realized, and it can operate stably within a set time window, meeting the technical requirements of the bridge health monitoring system for real-time response;

[0014] (2) In terms of data volume optimization, this method compresses the original data volume by more than 50% through effective data processing strategies, significantly reducing the storage and transmission costs, and is especially suitable for the long-term operation of the high-frequency lightweight monitoring system of medium and short-span bridges;

[0015] (3) Aiming at the interference problem in complex engineering environments, this method can still maintain a stable extraction ability under the noise interference in the range of 60 - 50 dB. Even under the condition of significant high-frequency noise interference, it can still effectively distinguish the vehicle-induced strain and temperature strain components, and has excellent anti-interference ability and environmental adaptability;

[0016] (4) This method can also accurately extract the strain response caused by vehicle loads in actual bridge monitoring and compress the data volume of the monitoring system in real time. Description of the Drawings

[0017] Figure 1 It is a schematic diagram of the overall process of the segmentation method of the present invention;

[0018] Figure 2 It is a schematic diagram of the geometric structure of the T-beam;

[0019] Figure 3 It is a schematic diagram of the annual temperature change curve in the embodiment of the present invention;

[0020] Figure 4 It is a schematic diagram of the vehicle weight change curve and the generated strain when the vehicle passes through the T-beam in the embodiment of the present invention;

[0021] Figure 5 It is a schematic diagram of the annual temperature strain curve in the embodiment of the present invention;

[0022] Figure 6 It is a schematic diagram of the annual vehicle-induced strain curve in the embodiment of the present invention;

[0023] Figure 7 It is a schematic diagram of the annual total strain curve with noise in the embodiment of the present invention;

[0024] Figure 8 It is a schematic diagram of the Allan variance calculation result of the present invention;

[0025] Figure 9 It is a schematic diagram of the initial smoothed mean and the results of 1 iteration, 2 iterations, and 6 iterations in the embodiment of the present invention;

[0026] Figure 10 It is a schematic diagram of the pseudo temperature strain data segment after reclassifying the vehicle-induced strain in the embodiment of the present invention;

[0027] Figure 11 It is a schematic diagram of the pseudo vehicle-induced strain data segment after removing the temperature strain in the embodiment of the present invention;

[0028] Figure 12 It is a schematic diagram of the cutting result within a certain 15-minute period in the embodiment of the present invention;

[0029] Figure 13 It is a schematic diagram of the annual strain segmentation result in the embodiment of the present invention;

[0030] Figure 14 It is a schematic diagram of the cutting result of each noise value strain part throughout the year in the embodiment of the present invention;

[0031] Figure 15 It is a schematic diagram of the sensor measuring point layout in the embodiment of the present invention;

[0032] Figure 16 Schematic diagram of the strain diagram of measuring point M-R-2-S in the embodiment of the present invention;

[0033] Figure 17 Schematic diagram of the final iteration result in the embodiment of the present invention;

[0034] Figure 18 Schematic diagram of the pseudo temperature strain data and the pseudo vehicle-induced strain data in the embodiment of the present invention;

[0035] Figure 19 Schematic diagram of the temperature strain data and the vehicle-induced strain data of the final cutting result in the embodiment of the present invention;

[0036] Figure 20 Schematic diagram of the cutting effect of the strain of measuring point M-R-2-S in the embodiment of the present invention;

[0037] Figure 21 Schematic diagram of the result of setting the vehicle-induced strain of measuring point M-R-2-S to zero in the embodiment of the present invention. Detailed implementation manners

[0038] Regarding the foregoing and other technical contents, features and effects of the present invention, they will be clearly presented in the following detailed description of the embodiments in conjunction with the attached Figures 1 - 21 In the detailed description of the embodiments, it will be clearly shown. The method contents mentioned in the following embodiments are all referenced to the accompanying drawings of the specification.

[0039] Embodiment 1, establish a method for real-time cutting of vehicle-induced strain, as Figure 1 shown, it is generally divided into three major steps: First, utilize the characteristic that the short-term fluctuation of vehicle-induced strain is large, and use the idea of smooth iteration to preliminarily divide the total strain data into a pseudo temperature strain data segment and a pseudo vehicle-induced strain data segment. Theoretically, the temperature strain data segment in the pseudo temperature strain data segment is only caused by temperature changes, and the vehicle-induced strain data segment in the pseudo vehicle-induced strain data segment is caused by the combined action of vehicle load and temperature. However, in actual operation, the obtained pseudo temperature strain data segment contains trivial vehicle-induced strain data segments, and the pseudo vehicle-induced strain data segment is mixed with temperature strain data segments with relatively small short-term ranges. Subsequently, according to the length characteristics of the trivial vehicle-induced strain data segments, the vehicle-induced strain data segments in the pseudo temperature strain data segment are correctly classified by a length segmentation method based on a threshold; finally, according to the short-term range characteristics of the temperature strain data segments in the pseudo vehicle-induced strain data segment, the temperature strain data segments are removed from the pseudo vehicle-induced strain data segment by a screening method based on the range.

[0040] Example 2. On the basis of Example 1, the vehicle-induced strain is initially segmented. Since the total strain data collected by the bridge health monitoring system within a certain period contains vehicle-induced strain and temperature strain, and the vehicle-induced strain has large fluctuations in a short time, while the temperature strain changes relatively smoothly. Therefore, when the size of the smoothing mean window is appropriate, the difference between the vehicle-induced strain and the smoothing mean will be significantly greater than the difference between the temperature strain and the smoothing mean. According to this principle, by setting an appropriate difference discrimination threshold, the vehicle-induced strain and temperature strain in the strain data can be initially separated. The specific steps are as follows:

[0041] Step 1: Based on the short-time fluctuation characteristics of the vehicle-induced strain and temperature strain, combined with the Allan variance to balance the smoothing effect of the signal and the retention of short-time fluctuations, adaptively determine the window size m of the smoothing mean, and perform smoothing mean processing on the strain data within the time period [t i , t i+1 to obtain the smoothing mean where is the k-th strain data in the j-th iteration within the time period [t

[0042] , t i , t i+1 , i = 1, 2, 3,..., n, j = 1, 2, 3..., N, k = 1, 2, 3,..., N j , N j is the number of strain data in the j-th iteration, is the k-th smoothing mean after the strain data in the j-th iteration is processed by the smoothing mean;

[0043] Step 2: Calculate the difference between the strain data and the smoothing mean the mean value of the differences and the standard deviation of the differences and set the difference discrimination threshold and σ j where is the k-th difference in the j-th iteration;

[0044] e j,k is the mean value of the differences in the j-th iteration; σ is the standard deviation of the differences in the j-th iteration; T j is the difference discrimination threshold in the j-th iteration; e,h Step 3: If the mean value of the differences e

[0045] is greater than the difference discrimination threshold T j,k , then mark e,j as pseudo vehicle-induced strain otherwise mark it as pseudo temperature strain ​

[0046] Step 4: Perform the (j + 1)-th smoothing process on the pseudo temperature strain to obtain the smoothed mean value Calculate the pseudo temperature strain and the difference e between the smoothed mean value j+1,k , and then calculate the change rate of the difference mean value

[0047] Step 5: If r is greater than 0.01, repeat Steps 2 - 5; otherwise, end the iteration. Since there is noise interference in the total strain data, the pseudo temperature strain contains a small amount of vehicle-induced strain data, and the pseudo vehicle-induced strain is mixed with temperature strain data.

[0048] Example 3: On the basis of Example 2, reclassify the vehicle-induced strain data segments in the pseudo temperature strain data segments. Since the initial iterative extraction of vehicle-induced strain is performed through a hard threshold, in the iterative segmentation process of the total strain, whether it is a vehicle-induced strain data segment or a temperature strain data segment, as long as the data greater than the difference threshold is regarded as vehicle-induced strain, and the strain data less than the difference threshold is classified as temperature strain. Therefore, as long as it is less than the set threshold, it will also be judged as temperature strain data. Because the vehicle-induced strain in the total strain fluctuates around the temperature strain, there must be a phenomenon that some local vehicle-induced strain segments are less than the difference threshold during iterative segmentation, resulting in the misclassification of vehicle-induced strain as temperature strain. This is also the reason why the data after the initial iterative segmentation is called pseudo vehicle-induced strain and pseudo temperature strain. In view of this inevitable phenomenon, it is necessary to reclassify the vehicle-induced strain in the pseudo temperature strain data segments. Considering that the vehicle-induced strain data segments mixed in the pseudo temperature strain data segments have the characteristics of unequal but relatively short lengths, such vehicle-induced strain data segments can be reclassified by setting an appropriate data segment length threshold to extract vehicle-induced strain to the greatest extent. The specific steps are as follows:

[0049] Step 1: Denote the respective lengths of all sub-data segments in the pseudo temperature strain data segment as l k , and count the occurrence frequencies of the sub-data segments with the same length l k ;

[0050] Step 2: Considering the characteristics of the length distribution of the pseudo temperature strain data segment , count the frequency distribution of the sub-data segments with the same length l k , and then, based on the amplitude of frequency change, determine the length of the sub-data segment corresponding to the maximum frequency change as the discrimination threshold T l of the sub-data segment length;

[0051] Step 3: The pseudo temperature data segment The length l of the neutron data segment k The sub-data segment smaller than the length discrimination threshold T l is regarded as the vehicle-induced strain data segment and reclassified into the pseudo vehicle-induced strain data segment Among them, the remaining data segments are marked as temperature strain data

[0052] Example 4. On the basis of Example 3, the temperature strain data segments in the pseudo vehicle-induced strain data segments are removed. Since the short-term fluctuation of vehicle-induced strain is relatively large, the vehicle-induced strain data segments in the pseudo vehicle-induced strain data segments have the characteristic that the short-term range is significantly greater than that of the temperature strain data segments. By setting the range discrimination threshold, the temperature strain data segments can be effectively removed from the pseudo vehicle-induced strain data segments. The specific steps are as follows:

[0053] Step 1: Calculate the range R of each sub-data segment in the pseudo vehicle-induced strain data segment k and arrange the range values in ascending order;

[0054] Step 2: Calculate the difference between two adjacent range values after sorting, select the point with the smallest difference as the jump point, and take one of the corresponding adjacent range values as the range discrimination threshold T R ;

[0055] Step 3: The data segments in the pseudo vehicle-induced strain data segment k whose range R is greater than the range discrimination threshold T R are regarded as vehicle-induced strain while the remaining data are marked as temperature strain

[0056] Example 5. On the basis of Example 2, the selection of the smoothing mean window size is carried out. Since the Allan variance can effectively evaluate the volatility of the signal and has been used to determine the window size of the simple moving average, especially in the case of noise and mutation, the Allan variance is also used in this method to set the size of the smoothing mean window. The Allan variance is defined as the mean square value of the second-order difference of the adjacent window means within a non-overlapping sliding window of length m, as shown in formula (1):

[0057]

[0058] In the formula, m is the smoothing mean window size; is the cumulative value of all data before the l-th strain data; N i is the time period [t i , t i+1The number of strain data received therein. When the smoothing window is small, the Allan variance increases rapidly as the window increases; when the smoothing window increases to a certain extent, the Allan variance tends to level off. The ideal window size should be in the stable region after the significant fluctuation stage of the Allan variance, so that sufficient signal characteristics can still be retained after the data is processed by the smoothed mean.

[0059] To adaptively determine the smoothing mean window size using the Allan variance, this method optimizes the window size by calculating the change rate CR of the Allan variance. When CR tends to be stable, the window size is determined. Specifically, if the smoothing mean window size changes continuously 50 times and CR remains basically unchanged, the calculation is stopped and the window size at the first change is selected as the final smoothing mean window size, as shown in formula (2):

[0060]

[0061] σ 2 (m) represents the Allan variance when the smoothing mean window size is m; σ 2 (m + 1) represents the Allan variance when the smoothing mean window size is m + 1.

[0062] Example 6, based on the above Examples 1 to 5, design the evaluation criteria for this method, which are divided into:

[0063] (1) Cutting accuracy, which is represented by the cutting rate and the false positive rate. The cutting rate is the ratio of the amount of target strain data in the cut strain data segment to the amount of target strain data in the total strain data, as shown in formula (3):

[0064]

[0065] P vs is the cutting rate of vehicle-induced strain, N v is the amount of vehicle-induced strain data in the cut strain data segment, N V is the amount of vehicle-induced strain data in the total strain data. The false positive rate is the proportion of non-target strain data in the cut strain data segment, as shown in formula (4):

[0066]

[0067] FPR is the false positive rate of vehicle-induced strain, N t is the amount of temperature strain data in the cut strain data segment, and N is the amount of cut strain data.

[0068] (2) Anti-noise ability. In practical applications, the data collected by the health monitoring system is often affected by various noises such as environmental interference, equipment errors, and signal attenuation. Therefore, it is necessary to discuss the impact of noise on the segmentation quality. The measurement noise is added to the original signal in the form of signal-to-noise ratio (SNR). By systematically adjusting the SNR, the monitoring conditions under different noise levels can be simulated. The calculation of SNR is shown in Equation (5):

[0069]

[0070] A signal represents the amplitude of the signal, and A noise represents the amplitude of the noise. A coefficient called anti-noise strength is used to quantify the anti-noise ability of this cutting method, as shown in Equation (6):

[0071]

[0072] D N and D S are the strain cutting rates obtained under the influence of noise and without noise respectively.

[0073] Example 7. On the basis of the above examples, this cutting method is verified. First, a finite element model is established. To verify the effectiveness of the proposed method, as Figure 2 shown, this method takes a certain T-beam bridge as the reference object, and a finite element model is established by selecting a single T-beam. The main beam is 2.3 m high, 2.5 m wide, and 50 m long, and it is a simply supported structure. To reduce the calculation amount and improve the calculation speed of the model, beam elements are used for modeling, and the selected element type is BEAM188. Three kinds of strains are required for the verification of the proposed method, namely pure temperature strain, pure vehicle-induced strain, and total strain under the simultaneous action of temperature and vehicle load. Therefore, this study conducts simulations of the working conditions of pure temperature action, pure vehicle load action, and simultaneous action of temperature and vehicle load. Among them, the simulation of the working condition of simultaneous action of temperature and vehicle load is realized by means of data combination.

[0074] The measured annual temperature changes at 6 measuring points of a certain concrete bridge are as Figure 3 shown. In this study, the pure temperature working condition is realized by applying only uniform temperature action on the T-beam section. The temperature of the beam section is the annual temperature change corresponding to the T6 measuring point (Note: Due to sensor failures, some annual temperature data is lost, but the selected temperature range still includes the maximum and minimum values of the annual temperature change and has a certain representativeness). The reference temperature in the finite element model is set to 20 °C, and the linear expansion coefficient is set to 1.2×10 -6 / °C.

[0075] For the pure vehicle operating conditions, only the case where the vehicle drives unidirectionally and uniformly through the center position of the beam section is considered, and concentrated forces of different magnitudes are used to simulate different vehicle weights. According to the fitting results of the measured vehicle weight data, the vehicle weights are selected as 15 kN, 25 kN, 35 kN, 45 kN, and 55 kN respectively. Given that rough road surfaces can cause fluctuations in the axle weights of vehicles on bridges, the concentrated forces simulated in this study vary randomly by 5%. Since the focus of this operating condition is on obtaining vehicle-induced strains and vehicle speed does not affect this cutting method, a simulated vehicle speed of 10 m / s is selected. To simulate high-frequency data sampling of medium and small-span bridges, the sampling frequency is set to 50 Hz. The magnitudes of the concentrated forces corresponding to different positions of the vehicle on the main beam are as shown in Figure 4 (a), and the strains at the bottom of the beam generated by each vehicle passing through the main beam are as shown in Figure 4 (b).

[0076] Then, the verification of the cutting effect of vehicle-induced strains is carried out. The strains under pure temperature conditions are as shown in Figure 5 , which has a significant positive correlation with the applied temperature. At the same time, when there is no temperature change, no temperature strains are generated. The strains under pure vehicle operating conditions are generated by randomly selecting vehicle-induced strains generated by different types of vehicles when passing through the T-beam and superimposing them in chronological order. The annual vehicle-induced strains are as shown in Figure 6 . The total strain is obtained by adding the pure vehicle-induced strain to the pure temperature strain. To more realistically simulate the actual strain data, Gaussian white noise of 60 dB is applied, and the total strain containing noise is as shown in Figure 7 .

[0077] To demonstrate the real-time cutting effect, in this embodiment, a 15-minute period, that is, a data segment of 45,000 data, is used to cut the annual strain data segment by segment to simulate the real-time cutting scenario. The variations of the Allan variance and the change rate of a certain 15-minute data under different smoothing window sizes are as shown in Figure 8 . When the smoothing window m increases from 0 to 30, the maximum increase in the Allan variance is up to 2.213. As the smoothing window further increases, the Allan variance gradually decreases to 0.5 and tends to be stable. When the smoothing window m is 256, the change rate of the Allan variance has remained basically unchanged for 50 consecutive times and automatically stops calculating. Since the larger the m, the smaller the calculation speed. To ensure the real-time performance of the algorithm as much as possible, the smoothing window size is finally selected as the m value corresponding to the initial stability of the change rate of the variance, that is, 206, as the optimal smoothing window. The strain data is processed by smoothing the mean according to the adaptively obtained smoothing window size.

[0078] The iterative process of the preliminary segmentation of vehicle-induced strains is as shown in Figure 9 , and the initial smoothing mean and the difference discrimination threshold are as shown in Figure 9As shown in (a), as expected, the data exceeding the differential discrimination threshold is the vehicle-induced strain part. The result of the first segmentation based on the initial differential discrimination threshold is shown in Figure 9 (b). Since with the increase of the number of iterations, the extracted vehicle-induced strain database may contain temperature strain data segments introduced due to excessive iterations. Therefore, the strain cut in this stage is called pseudo vehicle-induced strain. Subsequently, based on the analysis of the segmentation results, the differential discrimination threshold is dynamically adjusted, and multiple iterations are carried out. Each time the threshold is updated and the pseudo temperature strain data is segmented, so as to gradually cut out the pseudo vehicle-induced strain part. The result of the second iterative segmentation is shown in Figure 9 (c). As the iterative process progresses, the segmentation results of vehicle-induced strain and temperature strain data gradually tend to be stable. When the iteration reaches the convergence condition, the iteration stops. At this time, a total of 6 iterations are carried out. The result of the initial iterative segmentation is shown in Figure 9 (d).

[0079] To more intuitively analyze the pseudo temperature strain data segments obtained after the preliminary segmentation, a 5-minute pseudo temperature strain data segment is selected from the preliminary segmentation results for display, as shown in Figure 10 (a). The pseudo temperature strain data obtained by the preliminary segmentation consists of a large number of short-time data segments, which also contains some trivial vehicle-induced strain data segments. The frequency distribution of the data segment lengths is shown in Figure 10 (b), where the number of data segments with lengths less than 10 is significantly more than that of other lengths. Based on this, by adaptively identifying the position where the frequency of the data segment lengths changes significantly, and taking the data segment length at this position as the discrimination threshold. According to this threshold, the part with a data segment length less than 10 is reclassified into the pseudo vehicle-induced strain data segments. The reclassified pseudo temperature strain data segments are shown in Figure 10 (c). However, in the process of reclassifying the trivial vehicle-induced strain data segments into the pseudo vehicle-induced strain data segments, some temperature strain data segments with similar lengths may also be misclassified. Therefore, to improve the accuracy of data classification, the pseudo vehicle-induced strain data segments need to be further processed.

[0080] The reclassified pseudo vehicle-induced strain data is shown in Figure 11 (a), which contains many temperature strain data segments with very small ranges. The frequency distribution of the data segment ranges is shown in Figure 11 (b). The range can intuitively reflect the short-time fluctuation characteristics of the data segment. In contrast, the range of the vehicle-induced strain data segments is significantly larger than that of the temperature strain data segments. Adaptively identify the position where the range changes significantly, and take the corresponding range value at this position as the range discrimination threshold. Subsequently, the data segments with a range less than this threshold are marked as temperature strain data segments and excluded. The finally obtained vehicle-induced strain data segments after excluding the temperature strain data segments are shown in Figure 11(c). After the above processing, the vehicle-induced strain data segment was successfully cut. The final segmentation results of the 15-minute time segment data are as Figure 12 shown.

[0081] Next, in chronological order, segment-by-segment cutting was performed on the total annual strain data to simulate the real-time cutting effect in actual engineering. It should be noted that the data length can be adjusted according to actual needs, such as being selected as the duration or data volume corresponding to a single data storage file in the health monitoring system of medium and small-span bridges. In this embodiment, 15 minutes was selected as the time window unit for cutting, and the cutting process is as Figure 13 (a). The annual results of segment-by-segment cutting are as Figure 13 (b). The final cutting rate of vehicle-induced strain was 98.01%, and the misjudgment rate was 0.89%. The misjudged data of the proposed vehicle-induced cutting method was mainly the strain of the free vibration of the bridge after the vehicle crossed the bridge, as Figure 13 (b). In some cases, the strain data during the free vibration of the bridge was almost equal to the noise data in terms of the fluctuation amplitude. Since the proposed method automatically cuts vehicle-induced strain through the amplitude threshold, the result was finally misjudged. Since this method pays more attention to the strain of the main girder during vehicle loading, and the free vibration of the bridge occurs after the vehicle crosses the bridge, the dynamic strain during free vibration can be considered as both vehicle-induced strain and non-vehicle-induced strain, with a large degree of subjectivity. Considering that the main purpose of this study is to extract the dynamic strain data during vehicle loading, therefore, such dynamic strain data was not included in the statistical range of the misjudgment rate.

[0082] Finally, the influence of noise was verified. The cutting results of strain data under different degrees of noise are as Figure 14 shown. As the noise amplitude increases, the noise gradually approaches the vibration strain generated after the vehicle passes through the bridge. Therefore, the noise mainly affects the vibration strain part. As the noise amplitude increases, the cutting rate and misjudgment rate of vehicle-induced strain gradually increase. The cutting accuracy under different degrees of noise is shown in Table 1:

[0083] Table 1 Noise resistance strength

[0084]

[0085] Example 8 was applied to a real bridge to verify the general situation of bridge and lightweight monitoring. The proposed method was applied to a certain T-shaped continuous beam bridge structure. This bridge has several spans, each span is 50 meters long, the beam height is 2.5 meters, and the bridge deck is a one-way four-lane. Since heavy vehicles travel in the second and third lanes, and the mid-span section of the bridge structure is a typical stress area of the maximum bending moment of the main girder, it can directly reflect the characteristics of vehicle-induced strain. The diaphragm beam can synchronously characterize the lateral cooperative deformation and temperature effect by rigidly connecting adjacent main girders. For the above reasons, a total of 12 sensors were arranged at the mid-span position of the third main girder and its diaphragm beam, asFigure 15 As shown, it includes a strain sensor and a temperature sensor. To ensure the acquisition of complete vehicle-induced strain data, the sampling frequency of the monitoring system should be at least 8 times the natural vibration frequency of the bridge. The natural vibration frequency of this bridge is approximately 2.5 Hz, and the sampling frequency of the monitoring system is set to 50 Hz. The strain measured at measuring point M-R-2-S from January 4th to 6th, 2025 is as Figure 16 shown. It can be seen from the strain data that the strain amplitudes generated by some small vehicles are relatively small, which is not much different from the temperature strain fluctuations caused by the sensor accuracy limitations.

[0086] Since the M-R-2-S strain sensor is located at the bottom of the mid-span beam, the maximum dynamic strain change is collected, which is beneficial to the segmentation of vehicle-induced strain. Therefore, in this study, it is used as a reference to determine the time points corresponding to vehicle-induced strain to cut the data of all measuring points. First, calculate the Allan variance under different smoothing mean window sizes, and the adaptively selected window size is 468. Use this window size to perform initial smoothing mean processing on the strain data. Subsequently, automatically set the difference discrimination threshold and perform iterative segmentation. The final iterative result is as Figure 17 shown. The pseudo-temperature strain data segments and pseudo-vehicle-induced strain data segments generated during the iterative process are as Figure 18 shown. There are trivial vehicle-induced strain data segments in the pseudo-temperature strain data segments, and there are temperature strain data segments with relatively small ranges in the pseudo-vehicle-induced strain data segments. Then, adaptively calculate the length discrimination threshold and range discrimination threshold of the trivial data segments, and perform data cutting. The final cutting result is as Figure 19 shown. It can be seen from Figure 19 (a) that some relatively small vehicle-induced strains are misjudged as temperature strains. This phenomenon is mainly due to the fact that the strains generated by light vehicles are relatively small, and the difference from the noise signal is relatively weak, resulting in difficulty in accurately identifying them. Nevertheless, the assessment of the bridge bearing capacity mainly uses the vehicle-induced strains of heavy vehicles, and the influence of the unrecognized vehicle-induced strains of light vehicles on the bridge is very small. Therefore, the misjudged vehicle-induced strains of light vehicles can be accepted.

[0087] Perform real-time cutting on the strain data in chronological order. The final cutting effect is as Figure 20 shown, and the vehicle-induced strains generated by heavy vehicles have been successfully cut. To achieve the baseline zeroing of the vehicle-induced strain signal, first, average the strain values in the adjacent intervals at both ends of the cut strain segment as the estimated value of the temperature strain for this period. Subsequently, subtract this estimated value from the entire vehicle-induced strain data to eliminate the influence of temperature strain on the vehicle-induced strain signal and achieve the zeroing of the vehicle-induced strain signal. In terms of real-time performance, this method performs online cutting on the bridge vehicle-induced strain data in a block-by-time order, and can identify the characteristics of vehicle-induced strain in real time. This efficient data cutting method not only achieves a fast response in actual bridge monitoring, but also provides reliable technical support for real-time health monitoring.

[0088] All dynamic data of the bridge are cut according to the timestamps obtained by segmenting the measurement data at the M-R-2-S measuring point. The storage size of the strain data collected within 20 days is 7055568 KB. After cutting, only the vehicle-induced strain data segment is retained, and its storage capacity is 3282584 KB, a reduction of approximately 54%. It can be seen that the proposed method can compress the data storage capacity of the monitoring system in real time.

[0089] For the lightweight monitoring of medium and small-span bridges, the proposed method has three benefits: First, the proposed algorithm has high real-time performance, a simple process, and low computing power requirements, and is very easy to embed into the monitoring system; Second, it can significantly compress the data volume in the high-frequency health monitoring system, reducing the system storage and data transmission costs; Finally, the proposed algorithm completely retains the original vehicle-induced strain data while compressing the data storage size, ensuring the integrity of the vehicle-induced strain, and providing a reliable data source for subsequent structural analysis and evaluation based on vehicle-induced strain.

[0090] The above description is only for the purpose of illustrating the present invention. It should be understood that the present invention is not limited to the above embodiments, and various flexible forms conforming to the idea of the present invention are within the protection scope of the present invention.

Claims

1. An adaptive online intercepting and storing method for high-frequency vehicle-induced responses of bridges, characterized in that, It includes the following steps: S1: First, according to the characteristic that the vehicle-induced strain has large short-term fluctuations, the total strain data is preliminarily segmented into pseudo-temperature strain data segments and pseudo-vehicle-induced strain data segments by using the idea of smooth iteration; S2: Subsequently, according to the length characteristics of the trivial vehicle-induced strain data segments, the vehicle-induced strain data segments in the pseudo-temperature strain data segments in S1 are correctly classified by the length segmentation method based on a threshold; S3: Finally, according to the short-term range characteristics of the temperature strain data segments in the pseudo-vehicle-induced strain data segments, the temperature strain data segments are removed from the pseudo-vehicle-induced strain data segments in S1 by the screening method based on the range.

2. The adaptive online intercepting and storing method for high-frequency vehicle-induced response of a bridge according to claim 1, characterized in that The specific steps of S1 include: S1-1: Based on the short-time fluctuation characteristics of vehicle-induced strain and temperature strain, combined with Allan variance, balance the smoothing effect of the total strain data signal and the retention ability of short-time fluctuations, adaptively determine the window size m of the smoothing mean, and for the strain data within the time period [t i , t i+1 , perform smoothing mean processing to obtain the smoothing mean Among them, is the k-th strain data in the j-th iteration during the time period [t i , t i+1 . Here, i = 1, 2, 3, …, n, j = 1, 2, 3, …, N, k = 1, 2, 3, …, N j , and N j is the number of strain data in the j-th iteration. is the k-th smoothed mean value of the strain data after smoothed mean processing in the j-th iteration; S1-2: Calculate strain data The difference from the smoothed mean The difference The mean difference And the standard deviation of the difference And based on And σ j Set the difference discrimination threshold S1-3: If the difference e j,k is greater than the difference discrimination threshold T e,j , then is marked as pseudo vehicle-induced strain Otherwise, it is marked as pseudo temperature strain S1-4: Perform the (j + 1)-th smoothing process on the pseudo temperature strain to obtain the smoothed mean value Calculate the pseudo temperature strain and the difference e between the smoothed mean value j+1,k , and then calculate the change rate of the difference mean value S1-5: If r j is greater than 0.01, repeat steps S1-2 to S1-5; otherwise, end the iteration. Due to noise interference in the total strain data, a small amount of vehicle-induced strain data is included in the pseudo temperature strain, and temperature strain data is also mixed in the pseudo vehicle-induced strain.

3. An adaptive online intercepting and storing method for high-frequency vehicle-induced response of a bridge according to claim 2, characterized in that, The specific steps of S2 include: S2-1: Denote the respective lengths of all sub-data segments in the pseudo temperature strain data segment as l k , and count the occurrence frequencies of sub-data segments with the same length l k ; S2-2: Given the pseudo temperature strain data segment Characteristics of the length distribution, statistically count l k Frequency distribution of sub-data segments with the same length, and then based on the amplitude of the frequency change, determine the length of the sub-data segment corresponding to the largest frequency change as the discrimination threshold T of the sub-data segment length l ; S2-3: Treat the pseudo-temperature data segment The length l of the neutron data segment k Less than the length discrimination threshold T l The sub-data segment is regarded as the vehicle-induced strain data segment and reclassified into the pseudo-vehicle-induced strain data segment In it, and the remaining data segments are marked as temperature strain data 4. An adaptive online intercepting and storing method for high-frequency vehicle-induced response of a bridge according to claim 3, characterized in that, The specific steps of S3 include: S3-1: Calculate the range R of each sub-data segment in the pseudo vehicle-induced strain data segment and arrange the range values in ascending order; k ​ S3-2: Calculate the difference between adjacent extreme differences after sorting, select the point with the smallest difference as the jump point, and use one of the corresponding adjacent extreme differences as the extreme difference discrimination threshold T R ; S3-3: Consider the range R of each sub-data segment in the pseudo vehicle-induced strain data segment greater than the range discrimination threshold T k as the vehicle-induced strain R and mark the remaining data as temperature strain 5. An adaptive online intercepting and storing method for high-frequency vehicle-induced response of a bridge according to claim 2, characterized in that The process of balancing the signal by combining the Allan variance includes: defining the Allan variance as the mean square value of the second-order difference of the adjacent window means within a non-overlapping sliding window of length m, as shown in formula (1): where m is the size of the smoothing mean window, is the cumulative value of all data before the l-th strain data, and N i is the number of strain data received within the time period [t i , t i+1 .

6. An adaptive online intercepting and storing method for high-frequency vehicle-induced response of a bridge according to claim 2, characterized in that, The process of adaptively determining the window size m of the smooth mean includes: optimizing the window size by calculating the change rate CR of the Allan variance, as shown in formula (2): σ 2 (m) represents the Allan variance when the smoothing mean window size is m, σ 2 (m + 1) represents the Allan variance when the smoothing mean window size is m + 1. When CR tends to be stable, the window size is determined.

Images