High-precision dynamic acquisition instrument

Abstract

The invention relates to the technical field of high-precision acquisition instruments, and particularly discloses a high-precision dynamic acquisition instrument. Comprising a signal acquisition module, a signal processing module and a data output module. The signal acquisition module acquires bridge monitoring data according to a basic sampling frequency, and adjusts the sampling frequency in real time according to a unified threshold value corresponding to the data; the signal processing module obtains target data through multi-stage processing such as preprocessing, key information extraction, integration fusion and spatial-temporal characteristic analysis, error compensation and high-precision fitting; and the data output module processes and outputs the target data by means of a real-time complete output guarantee mechanism. The acquisition instrument effectively solves the problems that in the prior art, a bridge monitoring environment is complex, original data is prone to being interfered by noise of a sensor, environmental electromagnetic interference is caused, non-structural factors are mixed with structural signals, and the data precision is low due to the fact that noise is not filtered thoroughly in a traditional processing mode.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of high-precision acquisition instruments, and particularly relates to a high-precision dynamic acquisition instrument. BACKGROUND

[0002] In the field of bridge monitoring, accurately and efficiently obtaining data is crucial for evaluating the health status of bridge structures and ensuring traffic safety. With the development of technology, bridge monitoring has gradually shifted from traditional manual periodic detection to automation and real-time monitoring, and a large number of sensors have been deployed to collect various types of data. However, current bridge monitoring data acquisition faces many challenges.

[0003] The bridge monitoring environment is extremely complex, and the raw data is easily affected by various disturbances. On the one hand, sensors themselves have noise, such as zero drift of strain gauges and quantization noise of accelerometers, which will superimpose with the true signal and reduce the signal-to-noise ratio of the raw data. At the same time, electromagnetic radiation generated by power equipment, communication base stations and other devices in the surrounding environment of the bridge introduces random noise in the process of sensor signal transmission, seriously affecting the data quality. On the other hand, non-structural factors also interfere with the monitoring data. For example, temperature fluctuations can cause thermal expansion and contraction of bridge materials, mixing unrelated signals to the structure stress in the monitoring data; accidental factors such as temporary construction vibration can also mix with the structural signals of the bridge itself, making the data more complex.

[0004] Traditional data processing methods cannot effectively deal with these complex disturbances. Common data filtering algorithms, such as simple low-pass and high-pass filters, cannot accurately identify and remove the above complex noise, resulting in incomplete noise filtering, and ultimately making the processed data unreliable and difficult to accurately reflect the true structural state of the bridge. This not only affects the judgment of the current health status of the bridge, but in the long run, the analysis and prediction of the structural performance of the bridge based on these inaccurate data may lead to incorrect maintenance decisions, threatening the safe operation of the bridge.

[0005] For example, in "CN107340748A, Bridge Monitoring Data Acquisition Instrument", the acquisition instrument also aims to obtain bridge monitoring data, but when facing noise and interference problems, it only uses conventional hardware anti-interference measures, such as shielded cables to reduce electromagnetic interference, and simple data smoothing algorithms to process data. It lacks effective and targeted processing mechanisms for sensor self-noise and the mixing of non-structural factors and structural signals in complex environments, making it difficult to meet the high-precision requirements of today's bridge monitoring data. SUMMARY

[0006] In view of this, embodiments of the present invention provide a high-precision dynamic data acquisition instrument to solve the technical problems in the prior art where the raw data is easily affected by sensor noise, environmental electromagnetic interference, and the mixture of non-structural factors and structural signals due to the complex bridge monitoring environment, and the incomplete noise filtering of traditional processing methods, resulting in low data accuracy.

[0007] This invention provides a high-precision dynamic data acquisition instrument, comprising: The signal acquisition module uses a basic sampling frequency to sample dynamic bridge monitoring data and adjusts the sampling frequency in real time according to a unified threshold corresponding to the bridge monitoring data. The signal processing module acquires the bridge monitoring data collected in real time by the signal acquisition module, and processes the bridge monitoring data through multi-level processing to obtain the target data. The data output module processes the massive amount of target data through a real-time complete output guarantee mechanism and divides it into several data packets at expected time intervals, and then outputs the data packets to the bridge monitoring center system, data storage server and / or related analysis platform; The bridge monitoring data includes structural response data and external load data.

[0008] Preferably, the step of sampling dynamic bridge monitoring data using a basic sampling frequency and adjusting the sampling frequency in real time according to a unified threshold corresponding to the bridge monitoring data includes: Obtain the original data x; The original data is normalized to obtain a uniform threshold. Sampling is performed according to the sampling frequency corresponding to the unified threshold; The unified threshold includes a first threshold T1 and a second threshold T2, where T1 < T2; The sampling frequency includes a basic sampling frequency, a first sampling frequency, and a second sampling frequency.

[0009] Preferably, the normalization process includes: The original data x is normalized according to the formula x′=(x-x_min) / (x_max-x_min) to obtain normalized data x′; The normalized data x′ is matched with the unified threshold. If the match is successful, sampling is performed according to the corresponding sampling frequency. If the match fails, sampling is performed according to the basic sampling frequency. Wherein, a successful match is achieved when x′ is triggered at least N consecutively at the corresponding sampling frequency, and N is a positive integer; Where x_min is the minimum theoretical value of this type of data, and x_max is the maximum design value of this type of data.

[0010] Preferably, matching the normalized data x′ with the unified threshold, and if the match is successful, sampling is performed according to the corresponding sampling frequency, including: When x′<T0, the signal acquisition module is controlled to sample at the basic sampling frequency f1; When T0≤x′<T1, the signal acquisition module is controlled to sample at the second sampling frequency f2, where f2>f1; When x′≥T1, the signal acquisition module is controlled to acquire data at a third frequency f3, where f3>f2. Preferably, the multi-level processing includes: Clean data is obtained by preprocessing real-time bridge monitoring data; Key information reflecting the structural state of the bridge is extracted from clean data to obtain key frame data; Enhanced frame data is obtained by integrating and fusing keyframe data and strengthening effective information and weakening interference through spatiotemporal feature analysis. The target data is obtained by performing error compensation and high-precision fitting on the enhanced frame data and reducing the influence of random noise.

[0011] Preferably, the step of extracting key information reflecting the structural state of the bridge from the clean data and obtaining key frame data includes: Calculate time-domain statistics using a sliding window or a fixed time interval; The cleanliness data in the time domain is converted into a frequency domain signal using a Fourier transform; By combining the dynamic model of the bridge structure, a deep correlation analysis of the time domain and frequency domain characteristics is performed to extract modal frequencies and mode shape parameters. The keyframe data is obtained by calibrating the model parameters by comparing the natural frequencies identified in the frequency domain with the theoretical modal frequencies of the dynamic model.

[0012] Preferably, the process of integrating and fusing keyframe data and enhancing effective information and weakening interference through spatiotemporal feature analysis to obtain enhanced frame data includes: Kalman filtering algorithm is used to integrate key frame data of the same type of data collected by multiple sensors and different types of data at the same location; Analyze the spatial correlation of different keyframe data to determine whether it conforms to the mechanical transmission law of bridge structure. Determine whether the trend of keyframe data changes over time is reasonable, weaken or remove false features, and enhance effective information to obtain enhanced frame data.

[0013] Preferably, the real-time complete output guarantee mechanism includes: Compress the target data; Establish a tiered transmission mechanism; Establish a dynamic bandwidth adaptation mechanism; The data is packaged and output according to the collection frequency and type.

[0014] Preferably, the compression process of the target data includes: The sliding window technique is activated to dynamically scan the target data according to the set initial window size; Within each dynamically selected window range, the target data is scanned byte by byte using the LZ77 algorithm to identify repeating patterns and redundant information in the target data; By using the DEFLATE algorithm and combining the processing results of LZ77, these repetitive patterns and redundant information are encoded and converted to obtain compressed target data.

[0015] Preferably, the establishment of the hierarchical transmission mechanism includes: For critical data, dedicated network resource channels are allocated using 5G slicing technology to achieve real-time output; For regular data, it is cached in batches and then packaged and transmitted in batches according to the set time intervals.

[0016] The high-precision dynamic data acquisition instrument provided by this invention has the following beneficial effects: This high-precision dynamic data acquisition instrument, through a multi-level processing mechanism, effectively addresses issues such as sensor noise, environmental electromagnetic interference, and the mixing of non-structural factors with structural signals in bridge monitoring, significantly improving data reliability and accuracy. Compared to existing technologies that only employ conventional hardware anti-interference and simple data smoothing algorithms, its processed monitoring data more accurately reflects the true structural state of the bridge, providing a reliable basis for assessing bridge health status, analyzing and predicting structural performance, and making maintenance decisions, thus strongly ensuring bridge operational safety. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, and these are all within the protection scope of the present invention.

[0018] Figure 1 This is a structural schematic diagram of a high-precision dynamic data acquisition instrument; Figure 2 This is a flowchart illustrating the overall workflow; Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. In the description of the present invention, it should be understood that the terms "center," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements, but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, the element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. Where there is no conflict, embodiments of the present invention and the various features thereof can be combined with each other, all of which are within the scope of protection of the present invention.

[0020] Example 1 Please see Figure 1 This invention provides a high-precision dynamic data acquisition device; the acquisition device converts changes in the physical world (such as acceleration, temperature, and pressure) into electrical signals through sensors, then translates them into physical quantity data with timestamps, ultimately allowing a specific third party (engineer or system) to know "how something changed at a certain time".

[0021] For example, the working process of a data acquisition instrument for monitoring bridges is as follows: acquire bridge vibration - sensor voltage - data processing - output "vibration acceleration values ​​of each point on the bridge at each moment", and a specific third party can use these values ​​to determine whether the bridge has structural hazards.

[0022] In this embodiment, the data acquisition device includes a signal acquisition module.

[0023] The signal acquisition module includes a sensor and an AD module. The sensor is responsible for converting physical quantities into analog electrical signals, but these signals cannot be directly output as digital information. The AD module, as the built-in core unit of the information acquisition module, amplifies, filters, samples, and quantizes the analog signals output by the sensor, and finally converts them into digital signals. These digital signals are the original monitoring information output by the information acquisition module for use by the subsequent data processing module.

[0024] Specifically, when devices such as accelerometers and strain sensors on a bridge detect physical signals such as structural vibration and deformation, and convert them into analog voltage / current signals, these analog signals cannot be directly recognized and analyzed by subsequent digital systems. At this point, the AD module needs to intervene immediately, amplifying the weak analog signals output by the sensors through amplification circuits, removing environmental interference through filtering functions, and then performing analog-to-digital conversion at a sampling rate that matches the signal characteristics, converting the continuous analog signals into discrete digital signals.

[0025] Furthermore, at the start of signal acquisition, the signal acquisition module uses a basic sampling frequency to sample the dynamic bridge monitoring data, and adjusts the sampling frequency in real time according to the unified threshold corresponding to the bridge monitoring data; the bridge monitoring data includes structural response data and external load data.

[0026] The signal acquisition module includes various sensors, which must be directly fixed to key parts of the bridge (such as piers and beams). This is because their function is to sense the bridge's vibrations at close range; when the bridge vibrates, the sensitive elements inside the sensors move accordingly, generating minute voltage changes.

[0027] When the high-precision dynamic acquisition instrument is in the startup state, and the external sensor senses the dynamic signal and converts it into an electrical signal output, the signal acquisition module immediately begins to acquire the signal. Throughout the dynamic signal monitoring process, the signal acquisition module maintains a continuous connection with the external sensor. Once the sensor outputs a dynamic signal, it receives the signal in real time and immediately executes adaptive sampling control and multi-channel synchronous processing. In other words, as long as the acquisition instrument is operational and the sensor outputs a dynamic signal, the signal acquisition module will continuously acquire signals to ensure the complete capture of the dynamic signal's changes, providing continuous and effective raw data for subsequent processing modules.

[0028] Specifically, the structural response data includes: Vibration-related parameters: acceleration, velocity, displacement (vibration direction, frequency, amplitude); Mechanics: Stress, strain, cable force, support reaction force; Deformation-related: deflection, settlement, tilt, expansion joint displacement; Damage types: crack width / length, rebar corrosion rate, connector defects; The external load data includes: Traffic load: vehicle weight (axle load / gross weight), vehicle speed, traffic volume, wheel pressure; Environmental loads: wind speed / direction, temperature (environment / structure interior), rainfall, water level / scour depth.

[0029] Specifically, in this embodiment, the collection of structural response data and external load data complements each other for bridge monitoring. The combination of the two can achieve comprehensive monitoring from two dimensions: the structural state itself and the external stimuli.

[0030] The structural response data directly reflects the health status of a bridge and identifies potential risks. For example, vibration acceleration, strain, and displacement data provide immediate feedback on the bridge under external forces and can be used to assess structural safety in real time. Vibration frequency and peak strain data can directly determine whether the bridge is within its normal stress range. For instance, a sudden increase in strain in a span of a beam exceeding the design threshold may indicate cracking in local components or loosening of connections, requiring immediate warning. Continuous monitoring of cumulative displacement and changes in vibration damping ratio can reflect long-term trends such as structural aging and creep. For example, the gradual increase in foundation settlement can predict the potential risk of uneven settlement. External load data can clearly identify the causes of structural response and distinguish between normal and abnormal changes. For example, high summer temperatures cause increased thermal expansion displacement of the beam. If only displacement data is considered, it may be misjudged as structural deformation. Combining temperature data can distinguish between normal thermal expansion and abnormal settlement. By statistically analyzing data such as the frequency of heavy vehicle traffic and the proportion of overload, the impact of cumulative load on the fatigue life of bridges can be analyzed, providing a basis for maintenance plans.

[0031] In this embodiment, response-cause correlation analysis is achieved by combining two types of data, thereby improving the accuracy of monitoring. When the structural response is abnormal, it can be quickly determined whether it is a temporary impact of the load or damage to the structure itself by matching the external load at the same time. In the long term, by establishing a load-response correlation model, the structural state under future loads can be predicted, thereby upgrading from passive monitoring to active prevention.

[0032] When the signal acquisition module starts working, the sensor uses a basic sampling frequency to acquire raw data x; and adjusts the sampling frequency according to a unified threshold range reached by the raw data x; the unified threshold includes a basic threshold T0, a first threshold T1, and a second threshold T2, where T1 < T2; wherein the sampling frequency includes a basic sampling frequency, a first sampling frequency, and a second sampling frequency.

[0033] Furthermore, bridges themselves exhibit inherent, subtle vibrations that are difficult to detect. If data collection is continuously performed during bridge monitoring, this data generated by these inherent vibrations is not only meaningless for analysis and judgment (considered invalid data), but also consumes significant unnecessary resources, such as data storage space and processing time. Therefore, to conserve resources and improve monitoring efficiency, setting a low-power standby monitoring mode is essential. Thus, we set three different threshold values: a base threshold T0, a first threshold T1, and a second threshold T2. This multi-layered threshold setting allows for more precise control of monitoring mode switching, thereby achieving effective resource utilization and efficient monitoring operations.

[0034] The data collected by the data acquisition device is diverse and lacks uniformity, thus requiring the setting of unified thresholds for various data types. However, with numerous data categories, setting a threshold for each type would lead to an overly complex and difficult-to-manage system. Therefore, to achieve unified quantification of this data, specific methods or formulas must be used to transform this inconsistent data into a unified standard. Only after this is done can subsequent adjustments to the sensor's operating mode be made easier. If multiple different basic thresholds, first thresholds, and second thresholds exist, the sensor's operation will become chaotic and disordered, making effective management and operation difficult.

[0035] Therefore, the original data x needs to be normalized to obtain a uniform threshold. Specifically, the original data x is normalized according to the formula x′=(x-x_min) / (x_max-x_min) to obtain normalized data x′; The normalized data x′ is matched with the unified threshold. If the match is successful, sampling is performed according to the corresponding sampling frequency. If the match fails, sampling is performed according to the basic sampling frequency. Where x_min is the minimum theoretical value of this type of data, and x_max is the maximum design value of this type of data.

[0036] The basic threshold is set higher than the maximum value of environmental noise but lower than the upper limit of the 90% confidence interval for the bridge's own micro-motion. For small-to-medium span beam bridges, ±0.005g to ±0.008g is used; for large-span flexible bridges (such as cable-stayed bridges), due to their slightly larger micro-motion, ±0.008g to ±0.01g can be used. The sensor's built-in sleep / wake-up function is used, activating the main acquisition module only when the vibration amplitude exceeds the basic threshold.

[0037] The first threshold is set to cover the peak vibration when 95% of small vehicles (load < 5 tons) pass by. The reference value is ±0.01g to ±0.03g; due to vibration attenuation of bridge piers, it can be ±0.008g to ±0.02g.

[0038] The second threshold is set to be reached when heavy vehicles or abnormal events occur. It is set higher than the maximum vibration amplitude of small vehicles, covering 90% of heavy trucks (load > 10 tons) and possible abnormal events (such as impacts caused by localized damage to the bridge deck). The reference value is ±0.05g~±0.1g (bridge deck); for bridge piers, it is ±0.03g~±0.08g (because the threshold needs to be lower than that of the bridge deck due to vibration transmission attenuation).

[0039] Therefore, when x′<T0, the signal acquisition module is controlled to sample at the basic sampling frequency f1; when the basic threshold is below, the system operates in a sleep + intermittent wake-up mode (e.g., wake up once every 1 second and sample 1 data point), and the power consumption can be reduced to more than 1 / 100 of that of continuous acquisition.

[0040] When T0≤x′<T1, the signal acquisition module is controlled to sample at the second sampling frequency f2, where f2>f1; at this time, the sampling frequency is 50~100Hz, which is sufficient to capture the wheel impact frequency when the car passes by, corresponding to the main frequency components of the structural response.

[0041] When x′≥T1, the signal acquisition module is controlled to acquire data at a third frequency f3, where f3>f 2. At this point, the sampling frequency is 500~2000Hz, which needs to capture instantaneous impact peaks, such as the high-frequency vibrations during the sudden braking of a heavy truck, to avoid the peaks being flattened due to insufficient sampling rate. The sampling frequency automatically drops after the event ends to reduce unnecessary energy consumption.

[0042] Furthermore, it's possible that a certain value might suddenly exceed a set threshold, only to quickly fall below it in the next instant. This would cause the sampling frequency to fluctuate repeatedly. Such fluctuations in the sampling frequency directly increase energy consumption, which is an undesirable outcome.

[0043] Therefore, to address this situation and avoid unnecessary energy consumption increases, this embodiment sets the sampling frequency to be triggered at least N consecutive times for a successful match to be determined, where N is a positive integer. This setting effectively reduces unnecessary repeated changes in the sampling frequency caused by short-term numerical fluctuations, thereby reducing energy consumption.

[0044] For example, when a heavy truck crosses a bridge, the vibration signal suddenly increases from 0.05g to 0.5g within 0.1 seconds; the nth point is 0.05g; the (n+1)th point is 0.1g; and the (n+2)th point is 0.3g. The threshold of these three consecutive points exceeds the second threshold. The data acquisition instrument immediately increases the sampling rate from the basic sampling frequency or the first sampling frequency to the second sampling frequency to capture the instantaneous peak value during the impact process with higher density.

[0045] After the truck passed, the vibration signal gradually attenuated to 0.05g. When the threshold values ​​at three consecutive sampling points were lower than the base sampling frequency, the first sampling frequency, or the second sampling frequency, the data acquisition instrument determined that the signal had returned to a stable state. Then Change the sampling frequency to reduce redundant data.

[0046] Through the above mechanism, storage and computing resources can be saved when the signal is stable, and dynamic details can be accurately captured when the signal changes abruptly, so as to collect real-time bridge monitoring data and realize high-precision dynamic monitoring with on-demand sampling.

[0047] Furthermore, if no signal is received from the sensor for an extended period, the system will automatically enter standby mode. It's important to understand that sensors can malfunction or become inoperable during operation, which will prevent signal transmission.

[0048] In a high-precision dynamic data acquisition instrument, to determine whether a sensor is emitting no signal or has malfunctioned, the signal acquisition module also includes a sensor status detection unit. This unit is configured to initiate a sensor detection process when the data acquisition instrument is in the startup and running state and has not received a sensor output signal for a preset time (e.g., 5 minutes). Send a preset excitation signal to the sensor.

[0049] The monitoring sensor receives a feedback response after receiving an excitation signal. If a feedback signal that meets the preset characteristics is received within a preset time (e.g., 10 seconds), the sensor is determined to be in normal condition but no dynamic signal is emitted. At this time, the data acquisition instrument enters standby mode (basic sampling frequency sampling mode). If no feedback signal is received or the feedback signal does not meet the preset characteristics, it is determined that the sensor is malfunctioning or incapacitated. The data acquisition unit triggers the alarm mechanism and sends fault information to the terminal.

[0050] Through the above detection process, the sensor's no-signal state can be accurately distinguished, ensuring that the data acquisition instrument enters standby mode appropriately when the sensor is normal and has no signal, and provides timely warning when the sensor fails or becomes inoperable.

[0051] Furthermore, the data acquisition device also includes a signal processing module.

[0052] The bridge monitoring data is acquired in real time by the signal acquisition module, and the target data is obtained after processing the bridge monitoring data through multi-level processing.

[0053] Clean data is obtained by preprocessing real-time bridge monitoring data. The preprocessing stage addresses environmental noise, sensor drift, and outliers in the raw data, and operates in steps in the order of noise removal, drift correction, and handling of outliers and missing values. The specific process is as follows: The first step is to filter out high-frequency noise.

[0054] Adaptive filtering algorithms, such as wavelet threshold denoising, are employed. Continuous raw monitoring data (e.g., vibration acceleration, strain signals) are first decomposed into wavelet coefficients of different frequencies, with high-frequency coefficients primarily corresponding to noise components such as electromagnetic interference and sensor thermal noise. A threshold is dynamically set based on noise intensity, and high-frequency coefficients exceeding the threshold are truncated or reduced. The signal is then reconstructed through inverse wavelet transform, thereby filtering out high-frequency noise and retaining the effective low-frequency signal reflecting the true state of the structure. For example, after processing strain data containing electromagnetic interference, spikes (high-frequency noise) in the signal fluctuations are smoothed out, and the curve more closely approximates the actual stress variation trend of the structure.

[0055] When collecting data, bridge vibration acceleration sensors are susceptible to electromagnetic interference from motors in the surrounding environment and thermal noise from the sensor itself, resulting in irregular spikes in the raw data curve (such as high-frequency jitter mixed in with normal vibration waveforms). When using the wavelet thresholding denoising algorithm, the vibration data is first decomposed into wavelet coefficients at five frequency levels, with the high-frequency coefficients at levels 4 and 5 mainly corresponding to electromagnetic interference (frequency > 100Hz). A threshold of 0.02g (g is the unit of gravitational acceleration) is set based on the noise intensity. High-frequency coefficients exceeding this threshold are then contracted, and the signal is reconstructed through inverse wavelet transform. After processing, high-frequency spikes in the original data are smoothly removed, and the curve retains only the low-frequency vibration characteristics of the bridge under vehicle load (frequency 5–20Hz), more realistically reflecting the actual vibration state of the beam.

[0056] The second step is to correct the zero-point drift of the sensor.

[0057] Collect historical baseline data of the sensor during the initial installation or periodic calibration, i.e., the output value when the structure is in a stable state; and establish a baseline model (such as the initial zero-point value and the normal drift curve over time). Compare the real-time acquired data with the baseline model to calculate the deviation between the current output value and the baseline value; if the deviation is within the preset normal range, it is determined to be normal fluctuation; if the deviation continues to exceed the range, the data is compensated using the linear correction formula "real-time value = measured value - drift deviation" to bring the sensor output back to the true zero-point level and eliminate drift errors caused by long-term use.

[0058] During the initial installation phase, the baseline value of the strain sensors at the bridge bearings, when the structure is unloaded, is generally stable at 2. 。 "1" is the unit of microstrain, a physical quantity that measures the degree of deformation of a material. Equal to one part per million strain (i.e. ).

[0059] In bridge monitoring, strain reflects the magnitude of deformation of the bridge structure under external forces. For example, when a bridge is subjected to vehicle loads, the beam may undergo minute tensile or compressive deformations, which are represented by micro-strain. "Quantifying such minute deformations is very intuitive. For example, the strain at a support is typically in the range of..." Up to dozens Fluctuating within the range of "", if the value suddenly and significantly exceeds the normal range (e.g., from 2) It surged to 15 If the value is 0, it may indicate that the structure is under abnormal stress or has potential damage.

[0060] After 6 months of use, the sensor will exhibit zero-point drift due to temperature aging, and the output value will remain stable at 7 under no-load conditions. (Offset from baseline 5) This leads to an overall higher level of monitoring data.

[0061] A linear baseline model (zero point value =) is established by retrieving baseline data from the initial installation phase. ( / month × usage time), calculate the current drift deviation as: .

[0062] The corrected formula is used: "Real-time strain value = Measured value - 4.94". "Compensate the data, for example, if the measured value at a certain moment is 15." The corrected actual value is 10.06. This eliminates systematic errors caused by drift, allowing the data to return to its true level.

[0063] The third step is to handle outliers and missing values.

[0064] Firstly, through 3 Outliers are identified using criteria (based on data mean and standard deviation) or box plot methods: When the difference between a data point and its adjacent data exceeds three times the standard deviation, or exceeds 1.5 times the interquartile range, it is marked as an outlier, such as a jump or invalidity. For marked outliers, if they are transient pulse interference (such as signal transmission errors), the point is directly removed; if they are short-term continuous outliers (such as short-term sensor failures), they are replaced with adjacent normal data using linear interpolation. For missing data (such as data gaps caused by transmission interruptions), if the missing duration is short (e.g., within 10 seconds), linear interpolation is used (calculating the median value based on the values ​​before and after the missing point); if the missing period is long but the data trend is stable, polynomial interpolation is used to fit the missing segment to ensure the continuity of the data sequence. For example, if a displacement sensor loses 5 minutes of data due to a communication interruption, the missing data for that period can be interpolated based on the linear growth trend of the 30 minutes before the loss to maintain the integrity of the sequence.

[0065] For example, during heavy rain, bridge mid-span displacement sensors may exhibit abnormal values ​​due to signal transmission interference: at a certain moment, the data suddenly jumps from the normal 3mm to 18mm (far exceeding 3). The threshold is 5mm), and the data is blank for the next 10 seconds (transmission interruption). For a jump value of 18mm, it is judged as instantaneous pulse interference and the point is directly removed; for the missing data of 10 seconds, linear interpolation is used: the data before the missing moment is 3mm, and the data after the missing moment is 3.2mm. The filling value is calculated by the formula "missing point value = 3mm + (3.2mm - 3mm) × (t - missing start time) / 10 seconds" to make the displacement curve continuous and avoid the impact of data breakage on subsequent trend analysis.

[0066] Through the above three steps, noise and drift errors in the original data are effectively eliminated, and outliers and missing values ​​are reasonably corrected, ultimately forming a continuous and clean dataset, providing a reliable foundation for subsequent feature extraction.

[0067] Key information reflecting the structural state of the bridge is extracted from the clean data to obtain key frame data.

[0068] The process of extracting key information reflecting the structural state of the bridge from clean data and obtaining key frame data also includes: Calculate time-domain statistics using a sliding window or a fixed time interval; For the preprocessed clean data, time-domain statistics such as mean, peak value, variance, and kurtosis are calculated at fixed time intervals (e.g., 1 minute / 10 minutes) or sliding windows (the window size is dynamically adjusted according to data characteristics, such as shrinking the window to 30 seconds when the vibration is severe) to capture the amplitude and pattern of the dynamic changes in the structure and screen out abnormal periods as preliminary keyframe candidates.

[0069] For clean strain data at the mid-span of the bridge, statistics were calculated at fixed time intervals of 1 minute. Under normal conditions, the peak strain value per time interval remained stable within ±30. Variance ≤ 5 When the peak value suddenly increases to ±80 during a certain period The variance reached 25. The value was significantly higher than the historical average by 3 times, and this period was marked as the "high strain anomaly time domain frame", indicating that the structure was under abnormal stress.

[0070] The cleanliness data in the time domain is converted into a frequency domain signal using a Fourier transform; For the abnormal time-domain frame data selected in the first step, the time-domain signal is converted into a frequency-domain signal through Fourier transform to identify spectral characteristics such as natural frequency and resonant frequency. Natural frequency is an inherent property of the structure, while resonant frequency is related to external excitation. Abnormal changes in both can reflect the state of the structure.

[0071] Fourier transform was performed on the vibration acceleration data of the "high strain anomaly time domain frame" to obtain the spectrum. Under normal conditions, the bridge's natural frequency is 8Hz with no obvious resonance peak; however, in the spectrum of this period, the amplitude of the 8Hz main frequency decreased by 20%, and a new resonance peak appeared at 5Hz (matching the frequency of heavy vehicle traffic at that time). This spectral feature was marked as a "frequency shift frequency domain frame," indicating that the structural stiffness may have decreased or there is a risk of resonance.

[0072] Furthermore, performing a Fourier transform on the vibration acceleration data of the "high strain anomaly time domain frame" converts the "time-acceleration" signal in the time domain into a "frequency-amplitude" signal in the frequency domain, thereby identifying the frequency characteristics of structural vibration. The specific process is as follows: From the continuously monitored vibration acceleration clean data, the time period corresponding to the "high strain anomaly time domain frame" (e.g., 9:15-9:16) is extracted. The data in this time period is a discrete time domain signal. Taking a sampling frequency of 1000Hz as an example, there are 60,000 data points within 1 minute. Each point records the acceleration value (unit: g) at the corresponding moment, forming a "time (t) - acceleration (a)" sequence (a=0.02g at t=0.001s, a=0.03g at t=0.002s, etc.).

[0073] The extracted data is detrended to ensure that the signal contains only vibrational components and to avoid the trend term interfering with frequency analysis.

[0074] Time-domain data is processed by Fast Fourier Transform (FFT, an efficient algorithm for calculating Fourier transform). Vibration acceleration signals are essentially superimposed simple harmonic vibrations of different frequencies. The Fourier transform decomposes them into the amplitude and phase of each frequency component through mathematical operations, resulting in a frequency domain spectrum of "frequency (f) - amplitude (A)".

[0075] Specifically, zero-padding is performed on 60,000 time-domain data points to make the data length an integer power of 2, such as 65,536 points, thereby improving the efficiency of FFT calculation. The FFT algorithm is called to calculate the complex number result corresponding to each frequency point; Take the modulus of the complex number and normalize it using the formula "amplitude = 2 × modulus / data length" to obtain the vibration amplitude (unit: g) with a clear physical meaning.

[0076] The frequency domain spectrum of the FFT output is analyzed to extract frequency features related to the structural state. The effective frequency of bridge vibration is usually between 0-50Hz, so the analysis focuses on this range and ignores high-frequency noise.

[0077] The frequency point with the highest amplitude in the spectrum is the natural frequency of the structure. In the "high strain anomaly time domain frame", if the amplitude at 8 Hz decreases by 20% compared to the normal state, it indicates that the structural stiffness may have decreased.

[0078] If a new peak appears in the spectrum (e.g., at 5Hz), and this frequency matches the external excitation (e.g., the vibration frequency of a heavy vehicle passing by at that time, which is 5Hz), it indicates that the structure resonates at that frequency, which may exacerbate structural damage.

[0079] By performing a Fourier transform on the vibration acceleration data of the "9:15-9:16 high strain anomaly time domain frame", the final frequency domain spectrum is clearly displayed. The amplitude of the original 8Hz natural frequency decreased, indicating abnormal structural stiffness; The newly added 5Hz resonance peak corresponds to the excitation of heavy vehicles, which explains the cause of the high strain.

[0080] These frequency domain features provide key information for subsequent modal parameter extraction and model calibration, and also make the labeling of "frequency offset frequency domain frames" more targeted.

[0081] By combining the dynamic model of the bridge structure, a deep correlation analysis of the time domain and frequency domain characteristics is conducted to extract modal frequencies and mode shape parameters.

[0082] By calling the bridge finite element dynamics model and correlating time-domain statistics with frequency-domain spectral characteristics, core parameters such as modal frequencies and mode shapes can be extracted using algorithms such as the random subspace method and peak picking method. Modal frequencies directly reflect structural stiffness, and mode shape distortion can locate local damage.

[0083] Based on the finite element model of the bridge, the analysis of the "frequency offset frequency domain frame" revealed that 8Hz corresponds to the second bending mode frequency of the model, and the 5Hz resonance peak corresponds to the vehicle load excitation frequency.

[0084] Further calculation of the mode shape curves revealed that under normal conditions, the second mode shape is symmetrically distributed across the entire bridge, while the current mode shape exhibits a sudden increase in local displacement at K3+200. This feature was extracted as the "mode shape distortion frame," and the abnormal location was preliminarily identified.

[0085] The keyframe data is obtained by calibrating the model parameters by comparing the natural frequencies identified in the frequency domain with the theoretical modal frequencies of the dynamic model.

[0086] The measured natural frequencies extracted from the frequency domain are compared with the theoretical values ​​calculated by the dynamic model. If there is a deviation (such as the measured value being lower than the theoretical value), the model parameters are adjusted in reverse until the model calculation values ​​match the measured values. The calibrated model can output more accurate modal parameters, and finally form key frame data that can be used for defect localization.

[0087] For example, the theoretical natural frequency of the original dynamic model is 8.5 Hz, while the measured value is 8 Hz.

[0088] By calibrating the model parameters, the elastic modulus of the concrete in the span beam was increased from... Down to The model's calculated frequency was corrected to 8.05Hz, with an error of less than 1% compared to the measured value. After calibration, the displacement amplitude calculation at the mode distortion location (K3+200) was more accurate. Ultimately, the data "at K3+200, the modal frequency is 8Hz, and the mode amplitude increases abnormally by 40%" was identified as key frame data, indicating that there may be local damage at this location.

[0089] Through the above steps, the final keyframe data includes "time (9:15-9:16), location (K3+200), and eigenvalues ​​(strain peak 80)". (Natural frequency 8Hz, mode distortion 40%).

[0090] Enhanced frame data is obtained by integrating and fusing keyframe data and strengthening effective information and weakening interference through spatiotemporal feature analysis. For data of the same type collected by multiple sensors (such as strain keyframes of different beam segments) and data of different types at the same location (such as displacement and acceleration keyframes of a support), the Kalman filter algorithm is used for fusion.

[0091] For example, all three strain sensors acquired keyframes indicating "peak strain exceeding the threshold" at K3+200, with values ​​of 80. 85 78 Kalman filtering combines the error covariance of each sensor (e.g., sensor A error ±5). Sensor B±3 The optimal fusion value was calculated to be 81. This reduces the measurement bias of a single sensor.

[0092] Data integration of different types: strain keyframe "strain peak 81" at K3+200 The state equation is established by using Kalman filtering with the acceleration keyframe "vibration acceleration 0.15g". The resulting composite keyframe with abnormal force at that location is obtained by fusing the data with the acceleration keyframe "vibration acceleration 0.15g". This avoids the one-sidedness of a single data type.

[0093] Kalman filtering algorithm is used to integrate key frame data of the same type of data collected by multiple sensors and different types of data at the same location; Analyze the spatial correlation of different keyframe data to determine whether it conforms to the mechanical transmission law of bridge structure; The integrated keyframe data is mapped according to the physical location of the bridge to analyze whether its spatial distribution conforms to the laws of structural mechanics. For example, if the strain keyframe of the middle section of the third span (K3+200) of a bridge shows a "strain increase", it is necessary to verify whether there are correlation characteristics between adjacent positions (K3+190, K3+210). Under normal circumstances, the strain of the beam should change in a gradient along the load transfer direction (e.g., strain 81 at K3+200). , K3+190 is about 60 , K3+210 is about 55 If there is no response from adjacent points or a sudden change in value (e.g., K3+190 only responds to 10), If the condition is not met, then the keyframe may be affected by sensor malfunction and does not conform to the laws of mechanical transmission.

[0094] Determine whether the trend of keyframe data changes over time is reasonable, weaken or remove false features, and enhance effective information to obtain enhanced frame data; For example, the strain keyframe at K3+200 changes from 81 within 1 hour. Rise to 95 If the vehicle load keyframe for the corresponding time period shows "continuous passage of heavy trucks", then the trend is reasonable (increased load leads to increased strain); if there is no load change in the same time period but the strain rises or falls sharply, then it is judged as "false feature" (such as the sensor being affected by electromagnetic interference).

[0095] For keyframes that conform to the time trend of "continuous strain growth", their influence coefficient is amplified and strengthened in subsequent analysis by assigning higher weights; for spurious features such as "sudden strain change without load association", their weights are reduced or they are directly removed and weakened, ultimately forming enhanced frame data of "continuous stress anomaly at K3+200".

[0096] The target data is obtained by performing error compensation and high-precision fitting on the enhanced frame data and reducing the influence of random noise.

[0097] Although the enhanced frame data has weakened some interference through integration and fusion, there may still be systematic errors such as sensor temperature drift and installation deviation, which require targeted compensation. For data such as strain and displacement that are significantly affected by temperature, an ambient temperature is collected using a built-in temperature sensor. This data, combined with a preset "temperature-error" calibration curve, is used to correct the strain values ​​in the enhanced frame. For example, the strain at K3+200 in the enhanced frame is 81... If the temperature rises by 10°C compared to the calibration time, the compensated value is 81. -(10℃×2 / ℃) = 61 This eliminates systematic biases caused by temperature.

[0098] For measurement errors caused by sensor installation angle deviations, the acceleration values ​​in the enhanced frame are corrected to the true acceleration along the bridge axis using a three-dimensional coordinate transformation formula, ensuring the accuracy of the physical meaning of the data.

[0099] Enhanced frame data may exhibit discreteness due to sensor sampling intervals or local fluctuations, requiring fitting algorithms to reconstruct the continuous trend of the structural state. Polynomial fitting: for strain-enhanced frames at K3+200 (e.g., time series: t1=81) t2=85 t3=90 The least squares method was used to perform a quadratic polynomial fitting (y = at² + bt + c) to obtain a continuous strain-time curve. The fitted curve can smooth out local fluctuations, such as eliminating the 85° fluctuation caused by instantaneous vibration at time t2. A jump more accurately reflects the cumulative growth trend of response over time, such as from 81 linearly increased to 90 .

[0100] Segmented Fitting: To address the differences in mechanical properties across different structural segments of the bridge, the enhanced frame data is fitted in segments according to the structure. For example, the displacement enhanced frame data at the supports exhibits non-linear changes, so an exponential function is used for fitting; while the beam displacement shows linear growth, so a linear fitting is used, ensuring that the data trends at different locations match the structural characteristics.

[0101] Although the enhanced frame weakens most of the interference, random noise may still remain, requiring final purification through filtering algorithms: Generally, advanced algorithms such as Kalman filtering or wavelet thresholding are used to perform secondary filtering on the fitted continuous curve. For example, for the fitted strain curve, a noise variance threshold of 3 is set. For fluctuations in the curve that exceed the threshold, such as a sudden increase to 65 at a certain moment... The deviation from the fitted value is 61. Reaching 4 Smoothing is performed to preserve the true trend changes, such as 61. up to 63 The slow growth eliminates random noise.

[0102] For dynamic data such as vibration acceleration, the data is converted back to the frequency domain through inverse Fourier transform. High-frequency noise components (such as non-structural vibrations >50Hz) are truncated, and then converted back to the time domain to obtain purer acceleration enhancement frame data (such as 0.149g is corrected to 0.145g to remove high-frequency noise interference).

[0103] After error compensation, high-precision fitting, and reduction of random noise, the obtained target data possesses the characteristics of "high precision, low noise, and continuous trend." For example, the target data at K3+200 is "strain 61." "(After temperature compensation), acceleration 0.145g (after installation deviation compensation + filtering), displacement 2.3mm (after piecewise fitting)", can be directly used as the final target data for bridge health assessment and defect location.

[0104] The target data is compressed.

[0105] The compression process for the target data includes: The sliding window technique is activated to dynamically scan the target data according to the set initial window size; First, the sliding window technique is activated, and the initial window size is set based on the type and characteristics of the target data. Then, the window moves gradually along the target data sequence, performing a dynamic scan. During the scan, if a recurring pattern is detected with a high frequency, the window is appropriately enlarged to capture more redundant information; if the data sequence changes drastically and the recurring pattern is few, the window is reduced to improve scanning efficiency.

[0106] Within each dynamically selected window, the target data is scanned byte-by-byte using the LZ77 algorithm. This algorithm identifies repetition patterns and redundant information by searching for previous data blocks within the window that are identical to the current data block. Then, the DEFLATE algorithm, combined with the LZ77 processing results, encodes and transforms these repetition patterns and redundant information. For long, repetitive data sequences, shorter codes are used to replace them, thus completing the compression process and obtaining the compressed target data. This process strictly adheres to the principle of lossless compression, ensuring that the compressed data can be completely restored to the original target data while significantly reducing the data size.

[0107] Establish a tiered transmission mechanism.

[0108] For critical data, such as real-time vibration peak data and strain exceeding threshold data that directly reflect the safety of bridge structures, 5G slicing technology is used to allocate dedicated network resource channels for them. By logically dividing the 5G network, critical data has independent bandwidth, latency, and reliability guarantees during transmission, free from interference from other data transmissions, enabling real-time output and ensuring that critical information can quickly reach receiving terminals such as the bridge monitoring center system.

[0109] For routine data, such as temperature data and displacement data within normal ranges, a batch caching approach is used. The cached routine data is packaged together at set time intervals (e.g., every 5-10 minutes) and then transmitted. This method reduces the interaction overhead caused by frequent transmissions and improves transmission efficiency.

[0110] Establish a dynamic bandwidth adaptation mechanism.

[0111] Establish a mechanism for real-time monitoring of network bandwidth status, and continuously obtain the current bandwidth value through network monitoring tools.

[0112] When insufficient bandwidth is detected, the data compression rate is adjusted according to the actual situation. If the bandwidth is slightly lower than the normal transmission requirements, the compression level is appropriately increased to further reduce the data volume to adapt to the bandwidth limitation; if the bandwidth is severely insufficient, the core feature priority transmission mode is activated, transmitting only the most critical core feature information in the data. Once the bandwidth is restored, the complete data is transmitted to ensure the continuity and effectiveness of data transmission in complex network environments.

[0113] The data is packaged and output according to the collection frequency and type.

[0114] The compressed target data is encapsulated according to the data collection frequency and type. For high-frequency collected critical data, the complete compressed data within that time period and its corresponding timestamp are encapsulated into a data packet at fixed time intervals (e.g., 1 second). For regular low-frequency data, it is aggregated according to monitoring points or data types, packaged into batch data packets, and each data packet is added with a timestamp and index information.

[0115] Finally, these data packets are output to the bridge monitoring center system, data storage server, and related analysis platform. Timestamps and data packet indexes ensure the temporal integrity of continuous data, meeting the requirement for complete and real-time output of massive amounts of dynamic monitoring data.

[0116] Specifically, the processing described in this embodiment yields high-precision and continuous target data. This high precision allows the data to accurately reflect subtle changes in the bridge structure, such as precisely capturing micron-level strain or millimeter-level displacement changes in bridge components. This is crucial for the early detection of potential bridge defects. With this high-precision data, staff can detect anomalies at the nascent stage of defects, gaining time to take timely maintenance measures, preventing small problems from escalating into major malfunctions, and significantly reducing the risk of bridge safety accidents.

[0117] The dynamic nature of the data allows for real-time monitoring of the bridge's condition under various loads and environmental conditions. For example, the data can be updated in real time under dynamic influences such as vehicle traffic and wind changes, enabling monitoring personnel to promptly grasp the bridge's dynamic response. Based on this, the bridge's load-bearing capacity and safety reserves can be dynamically assessed. In the event of emergencies such as overloaded vehicles, it is possible to quickly determine whether the bridge is in a safe state, providing timely information for traffic control and emergency response.

[0118] Meanwhile, the combination of high precision and dynamic data provides reliable data support for the long-term health assessment of bridges. Through the accumulation of high-precision dynamic data over a long period, the degradation trend of bridge structural performance can be analyzed, its remaining service life predicted, and a scientifically sound maintenance plan developed. This enables refined management of bridges, ensuring bridge safety while effectively saving maintenance costs.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A high-precision dynamic data acquisition instrument, characterized in that, include: The signal acquisition module uses a basic sampling frequency to sample dynamic bridge monitoring data and adjusts the sampling frequency in real time according to a unified threshold corresponding to the bridge monitoring data. The signal processing module acquires the bridge monitoring data collected in real time by the signal acquisition module, and processes the bridge monitoring data through multi-level processing to obtain the target data. The data output module processes the massive amount of target data through a real-time complete output guarantee mechanism and divides it into several data packets at expected time intervals, and then outputs the data packets to the bridge monitoring center system, data storage server and / or related analysis platform; The bridge monitoring data includes structural response data and external load data.

2. The high-precision dynamic acquisition instrument according to claim 1, characterized in that, The method of sampling dynamic bridge monitoring data using a basic sampling frequency and adjusting the sampling frequency in real time based on a unified threshold corresponding to the bridge monitoring data includes: Obtain the original data x; The original data is normalized to obtain a uniform threshold. Sampling is performed according to the sampling frequency corresponding to the unified threshold; The unified threshold includes a first threshold T1 and a second threshold T2, where T1 < T2; The sampling frequency includes a basic sampling frequency, a first sampling frequency, and a second sampling frequency.

3. A high-precision dynamic data acquisition instrument according to claim 2, characterized in that, The normalization process includes: The original data x is normalized according to the formula x′=(x-x_min) / (x_max-x_min) to obtain normalized data x′; The normalized data x′ is matched with the unified threshold. If the match is successful, sampling is performed according to the corresponding sampling frequency. If the match fails, sampling is performed according to the basic sampling frequency. Wherein, a successful match is achieved when x′ is triggered at least N consecutively at the corresponding sampling frequency, and N is a positive integer; Where x_min is the minimum theoretical value of this type of data, and x_max is the maximum design value of this type of data.

4. A high-precision dynamic data acquisition instrument according to claim 3, characterized in that, The normalized data x′ is matched with the unified threshold. If the match is successful, sampling is performed according to the corresponding sampling frequency, including: When x′<T0, the signal acquisition module is controlled to sample at the basic sampling frequency f1; When T0≤x′<T1, the signal acquisition module is controlled to sample at the second sampling frequency f2, where f2>f1; When x′≥T1, the signal acquisition module is controlled to acquire data at a third frequency f3, where f3>f2.

5. A high-precision dynamic data acquisition instrument according to claim 1, characterized in that, The multi-level processing includes: Clean data is obtained by preprocessing real-time bridge monitoring data; Key information reflecting the structural state of the bridge is extracted from clean data to obtain key frame data; Enhanced frame data is obtained by integrating and fusing keyframe data and strengthening effective information and weakening interference through spatiotemporal feature analysis. The target data is obtained by performing error compensation and high-precision fitting on the enhanced frame data and reducing the influence of random noise.

6. A high-precision dynamic data acquisition instrument according to claim 5, characterized in that, The process of extracting key information reflecting the structural state of the bridge from clean data and obtaining key frame data includes: Calculate time-domain statistics using a sliding window or a fixed time interval; The cleanliness data in the time domain is converted into a frequency domain signal using a Fourier transform; By combining the dynamic model of the bridge structure, a deep correlation analysis of the time domain and frequency domain characteristics is performed to extract modal frequencies and mode shape parameters. The keyframe data is obtained by calibrating the model parameters by comparing the natural frequencies identified in the frequency domain with the theoretical modal frequencies of the dynamic model.

7. A high-precision dynamic data acquisition instrument according to claim 5, characterized in that, The process of integrating and fusing keyframe data, enhancing effective information, and weakening interference through spatiotemporal feature analysis to obtain enhanced frame data includes: Kalman filtering algorithm is used to integrate key frame data of the same type of data collected by multiple sensors and different types of data at the same location; Analyze the spatial correlation of different keyframe data to determine whether it conforms to the mechanical transmission law of bridge structure; Determine whether the trend of keyframe data changes over time is reasonable, weaken or remove false features, and enhance effective information to obtain enhanced frame data.

8. A high-precision dynamic data acquisition instrument according to claim 1, characterized in that, The real-time complete output guarantee mechanism includes: Compress the target data; Establish a tiered transmission mechanism; Establish a dynamic bandwidth adaptation mechanism; The data is packaged and output according to the collection frequency and type.

9. A high-precision dynamic data acquisition instrument according to claim 8, characterized in that, The compression process for the target data includes: The sliding window technique is activated to dynamically scan the target data according to the set initial window size; Within each dynamically selected window range, the target data is scanned byte by byte using the LZ77 algorithm to identify repeating patterns and redundant information in the target data; By using the DEFLATE algorithm and combining the processing results of LZ77, these repetitive patterns and redundant information are encoded and converted to obtain compressed target data.

10. A high-precision dynamic data acquisition instrument according to claim 8, characterized in that, The aforementioned hierarchical transmission mechanism includes: For critical data, dedicated network resource channels are allocated using 5G slicing technology to achieve real-time output; For regular data, it is cached in batches and then packaged and transmitted in batches according to the set time intervals.

Images