A prestressed tensioning data management system and method for highway bridge construction projects

Abstract

The present invention relates to the technical field of highway bridge construction projects, and discloses a prestressed tensioning data management system and method for highway bridge construction projects. The system includes: a multidimensional data acquisition module for synchronously collecting tensioning force, steel strand elongation, acoustic emission signals, and equipment posture data during the tensioning process; a dynamic health baseline generation module for generating a multidimensional dynamic health baseline model based on design parameters, material properties, ambient temperature, and historical excellent data; a process feature fingerprint extraction module for extracting feature vectors representing the tensioning process state from multidimensional data as process feature fingerprints; a real-time diagnosis and health quantification module; and a causal knowledge graph and adaptive control module. Through the causal knowledge graph and adaptive control module, intelligent identification of fault modes and real-time decision-making are achieved, achieving the effect of dynamically adjusting intervention strategies. This effectively reduces the probability of major hidden dangers and ensures project safety.

Description

Technical Field

[0001] The present invention relates to the technical field of highway bridge construction engineering, in particular to a prestressed tensioning data management system and method for highway bridge construction engineering. Background Art

[0002] In highway bridge construction, prestressing is a critical step in ensuring structural safety and durability. Currently, many projects still rely on traditional data monitoring methods, which have several significant shortcomings.

[0003] First, traditional methods often rely on manual inspection. While manual inspection can identify problems, it has a slow response time. Once an anomaly occurs, it often takes a long time to address it promptly. This can lead to small issues becoming major risks, threatening project safety. Therefore, rapid response and automated monitoring are crucial.

[0004] Secondly, the static health baseline models used in existing technologies lack flexibility in adapting to environmental changes. For example, changes in temperature or humidity can directly affect material performance. If the baseline model cannot be updated in a timely manner, monitoring results may be distorted, leading to erroneous decisions. This limitation means that the actual condition of the bridge may deviate significantly from the monitoring results, thus affecting safety assessments.

[0005] Furthermore, traditional data collection methods often rely on a single sensor, which is susceptible to signal interference. Readings from multiple sensors may not be perfectly synchronized, which not only affects data accuracy but can also lead to erroneous analysis. This situation makes it difficult for project teams to obtain comprehensive and reliable information at critical moments. Summary of the Invention

[0006] In response to the shortcomings of the existing technology, the present invention provides a prestressed tensioning data management system and method for highway bridge construction projects, which solves the problems of untimely response, poor adaptability of baseline models and insufficient data acquisition accuracy in existing prestressed tensioning monitoring.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A prestressed tensioning data management system for highway bridge construction projects, comprising:

[0008] Multi-dimensional data acquisition module, used to synchronously collect tensioning force, steel strand elongation value, acoustic emission signal and equipment posture data during the tensioning process to generate multi-dimensional data;

[0009] Dynamic health baseline generation module, used to generate a multi-dimensional dynamic health baseline model based on design parameters, material properties, ambient temperature and historical good data;

[0010] A process feature fingerprint extraction module is used to extract a feature vector representing the state of the tensioning process from multidimensional data as a process feature fingerprint;

[0011] Real-time diagnosis and health quantification module, used to compare process characteristic fingerprints with multi-dimensional dynamic health baseline models and calculate the tensile health index;

[0012] The causal knowledge graph and the adaptive control module store the mapping relationship between fault mode, feature fingerprint and intervention strategy, and generate and send control instructions to the front-end execution mechanism according to the abnormality of the tension health index or the specific pattern of the process feature fingerprint.

[0013] Preferably, the multidimensional data acquisition module generates multidimensional data by the following steps:

[0014] Add a unified synchronization timestamp to the data streams from tension force, steel strand elongation, acoustic emission signals processed by the acoustic emission signal preprocessing unit, and equipment attitude sensors;

[0015] Aligning the data streams stamped with the synchronization timestamps according to a preset sampling frequency;

[0016] The aligned data streams are encapsulated into time series data frames with a unified data structure and output as the multidimensional data.

[0017] Preferably, the dynamic health baseline generation module includes:

[0018] A theoretical model building unit, configured to calculate a force-displacement reference curve based on the pore geometry information in the design parameters and the elastic modulus in the material properties;

[0019] A statistical model building unit, used to analyze historical excellent data, extract and establish a fluctuation range model of the acoustic emission signal energy;

[0020] The dynamic correction unit is used to perform linear compensation correction on the force-displacement reference curve according to the ambient temperature collected in real time.

[0021] Preferably, the statistical model building unit establishes the fluctuation range model of the acoustic emission signal energy through the following steps:

[0022] Normalize the tension force or tension time axis in multiple sets of historical high-quality data to form a unified analysis dimension;

[0023] Divide the normalized analysis dimension into multiple continuous intervals;

[0024] Count the acoustic emission signal energy values ​​corresponding to all historical good data falling within each interval, and calculate their statistical distribution characteristics;

[0025] A confidence interval is determined for each interval according to the statistical distribution characteristics, and the set of confidence intervals of all intervals together constitutes the fluctuation range model of the acoustic emission signal energy.

[0026] Preferably, the process feature fingerprint extraction module includes:

[0027] A signal synchronization and preprocessing unit, configured to perform time stamp alignment on the data streams of the tension force, the steel strand elongation value, and the acoustic emission signal;

[0028] Differentiation and integration operation unit, used to calculate the real-time slope, stress relaxation rate and energy integral value of the data after time stamp alignment;

[0029] The feature vector synthesis unit is used to combine the calculation results of the differential and integral operation units into a multi-dimensional process feature fingerprint at each time stamp.

[0030] Preferably, the differential and integral operation unit calculates the real-time slope of the force-displacement curve by the following steps:

[0031] Set a sliding calculation window containing a preset number of data points;

[0032] Continuously moving the sliding calculation window along the tension force and steel strand elongation value data sequence aligned with the timestamps;

[0033] At each position of the sliding calculation window, applying the least squares method to the set of data points contained in the window to perform linear fitting to obtain a fitting straight line;

[0034] The slope value of the fitting straight line is used as the real-time slope of the force-displacement curve corresponding to the center point of the sliding calculation window.

[0035] Preferably, the real-time diagnosis and health quantification module further includes:

[0036] A geometric deviation calculation unit, used to calculate the degree of deviation between the force-displacement trajectory in the process characteristic fingerprint and the corresponding reference trajectory in the multidimensional dynamic health baseline model;

[0037] an acoustic anomaly calculation unit, configured to calculate a degree of deviation between an acoustic emission signal feature in a process characteristic fingerprint and a corresponding acoustic benchmark in the multidimensional dynamic health baseline model;

[0038] A nonlinear fusion unit is used to perform weighted fusion on the geometric deviation and the acoustic anomaly to generate a tensile health index.

[0039] Preferably, the real-time diagnosis and health quantification module calculates the tension health index through the following steps:

[0040] At multiple preset tension sampling points, the Euclidean distances between the real-time force-displacement trajectory coordinates and the corresponding coordinates in the benchmark trajectory of the multidimensional dynamic health baseline model are calculated, and the set of Euclidean distances is used as the geometric deviation.

[0041] Calculating the standard deviation multiple of the real-time acoustic emission signal energy integral value and the acoustic baseline energy in the multidimensional dynamic health baseline model, and using the standard deviation multiple as the acoustic abnormality degree;

[0042] A nonlinear function with a preset weight coefficient is applied to calculate the geometric deviation and acoustic anomaly to obtain a standardized tensile health index.

[0043] Preferably, the mapping relationship between the causal knowledge graph and the adaptive control module includes:

[0044] The excessive friction failure mode of the pores is mapped to a characteristic fingerprint with a continuously low slope of the force-displacement curve and a steadily increasing acoustic emission energy, and is then linked to an intervention strategy of slight retreat and then re-advance.

[0045] The strand breakage failure mode is mapped to the characteristic fingerprint of an instantaneous drop in tensioning force and the appearance of high-energy pulses in the acoustic emission signal, and is then linked to an intervention strategy of immediately suspending tensioning.

[0046] A method for managing prestressed tensioning data in a highway bridge construction project comprises the following steps:

[0047] Generate a multi-dimensional dynamic health baseline model including a force-displacement benchmark curve and an acoustic emission signal energy fluctuation range model;

[0048] During the tensioning process, the tensioning force, steel strand elongation, acoustic emission signal and equipment posture data are collected synchronously;

[0049] Extract process feature fingerprints including the slope of the force-displacement curve and the energy characteristics of the acoustic emission signal from the collected data in real time;

[0050] Comparing the process characteristic fingerprint with the multi-dimensional dynamic health baseline model to calculate a quantitative tensile health index;

[0051] determining whether the tension health index is lower than a preset threshold or whether the process characteristic fingerprint matches a preset failure mode;

[0052] If so, the corresponding adaptive intervention strategy is generated and executed based on the predefined causal knowledge graph;

[0053] All data of the tensioning process, extracted feature fingerprints, health index and intervention records are structured and archived for subsequent optimization of the multidimensional dynamic health baseline model and the causal knowledge graph.

[0054] The present invention provides a prestressed tensioning data management system and method for highway bridge construction projects. It has the following beneficial effects:

[0055] 1. This invention utilizes a causal knowledge graph and an adaptive control module to achieve intelligent identification of fault modes and real-time decision-making, enabling dynamic adjustment of intervention strategies. Compared to traditional manual detection methods, this solves the problem of delayed response to issues, effectively reducing the probability of major hidden dangers and ensuring project safety.

[0056] 2. This invention introduces a dynamic health baseline generation module, constructing a comprehensive health baseline model through theoretical, statistical, and real-time corrections. This model effectively compensates for real-time environmental factors. Compared to the static models used in existing technologies, this solves the problem of baseline models' poor adaptability to environmental changes, ensuring more accurate safety monitoring of the tensioning process and significantly reducing the risk of misjudgment due to environmental factors.

[0057] 3. This invention utilizes multidimensional data acquisition technology and ensures the timeliness of tension monitoring data through real-time synchronization, achieving high-precision, low-noise data acquisition. Compared to existing methods using a single sensor, this method solves the problems of signal interference and varying timeliness between different sensors during data acquisition, enabling accurate integration of multiple signals and enhancing the reliability of tension process monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a system framework diagram of the present invention;

[0059] Figure 2 This is a schematic diagram of a dynamic health baseline generation module of the present invention;

[0060] Figure 3 This is a schematic diagram of the process feature fingerprint extraction module of the present invention;

[0061] Figure 4 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0062] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0063] Please see the attached Figure 1 -Attached Figure 3 The embodiment of the present invention provides a prestressed tensioning data management system and method for highway bridge construction projects, including:

[0064] Multi-dimensional data acquisition module, used to synchronously collect tensioning force, steel strand elongation value, acoustic emission signal and equipment posture data during the tensioning process to generate multi-dimensional data;

[0065] Specifically, the multidimensional data acquisition module in this embodiment physically comprises multiple sensor assemblies, an acoustic emission signal preprocessing unit, and a data synchronization and packaging unit. The sensor assemblies, used to collect raw physical signals, illustratively include a pressure sensor connected to the hydraulic system oil circuit, a displacement sensor for measuring the elongation of the steel strand, an acoustic emission sensor positioned close to or adjacent to the anchoring area, and an equipment attitude sensor fixed to the tensioning jack.

[0066] The acoustic emission signal preprocessing unit is directly connected to the acoustic emission sensor and is used to perform hardware-level real-time filtering on the signal before it enters the subsequent processing unit.

[0067] The pre-processing unit is equipped with a hardware bandpass filter. Due to the large amount of low-frequency mechanical vibration noise and high-frequency electromagnetic interference at the tensioning construction site, the noise will interfere with the effective acoustic emission signal generated by the microscopic damage of the steel strand.

[0068] When the raw electrical signal captured by the acoustic emission sensor is fed into the preprocessing unit, mechanical vibration noise with frequencies below 20kHz and electromagnetic interference with frequencies above 1MHz are effectively filtered out, retaining only the signal within this frequency band. This approach improves the signal-to-noise ratio at the very beginning of data acquisition, providing a foundation for the accuracy of subsequent feature extraction.

[0069] After completing the above-mentioned sensor layout and signal preprocessing, the multi-dimensional data acquisition module integrates the independent physical signals into unified multi-dimensional data through the following steps.

[0070] First, a unified synchronization timestamp is added to each independent data stream from the pressure sensor, displacement sensor, acoustic emission signal processed by the acoustic emission signal preprocessing unit, and the equipment attitude sensor. The system is equipped with a high-precision clock source, which provides a synchronized clock signal for all data acquisition channels. Each channel will store the current The values ​​are appended to the sampled data to ensure consistency in the time base for all data points.

[0071] According to a preset system sampling frequency , align the data streams with the synchronization timestamp. Since the inherent response frequencies and data output rates of different sensors may be different, direct combination will cause the data points to be misaligned on the time axis. The alignment process aims to unify all data streams to a common sampling frequency through upsampling or downsampling algorithms. superior.

[0072] The aligned data streams are encapsulated into a time series data frame with a unified data structure and output as the multidimensional data. It is in At a sampling time point, the vector composed of all sensor data can be mathematically expressed as:

[0073] ;

[0074] Where;

[0075] For the Data frames at sampling moments;

[0076] is the tension data at that moment;

[0077] is the steel strand elongation value data at that moment;

[0078] is the amplitude data of the filtered acoustic emission signal at that moment;

[0079] They are the posture data in the three spatial axes output by the device posture sensor at that moment.

[0080] The multi-dimensional data acquisition module can convert sensor signals from different physical sources and with different characteristics into a multi-dimensional data stream that is strictly synchronized in time and highly regular in structure.

[0081] Dynamic health baseline generation module, used to generate a multi-dimensional dynamic health baseline model based on design parameters, material properties, ambient temperature and historical good data;

[0082] Specifically, the dynamic health baseline generation module in this embodiment logically comprises a theoretical model building unit, a statistical model building unit, and a dynamic correction unit. These three units work together to build a baseline from three perspectives: theoretical calculation, historical data statistics, and real-time environmental adaptation.

[0083] The theoretical model building unit is used to calculate the force-displacement reference curve under ideal conditions based on deterministic physical laws. This unit receives the input of the design parameters, specifically the duct geometry information, and the material properties, specifically the measured elastic modulus of the steel strand. and cross-sectional area .

[0084] The unit is calculated based on the prestress theory under different tension The theoretical elongation of the steel strand under the action of This calculation takes into account both the friction loss in the hole and the deformation and shrinkage loss of the anchor. For example, the calculation relationship can be expressed as:

[0085] ;

[0086] Where;

[0087] Indicates in At the sampling moment, the tension reaches Theoretical elongation value when

[0088] For the The tension value at the moment;

[0089] is the total length of the steel strand;

[0090] is the effective cross-sectional area of ​​the steel strand;

[0091] is the elastic modulus of the steel strand material;

[0092] The friction coefficient caused by the local deviation of the hole per meter;

[0093] Indicates the length of the straight line segment from the tensioning end to the current calculation point;

[0094] is the friction coefficient between the steel strand and the duct wall;

[0095] It represents the sum of the angles of the channel curve path from the tensioning end to the calculation point.

[0096] By calculating a series of different Corresponding , the unit constructs a complete force-displacement reference curve.

[0097] The statistical model building unit is used to build a model for the energy fluctuation range of the non-deterministic acoustic emission signal to represent the normal state. Since the acoustic emission signal is random and cannot be directly predicted by theoretical formulas, this unit achieves this by analyzing the historical good data.

[0098] This unit models the fluctuation range of the acoustic emission signal energy through the following steps: First, the tension force or tension time axis in multiple sets of historical high-quality data is normalized to form a unified analysis dimension ranging from 0 to 1. This eliminates differences in the total duration or total tension force of different tensioning tasks, making them comparable.

[0099] On the normalized analysis dimension, multiple continuous intervals are divided , the statistics fall into each interval The energy values ​​of the acoustic emission signals corresponding to all the historical good data in the interval are calculated, and their statistical distribution characteristics are calculated. For example, the energy values ​​of the acoustic emission signals corresponding to all the historical good data in the interval are calculated. The mean internal acoustic emission energy and standard deviation .

[0100] Finally, a confidence interval is determined for each interval based on the statistical distribution characteristics. The confidence interval can be set as:

[0101] ;

[0102] Where;

[0103] Indicates the The mean of an acoustic emission signal data set, usually the center of the data;

[0104] Indicates the The standard deviation of an acoustic emission signal data set is used to measure the degree of data fluctuation or the range of signal variation;

[0105] To determine the coefficients for the signal data set interval, we usually choose the appropriate value to determine the interval width.

[0106] The dynamic correction unit is used to compensate and correct the theoretical model according to the real-time environmental factors to improve the field adaptability of the baseline model. .

[0107] Considering that temperature changes will cause changes in the elastic modulus and length of the steel strand, this unit performs linear compensation correction on the force-displacement reference curve generated by the theoretical model construction unit. The corrected theoretical elongation value The calculation relationship is

[0108] ;

[0109] Where:

[0110] Indicates the Theoretical elongation value after temperature correction at the moment;

[0111] Indicates the The original theoretical elongation value at the moment;

[0112] Indicates the influence coefficient of elastic modulus of steel strand on temperature;

[0113] Indicates the current real-time temperature;

[0114] Indicates the reference temperature;

[0115] Indicates the thermal expansion coefficient of the steel strand;

[0116] Indicates the total length of the steel strand.

[0117] The Dynamic Health Baseline Generation Module outputs a complete multi-dimensional dynamic health baseline model. This model not only includes an accurate theoretical force-displacement baseline curve dynamically corrected for temperature, but also includes a reliable normal behavior range for acoustic emission signals based on extensive historical data statistics.

[0118] A process feature fingerprint extraction module is used to extract a feature vector representing the state of the tensioning process from multidimensional data as a process feature fingerprint;

[0119] Specifically, the process feature fingerprint extraction module in this embodiment converts the high-dimensional original time series data output by the multi-dimensional data acquisition module into a low-dimensional feature vector that can more sensitively reflect the internal state changes of the tensioning process, namely the process feature fingerprint.

[0120] The process fingerprint extraction module logically consists of a signal synchronization and preprocessing unit, a differential and integral operation unit, and a feature vector synthesis unit. These three units sequentially process the data stream to transform the raw data into feature vectors.

[0121] The signal synchronization and pre-processing unit receives data frames from the multi-dimensional data acquisition module and performs final timestamp alignment on the data streams required for subsequent operations. This ensures that the tension force and steel strand elongation data points used to calculate the real-time slope, as well as the acoustic emission signal data points, are strictly aligned in terms of timestamps, ensuring the accuracy of subsequent differential and integral operations.

[0122] Furthermore, the unit calculates the real-time slope of the force-displacement curve by the following steps: First, a set of data points is set up. Sliding calculation window . The value of determines the balance between the smoothness and sensitivity of the slope calculation. Tensile force after alignment along timestamps and the elongation value of steel strand The data sequence is moved continuously.

[0123] At each position in the sliding calculation window , for the window contained A collection of data points Apply the least squares method to perform linear fitting to obtain a fitting straight line ;

[0124] The slope value of the fitting line , as the center point of the sliding calculation window The corresponding real-time slope of the force-displacement curve. The specific calculation formula is:

[0125] ;

[0126] Where;

[0127] Indicates at a point in time Real-time slope of the force-displacement curve;

[0128] Indicates the number of data points in the sliding calculation window;

[0129] Indicates the first The steel strand elongation value of each data point;

[0130] Indicates the first The tensile force value corresponding to each data point;

[0131] Indicates that all the The data points are summed.

[0132] In addition, the differential and integral operation unit is also used to calculate the stress relaxation rate in the load holding stage. When the tension reaches the design value and starts to hold the load, the unit calculates the stress relaxation rate in the load holding stage by aligning the tension data with the timestamp. The first-order differential in time is performed to calculate the feature, and its discretization calculation can be expressed as;

[0133] ;

[0134] Where:

[0135] Indicates at a point in time stress relaxation rate;

[0136] Indicates that at the current moment The tensile force value;

[0137] Indicates that at the previous moment The tensile force value;

[0138] represents the time step, that is, the time interval used to calculate the stress relaxation rate.

[0139] The differential and integral operation unit is also used to calculate the energy integral value of the acoustic emission signal. This feature reflects the intensity of the acoustic emission event. This unit is the amplitude of the acoustic emission signal after the timestamp is aligned and filtered. The square of is integrated over time, and its discretization calculation can be expressed as:

[0140] ;

[0141] Where:

[0142] Indicates at a point in time A previous time window Energy accumulation value within;

[0143] Indicates at a point in time The acoustic emission signal amplitude;

[0144] Indicates the time window size;

[0145] represents the time step;

[0146] Indicates the time window from arrive Sum all time points.

[0147] The function of the feature vector synthesis unit is to combine all the feature parameters calculated by the differential and integral operation units at the same time stamp to form a multi-dimensional process feature fingerprint vector. .

[0148] At the time point , the unit will calculate the real-time slope , stress relaxation rate And the energy integral value , synthesized into a column vector.

[0149] ;

[0150] Where;

[0151] Indicates at a point in time Characteristic vector of the tensioning process when ;

[0152] Indicates at a point in time The slope of the force-displacement curve;

[0153] Indicates at a point in time stress relaxation rate;

[0154] Indicates at a point in time The accumulated value of acoustic emission energy at ;

[0155] Represents a transpose operation.

[0156] The vector ) is the characteristic fingerprint of the tensioning process at that moment and is output to the real-time diagnosis and health quantification module.

[0157] Through the above implementation, the process feature fingerprint extraction module can efficiently extract a set of quantitative features that are highly sensitive to friction changes, anchor system stability and micro-damage events during the tensioning process from massive amounts of raw sensor data.

[0158] Real-time diagnosis and health quantification module, used to compare process characteristic fingerprints with multi-dimensional dynamic health baseline models and calculate the tensile health index;

[0159] Specifically, in this embodiment, the real-time feature fingerprint output by the process feature fingerprint extraction module is continuously compared with the baseline model generated by the dynamic health baseline generation module, and finally a quantitative index that can intuitively reflect the health status of the current tensioning process is output.

[0160] The real-time diagnosis and health quantification module further includes a geometric deviation calculation unit, an acoustic anomaly calculation unit, and a nonlinear fusion unit in terms of logical structure;

[0161] A geometric deviation calculation unit is configured to calculate the degree of deviation between the real-time force-displacement trajectory in the process signature fingerprint and the reference trajectory in the multi-dimensional dynamic health baseline model. The unit receives real-time force-displacement data points and a reference force-displacement curve.

[0162] The unit calculates the geometric deviation by the following steps:

[0163] First, calculations are performed at multiple preset tension sampling points. Get the real-time steel strand elongation value , forming the real-time trajectory coordinate points At the same time, the benchmark elongation value corresponding to the sampling point is obtained from the benchmark trajectory , forming the reference trajectory coordinate points .

[0164] Then, calculate the Euclidean distance between these two coordinate points Since the two points are on the same tensile coordinate, the distance simplifies to the difference in elongation values.

[0165] ;

[0166] Where:

[0167] Indicates the Euclidean distance on the sampling points;

[0168] Indicates the The actual steel strand elongation value corresponding to each sampling point;

[0169] Indicates the The elongation value of the reference steel strand corresponding to each sampling point;

[0170] Indicates the The tensile force value corresponding to each sampling point.

[0171] The Euclidean distance calculated for all sample points The set is aggregated by the root mean square algorithm to obtain a comprehensive geometric deviation index .

[0172] ;

[0173] It represents the overall tension deviation index;

[0174] Indicates the total number of preset tension sampling points;

[0175] Indicates that To Sum up all the error values ​​of the sampling points;

[0176] Indicates the The error Euclidean distance at each sampling point.

[0177] The acoustic anomaly calculation unit is used to calculate the degree of deviation between the real-time acoustic emission signal characteristics in the process characteristic fingerprint and the acoustic benchmark in the multi-dimensional dynamic health baseline model.

[0178] The unit receives the real-time acoustic emission signal energy integral value , and the statistical distribution characteristics of the acoustic reference energy corresponding to the tension stage in the baseline model, namely the mean and standard deviation This unit quantifies the acoustic anomaly D by calculating the difference between the real-time value and the mean relative to the multiple of the standard deviation. .

[0179] ;

[0180] Where;

[0181] Indicates the degree of acoustic anomaly;

[0182] Indicates the actual measured energy integral value of the acoustic emission signal;

[0183] represents the mean acoustic energy of the corresponding stage in the baseline model;

[0184] represents the standard deviation of acoustic energy at the corresponding stage in the baseline model.

[0185] The nonlinear fusion unit is used to convert the geometric deviation With the acoustic anomaly Perform weighted fusion to generate a single normalized tensile health index .

[0186] The unit calculates the two deviation indices using a nonlinear function with a preset weight coefficient. For example, a negative exponential function may be used for fusion, so that the greater the deviation, the lower the health index.

[0187] ;

[0188] Where:

[0189] Indicates at a point in time Tensile health index at ;

[0190] Preset weight coefficient representing tension deviation;

[0191] Indicates the overall deviation of the tensioning process;

[0192] A preset weight coefficient indicating the degree of acoustic anomaly;

[0193] Indicates the degree of acoustic abnormality;

[0194] represents the natural exponential function.

[0195] Through the above implementation, the real-time diagnosis and health quantification module can transform multi-dimensional, abstract process characteristics into a continuously changing health score ranging from 0 to 100. This score objectively and quantitatively reflects the degree to which the real-time tensioning process conforms to the ideal health state, providing a direct and reliable basis for on-site personnel's judgment and subsequent decision-making of the adaptive control module.

[0196] The causal knowledge graph and the adaptive control module store the mapping relationship between fault mode, feature fingerprint and intervention strategy, and generate and send control instructions to the front-end execution mechanism according to the abnormality of the tension health index or the specific pattern of the process feature fingerprint.

[0197] Specifically, the causal knowledge graph and adaptive control module of this embodiment are the core of realizing the closed-loop control of the present invention. Its function is to make intelligent decisions based on real-time diagnostic results and automatically generate intervention instructions to realize adaptive control of the tensioning process.

[0198] In terms of logical structure, this module includes a knowledge graph repository, a decision triggering and pattern matching unit, and an intervention instruction generation unit.

[0199] The knowledge graph repository is used to consolidate and store domain expert knowledge and historical data mining results. It stores a series of "fault mode-feature fingerprint-intervention strategy" mappings, each structured as a triple.

[0200] ;

[0201] Where;

[0202] Indicates the "Fault-Indication-Intervention" mapping rules;

[0203] Indicates the A specific failure mode;

[0204] Indication and failure modes A collection of relevant feature fingerprints / feature indicators;

[0205] Indicates a fault and indications The intervention or correction strategies adopted.

[0206] For failure mode The quantitative definition of the corresponding process characteristic fingerprint. It is not a single value, but a set of logical conditions consisting of multiple characteristic parameters.

[0207] Exemplarily, the knowledge graph repository includes at least the following two mapping relationships:

[0208] The first mapping relationship is used to identify the fault of excessive friction resistance in the hole. Its process characteristic fingerprint Quantitatively defined as: over a continuous period of time Internal, real-time slope of force-displacement curve Consistently below the preset lower limit of the baseline model slope , and the acoustic emission energy integral value It shows a steady growth trend. Its related intervention strategies It means "slight retreat-advance again".

[0209] The second mapping relationship is used to identify broken wire faults in steel strands. Its process feature fingerprint Quantified as: Tension In a very short time The occurrence of the preset threshold The instantaneous drop of the acoustic emission signal energy A high energy event threshold is exceeded The pulse of the disease. Its associated intervention strategies "Immediately suspend tensioning";

[0210] The decision triggering and pattern matching unit is used to monitor the system status in real time and trigger the decision process according to the preset rules. This unit receives the tensile health index output by the real-time diagnosis and health quantification module. , and the process feature fingerprint vector output by the process feature fingerprint extraction module .

[0211] The trigger logic of this unit includes two parallel judgment conditions:

[0212] The first is based on the threshold judgment of the tension health index. This unit will With a preset alarm threshold Compare. , then the system is considered abnormal and the decision-making process is triggered.

[0213] The second is pattern matching based on process feature fingerprints. This unit converts the real-time feature fingerprint vector All predefined fault feature fingerprints in the knowledge graph repository Match one by one. The matching process is a logical judgment process, and its mathematical expression can be defined as;

[0214] ;

[0215] Where;

[0216] Indicates at a point in time Is the match successful? Boolean indicator function of a failure mode;

[0217] Indicates the failure mode number;

[0218] Represents the current time point, used to obtain time-related feature vectors ;

[0219] Indicates the current time observed eigenvectors;

[0220] Indicates the The set of feature conditions defined in the rule.

[0221] The function of the intervention instruction generation unit is to generate a control instruction sequence that can be directly executed by the front-end execution mechanism after the decision is triggered.

[0222] When the decision trigger and pattern matching unit determines that intervention is required, the unit first determines the trigger source (whether the health index is lower than the threshold or a specific fault mode is matched) ) Retrieve the corresponding intervention strategy identification code from the knowledge graph repository .

[0223] The unit then codes the intervention strategy Parse it into a set of specific control instructions with a time sequence relationship.

[0224] Finally, the unit sends the generated instruction sequence to the actuator in the front-end tensioning equipment through the preset communication interface to complete closed-loop control.

[0225] The causal knowledge graph and adaptive control module transform expert experience and data patterns into machine-executable logic, automating the entire process from data perception to diagnostic evaluation and decision-making intervention. This enables timely correction of deviations during the tensioning process, effectively avoiding major quality and safety incidents, and ensuring the quality and safety of tensioning construction.

[0226] The device for managing prestressed tensioning data of a highway bridge construction project described below and the prestressed tensioning data management system for a highway bridge construction project described above can correspond to each other.

[0227] Please see the attached Figure 4 The present invention also provides a method for managing prestressed tensioning data in a highway bridge construction project, comprising the following steps:

[0228] Generate a multi-dimensional dynamic health baseline model including a force-displacement benchmark curve and an acoustic emission signal energy fluctuation range model;

[0229] During the tensioning process, the tensioning force, steel strand elongation, acoustic emission signal and equipment posture data are collected synchronously;

[0230] Extract process feature fingerprints including the slope of the force-displacement curve and the energy characteristics of the acoustic emission signal from the collected data in real time;

[0231] Comparing the process characteristic fingerprint with the multi-dimensional dynamic health baseline model to calculate a quantitative tensile health index;

[0232] determining whether the tension health index is lower than a preset threshold or whether the process characteristic fingerprint matches a preset failure mode;

[0233] If so, the corresponding adaptive intervention strategy is generated and executed based on the predefined causal knowledge graph;

[0234] All the data of the tensioning process, the extracted characteristic fingerprints, the health index and the intervention records are structured and archived for the subsequent optimization of the multidimensional dynamic health baseline model and the causal knowledge graph.

[0235] The device of this embodiment can be used to execute the above method embodiment, and its principles and technical effects are similar, so they will not be repeated here.

[0236] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A prestressed tensioning data management system for highway bridge construction projects, characterized in that: include: Multi-dimensional data acquisition module, used to synchronously collect tensioning force, steel strand elongation value, acoustic emission signal and equipment posture data during the tensioning process to generate multi-dimensional data; Dynamic health baseline generation module, used to generate a multi-dimensional dynamic health baseline model based on design parameters, material properties, ambient temperature and historical good data; The dynamic health baseline generation module includes: A theoretical model building unit, configured to calculate a force-displacement reference curve based on the pore geometry information in the design parameters and the elastic modulus in the material properties; A statistical model building unit, used to analyze historical excellent data, extract and establish a fluctuation range model of the acoustic emission signal energy; A dynamic correction unit, configured to perform linear compensation correction on the force-displacement reference curve according to the ambient temperature collected in real time; A process feature fingerprint extraction module is used to extract a feature vector representing the state of the tensioning process from multidimensional data as a process feature fingerprint; The process feature fingerprint extraction module includes: A signal synchronization and preprocessing unit, configured to perform time stamp alignment on the data streams of the tension force, the steel strand elongation value, and the acoustic emission signal; Differentiation and integration operation unit, used to calculate the real-time slope of the force-displacement curve, stress relaxation rate and acoustic emission signal energy integral value of the data after time stamp alignment; The feature vector synthesis unit is used to combine the calculation results of the differential and integral operation units into a multi-dimensional process feature fingerprint at each time stamp; Real-time diagnosis and health quantification module, used to compare process characteristic fingerprints with multi-dimensional dynamic health baseline models and calculate the tensile health index; The causal knowledge graph and the adaptive control module store the mapping relationship between fault mode, feature fingerprint and intervention strategy, and generate and send control instructions to the front-end execution mechanism according to the abnormality of the tension health index or the specific pattern of the process feature fingerprint.

2. A prestressed tensioning data management system for highway bridge construction projects according to claim 1, characterized in that: The multidimensional data acquisition module generates multidimensional data by the following steps: Add a unified synchronization timestamp to the data streams from tension force, steel strand elongation, acoustic emission signals processed by the acoustic emission signal preprocessing unit, and equipment attitude sensors; Aligning the data streams stamped with the synchronization timestamps according to a preset sampling frequency; The aligned data streams are encapsulated into time series data frames with a unified data structure and output as the multidimensional data.

3. A prestressed tensioning data management system for highway bridge construction projects according to claim 1, characterized in that: The statistical model building unit establishes the fluctuation range model of the acoustic emission signal energy through the following steps: Normalize the tension force or tension time axis in multiple sets of historical high-quality data to form a unified analysis dimension; Divide the normalized analysis dimension into multiple continuous intervals; Count the acoustic emission signal energy values ​​corresponding to all historical good data falling within each interval, and calculate their statistical distribution characteristics; A confidence interval is determined for each interval according to the statistical distribution characteristics, and the set of confidence intervals of all intervals together constitutes the fluctuation range model of the acoustic emission signal energy.

4. A prestressed tensioning data management system for highway bridge construction projects according to claim 1, characterized in that: The differential and integral operation unit calculates the real-time slope of the force-displacement curve through the following steps: Set a sliding calculation window containing a preset number of data points; Continuously moving the sliding calculation window along the tension force and steel strand elongation value data sequence aligned with the timestamps; At each position of the sliding calculation window, applying the least squares method to the set of data points contained in the window to perform linear fitting to obtain a fitting straight line; The slope value of the fitting straight line is used as the real-time slope of the force-displacement curve corresponding to the center point of the sliding calculation window.

5. A prestressed tensioning data management system for highway bridge construction projects according to claim 1, characterized in that: The real-time diagnosis and health quantification module further includes: A geometric deviation calculation unit, used to calculate the degree of deviation between the force-displacement trajectory in the process characteristic fingerprint and the corresponding reference trajectory in the multidimensional dynamic health baseline model; an acoustic anomaly calculation unit, configured to calculate a degree of deviation between an acoustic emission signal feature in a process characteristic fingerprint and a corresponding acoustic benchmark in the multidimensional dynamic health baseline model; A nonlinear fusion unit is used to perform weighted fusion on the geometric deviation and the acoustic anomaly to generate a tensile health index.

6. A prestressed tensioning data management system for highway bridge construction projects according to claim 5, characterized in that: The real-time diagnosis and health quantification module calculates the tensile health index through the following steps: At multiple preset tension sampling points, the Euclidean distances between the real-time force-displacement trajectory coordinates and the corresponding coordinates in the benchmark trajectory of the multidimensional dynamic health baseline model are calculated, and the set of Euclidean distances is used as the geometric deviation. Calculating the standard deviation multiple of the real-time acoustic emission signal energy integral value and the acoustic baseline energy in the multidimensional dynamic health baseline model, and using the standard deviation multiple as the acoustic abnormality degree; A nonlinear function with a preset weight coefficient is applied to calculate the geometric deviation and acoustic anomaly to obtain a standardized tensile health index.

7. A prestressed tensioning data management system for highway bridge construction projects according to claim 1, characterized in that: The mapping relationship between the causal knowledge graph and the adaptive control module includes: The excessive friction failure mode of the pores is mapped to a characteristic fingerprint with a continuously low slope of the force-displacement curve and a steadily increasing acoustic emission energy, and is then linked to an intervention strategy of slight retreat and then re-advance. The strand breakage failure mode is mapped to the characteristic fingerprint of an instantaneous drop in tensioning force and the appearance of high-energy pulses in the acoustic emission signal, and is then linked to an intervention strategy of immediately suspending tensioning.

8. A method for managing prestressed tensioning data for a highway bridge construction project, applied to a prestressed tensioning data management system for a highway bridge construction project as claimed in any one of claims 1 to 7, characterized in that: The following steps are involved: Generate a multi-dimensional dynamic health baseline model including a force-displacement benchmark curve and an acoustic emission signal energy fluctuation range model; During the tensioning process, the tensioning force, steel strand elongation, acoustic emission signal and equipment posture data are collected synchronously; Extract process feature fingerprints including the slope of the force-displacement curve and the energy characteristics of the acoustic emission signal from the collected data in real time; Comparing the process characteristic fingerprint with the multi-dimensional dynamic health baseline model to calculate a quantitative tensile health index; determining whether the tension health index is lower than a preset threshold or whether the process characteristic fingerprint matches a preset failure mode; If so, the corresponding adaptive intervention strategy is generated and executed based on the predefined causal knowledge graph; All data of the tensioning process, extracted feature fingerprints, health index and intervention records are structured and archived for subsequent optimization of the multidimensional dynamic health baseline model and the causal knowledge graph.

Images