A prestressed plate beam intelligent tensioning control method and system based on multi-source sensing information fusion

Abstract

This invention discloses an intelligent tension control method and system for prestressed slab beams based on multi-source sensor information fusion. The method includes: real-time acquisition of tension force F(t), strain ε(t), elongation L(t), visual feature D(t), and temperature T(t); constructing a state vector X(t) = [F(t), ε(t), L(t), D(t), T(t)]; and standardizing each parameter to obtain Z(t), with the formula Z... i =(X i -μ i ) / σ i , where μ i σ is the historical mean. i Let be the standard deviation; construct a weighted fusion evaluation model S(t)=ω1Z F +ω2Z ε +ω3Z L +ω4Z D +ω5Z T The system performs hierarchical closed-loop control based on S(t), and simultaneously determines single-source anomalies based on visual characteristics and temperature, triggering alarms or shutdowns. The system includes a base, jacks, various sensors, an industrial camera, an infrared camera, and a control module. This invention achieves multi-dimensional real-time monitoring and closed-loop adaptive control, and can automatically identify wire slippage, eccentric tensioning, and temperature anomalies, thereby improving tensioning accuracy and construction safety.

Description

Technical Field

[0001] This invention relates to the field of bridge construction technology, specifically to a method and system for intelligent tensioning control of prestressed slab beams based on multi-source sensor information fusion. Background Technology

[0002] Prestressing technology is a key technology widely used in bridge engineering. The quality of tensioning construction of prestressed slab girders directly determines the structure's load-bearing capacity, crack resistance, and durability. Traditional tensioning construction mainly relies on manual operation of hydraulic jacks, using hydraulic pressure measurement to calculate tension force and steel ruler measurement for elongation control. This method has significant drawbacks: the accuracy of tension force control is greatly affected by hydraulic pressure gauge reading errors and human factors; deformation measurement is inaccurate and cannot reflect the overall deformation of the slab girder in real time; the entire tensioning process lacks real-time monitoring of the girder's condition, making it difficult to detect potential problems such as micro-cracks on the concrete surface or anchorage abnormalities caused by stress concentration in the first instance.

[0003] With the development of automation technology, some intelligent tensioning systems have emerged, most of which focus on the automatic control of tension force and elongation. However, the monitoring dimensions of existing intelligent tensioning systems are relatively limited, mostly focusing only on force and displacement parameters, and failing to effectively monitor the overall deformation of the beam, the uniformity of stress distribution, and local damage caused by tensioning during the tensioning process. For highly concealed hidden dangers such as eccentric tensioning of steel strands, wire slippage, and abnormal local temperatures, existing systems struggle to achieve automatic identification and timely warning.

[0004] Therefore, developing a control method and system capable of multi-dimensional, real-time monitoring of the entire tensioning process, with intelligent early warning and closed-loop control capabilities, is of great significance for ensuring the quality and safety of prestressed construction. Specifically, the following technical problems urgently need to be solved: inability to accurately obtain the true stress and strain of the steel strands in real time; inability to monitor changes in the shape and displacement of the steel strands during tensioning; lack of automatic identification capability for eccentric tensioning and local damage to the steel strands; lack of temperature monitoring leading to the inability to detect slippage and friction anomalies; and low precision due to the inability to form a closed-loop adjustment in tensioning control. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for intelligent tensioning control of prestressed slab beams based on multi-source sensor information fusion.

[0006] To achieve the above objectives, the present invention provides the following technical solution: This application provides a method for intelligent tension control of prestressed slab beams based on multi-source sensor information fusion, including: Real-time acquisition of multi-source data during the tensioning process: tension force F(t) is obtained through stress sensor, strain ε(t) is obtained through strain gauge, elongation L(t) of steel strand is obtained through displacement sensor, images of steel strand are obtained through industrial camera and geometric features are extracted to form visual features D(t), and temperature T(t) of anchorage area and steel strand is obtained through infrared camera. The multi-source data is constructed as a state vector X(t) = [F(t), ε(t), L(t), D(t), T(t)]; The parameters in the state vector are standardized to obtain a standardized vector. The standardized formula is , Let i be the i-th parameter in the state vector X(t). The historical average of the parameter. Standard deviation; A weighted fusion evaluation model is constructed to obtain comprehensive security evaluation indicators. ,in to Weights for each indicator; Based on the comprehensive safety evaluation index S(t), hierarchical closed-loop control is executed to adjust the tensioning operation. At the same time, single-source anomaly is determined according to the visual feature D(t) and temperature T(t), and an alarm or shutdown is triggered when an anomaly is determined.

[0007] Preferably, extracting visual features D(t) from the steel strand image and performing anomaly detection includes: The outline boundary of the steel strand is extracted by the edge detection operator, and the equivalent diameter d(t) is obtained by least squares circle fitting. Calculate the rate of change of diameter ,in The initial diameter; Calculate the lateral offset of the center of the steel strand profile relative to the initial position. ; The visual feature D(t) includes the diameter change rate. and lateral offset ; when Greater than the preset maximum diameter change rate threshold When it is determined that the thread has slipped, Greater than the preset maximum offset threshold The system detects eccentric tensioning and triggers an alarm.

[0008] Preferably, the single-source anomaly determination based on temperature T(t) includes: Calculate the rate of temperature rise , where dT / dt is the derivative of temperature with respect to time; When T(t) is greater than the preset maximum temperature threshold ,or Greater than the preset maximum temperature rise rate threshold When an abnormal friction or localized damage is detected, an emergency shutdown is triggered.

[0009] Preferably, before standardizing the parameters in the state vector, a theoretical model comparison is also included: Calculate theoretical elongation ,in Where E is the initial length of the steel strand, E is the elastic modulus, and A is the cross-sectional area; The actual elongation L(t) is compared with the theoretical elongation. Deviations are incorporated into the standardization process.

[0010] Preferably, the hierarchical closed-loop control includes: Set multiple threshold levels: α < β < γ; When S(t) < α, maintain the normal tensioning rate; when α ≤ S(t) < β, reduce the tensioning rate; when β ≤ S(t) < γ, pause tensioning and perform a check; when S(t) ≥ γ, immediately stop and lock the machine. The tensioning rate v(t) is adjusted by the following formula: ,in The initial tensioning rate is set, and k is the adjustment coefficient.

[0011] Preferably, it also includes a data-driven self-learning step: Record historical data for each tensioning process to form a sample set Ω={(X(t),S(t))}; Update the weight parameters using the sample set Ω to And the thresholds α, β, and γ in hierarchical closed-loop control, to achieve adaptive optimization under different operating conditions.

[0012] A prestressed slab beam intelligent tensioning system based on multi-source sensor information fusion, used to implement the method described above, includes: The pedestal has a fixed end seat and a tensioning end seat at each end; An electric hydraulic jack and a movable crossbeam are sequentially arranged on the outside of the tensioning end seat; Prestressed steel bars extend axially along the platform, with one end anchored to the fixed end seat and the other end passing through the tensioning end seat and the movable crossbeam before being connected to the electric hydraulic jack. A stress sensor is installed in the force transmission path at the tension end of the prestressed steel bar to collect the tension force F(t); A strain sensor is arranged on the prestressed steel bar to collect strain ε(t); A displacement sensor is installed at the movable crossbeam to collect the elongation L(t) of the steel strand; An industrial camera, mounted on the side of the platform, is used to acquire images of prestressed steel bars; An infrared camera, positioned above the tensioning end, is used to collect temperature T(t); The data acquisition and fusion module is electrically connected to the stress sensor, strain sensor, displacement sensor, industrial camera and infrared camera. It is configured to perform state vector construction, standardization processing and weighted fusion evaluation to obtain a comprehensive safety evaluation index S(t), and to determine single-source anomalies based on visual features extracted from the images of the industrial camera and the temperature collected by the infrared camera. The closed-loop tensioning control module is connected to the data acquisition and fusion module and the electro-hydraulic jack, and is used to control the action of the electro-hydraulic jack according to S(t) and the anomaly judgment result.

[0013] Preferably, the industrial camera and the data acquisition and fusion module are configured as follows: The equivalent diameter d(t) is obtained from the image through edge detection and least-squares circle fitting; Calculate the rate of change of diameter and lateral offset ; when When determining slippage, The system detects eccentric tensioning and triggers an alarm. in The initial diameter, To preset the maximum diameter change rate threshold, This is the preset maximum offset threshold.

[0014] Preferably, the infrared camera and the data acquisition and fusion module are configured as follows: Calculate the rate of temperature rise ; when or If abnormal friction or localized damage is detected, an emergency shutdown will be triggered. in To preset the maximum temperature threshold, This is the preset maximum temperature rise rate threshold.

[0015] Preferably, the data acquisition and fusion module further includes a historical database and a self-learning unit. The self-learning unit is used to store a sample set Ω={(X(t),S(t))} of the historical tensioning process and to update the weights in the weighted fusion using the sample set Ω. to And the thresholds α, β, and γ in hierarchical closed-loop control.

[0016] Compared with the prior art, this application has the following beneficial effects: This invention proposes an intelligent tensioning control method and system for prestressed slab beams based on multi-source sensor information fusion. By integrating multi-source sensor information such as mechanics, vision, and temperature, a standardized state vector is constructed and a comprehensive safety evaluation index is generated through weighted calculation. Simultaneously, single-source anomaly judgment is performed on visual and temperature data. Combined with closed-loop adaptive adjustment under multi-level thresholds, real-time and accurate perception of actual stress, deformation, diameter changes, lateral offset, and temperature distribution during prestressing tensioning is achieved. It can automatically identify potential hazards such as wire slippage, eccentric tensioning, abnormal friction, and local damage, and promptly issue alarms or automatically adjust the tensioning rate, pause, or even stop the machine in an emergency. This significantly improves the accuracy and uniformity of tension force control, effectively avoids safety accidents such as wire breakage and structural damage, improves the finished quality of prestressed slab beams, and ensures that the entire process data is traceable. It also has self-learning optimization capabilities and can autonomously evolve according to construction conditions. Attached Figure Description

[0017] Figure 1 This is a structural schematic diagram of the intelligent tensioning system for prestressed slab beams of the present invention.

[0018] Figure 2 This is a system flowchart of the method of the present invention.

[0019] Figure label: 1. Electric hydraulic jack; 2. Base; 3. Prestressed steel bars; 4. Movable crossbeam; 5. Industrial camera; 6. Infrared camera. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Furthermore, in this invention, an element referred to as fixed to or disposed on another element may be directly disposed on the other element, or there may be an intermediate element. When an element is considered to be connected to another element, it may be directly connected to the other element, or there may be an intermediate element present simultaneously. The terms vertical, horizontal, left, right, and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0022] See Figures 1-2 This application provides a method for intelligent tension control of prestressed slab beams based on multi-source sensor information fusion, including steps one to five: Step 1. Real-time acquisition of multi-source data during the tensioning process: obtain tension force F(t) through stress sensor, strain ε(t) through strain gauge, elongation L(t) of steel strand through displacement sensor, obtain image of steel strand through industrial camera 5 and extract geometric features to form visual features D(t), and obtain temperature T(t) of anchor area and steel strand through infrared camera 6.

[0023] Before the tensioning operation begins, the installation and calibration of all sensors are completed. Stress sensors are fixedly installed on the force transmission path near the tensioning and fixing ends of the prestressed steel reinforcement 3 to measure the actual force F(t) of the steel strand during tensioning in real time. Strain sensors are uniformly attached to the surface of the prestressed steel reinforcement 3 to monitor the local strain of the steel strand and the minute deformation of the beam, obtaining the strain ε(t). Displacement sensors are installed at the movable crossbeam 4 to detect the total elongation L(t) of the steel strand in real time. An industrial camera 5 is fixed above the center of the platform 2, covering the exposed section of the steel strand and the beam outline, to acquire high-definition images of the steel strands. An infrared camera 6 is positioned above the tensioning end or on the side of the beam to monitor the temperature field distribution in the anchorage area and on the surface of the steel strands during tensioning, obtaining the temperature T(t). All sensor signals are connected to the data acquisition host, where the control system performs zero-point calibration, data synchronization, frame rate setting, and trigger mode setting to ensure that all data are acquired under the same time reference.

[0024] After tensioning begins, the electric hydraulic jack 1 gradually increases the force according to the preset graded loading requirements, and each sensor synchronously and continuously collects data to form a real-time multi-dimensional data stream.

[0025] Step 2. Construct the multi-source data into a state vector X(t)=[F(t),ε(t),L(t),D(t),T(t)].

[0026] The data acquisition and fusion module unifies the multi-source data collected in step one into a five-dimensional state vector: X(t)=[F(t),ε(t),L(t),D(t),T(t)] Wherein, F(t) is the tension force value collected in real time by the stress sensor, ε(t) is the strain value collected by the strain gauge, L(t) is the elongation of the steel strand collected by the displacement sensor, D(t) is the visual feature extracted from the image of industrial camera 5, and T(t) is the temperature value collected by infrared camera 6; this state vector gathers information from five dimensions of force, strain, displacement, vision and temperature at the same timestamp, and fully reflects the instantaneous state of the tensioning process.

[0027] Step 3. Standardize the parameters in the state vector to obtain a standardized vector. The standardized formula is , Let i be the i-th parameter in the state vector X(t). The historical average of the parameter. The standard deviation is denoted as .

[0028] To eliminate dimensional differences between parameters and enable comparison and fusion of different physical quantities on a unified scale, it is necessary to standardize the parameters in the state vector X(t). Specifically, the Z-score standardization method is used, with the following formula: in, Let F(t), ε(t), L(t), D(t), and T(t) represent the i-th parameter in the state vector X(t), which are F(t), ε(t), L(t), D(t), and T(t) in sequence. This represents the historical average of the corresponding parameter. This represents the standard deviation. In practice, if it's the first tensioning operation or historical data is insufficient, and These values ​​can be obtained using design experience or through pre-tensioning calibration tests. As the number of tensioning batches accumulates, and It can be continuously updated based on historical databases.

[0029] After substituting each parameter into the standardization formula, we obtain the standardized vector: in Corresponding to the standardized tension value, Corresponding to the standardized strain value, Corresponding to the standardized elongation value, Corresponding to the standardized visual feature values, The corresponding standardized temperature value.

[0030] Step 4. Construct a weighted fusion evaluation model to obtain comprehensive safety evaluation indicators. ,in to The weights of each indicator are given.

[0031] Based on the standardized vector Z(t), a weighted fusion evaluation model is constructed to calculate the comprehensive safety evaluation index S(t), as shown in the formula: ;in, As the weight of the tension index, As the weight of the strain index, The weight of the elongation index. The weights of visual feature indicators, The weights of temperature indicators are set based on the importance of each indicator in historical tensioning data and can be updated through a self-learning mechanism during subsequent operation. The value of S(t) comprehensively reflects the degree of deviation between the current tensioning state and the normal working condition. The larger the value, the higher the degree of abnormality.

[0032] Step 5. Based on the comprehensive safety evaluation index S(t), perform hierarchical closed-loop control to adjust the tensioning operation. At the same time, perform single-source anomaly judgment based on the visual feature D(t) and temperature T(t), and trigger an alarm or shutdown when an anomaly is judged.

[0033] In one specific implementation, extracting visual features D(t) from the steel strand image and performing anomaly detection includes: The outline boundary of the steel strand is extracted by the edge detection operator, and the equivalent diameter d(t) is obtained by least squares circle fitting. Calculate the rate of change of diameter ,in The initial diameter; Calculate the lateral offset of the center of the steel strand profile relative to the initial position. ; The visual feature D(t) includes the diameter change rate. and lateral offset ; when Greater than the preset maximum diameter change rate threshold When it is determined that the thread has slipped, Greater than the preset maximum offset threshold The system detects eccentric tensioning and triggers an alarm.

[0034] While performing the fusion evaluation in steps two through four, the system performs parallel special processing on the visual data D(t) to achieve rapid identification of wire slippage and eccentric tensioning. The specific steps are as follows: First, the images of steel strands captured by industrial camera 5 are preprocessed, and the outline boundary of the steel strands is extracted using the Canny edge detection operator. The Canny operator can accurately identify the edge pixels between the steel strands and the background by calculating the magnitude and direction of the image gradient, thereby obtaining a clear outline.

[0035] Then, based on the extracted contour boundary point set, the least squares circle fitting algorithm is used to iteratively solve for a circle that best matches the contour. The diameter of this circle is the equivalent diameter d(t) of the steel strand under the current tension state. The least squares circle fitting obtains the optimal center position and diameter by minimizing the sum of squared distances from each edge point to the center of the fitted circle.

[0036] Based on the above fitting results, two key visual feature indices are calculated: First, the rate of change in diameter The calculation formula is: in The initial diameter of the steel strand was determined using the same image processing method before tensioning. It reflects the degree of diameter reduction of the steel strand cross section during the tensioning process.

[0037] Secondly, lateral offset , is the offset distance of the lateral position of the fitted center of the steel strand in the current image relative to the initial calibration position, which is obtained by comparing the difference between the coordinates of the fitted center in the current image and the initial image.

[0038] The above and As a component of the visual feature D(t), it is incorporated into the state vector in step two.

[0039] In determining single-source anomalies, a preset threshold σ for the maximum diameter change rate is used. max and maximum offset threshold X max Two thresholds; when > When this occurs, it indicates an abnormal reduction in the diameter of the steel strand, which is determined to be a slippage; when >X max When this occurs, it indicates a significant lateral shift in the centerline of the steel strand, indicating eccentric tensioning. Regardless of the triggering condition, the system immediately activates the alarm procedure, issuing an audible and visual warning signal.

[0040] In one specific implementation, the single-source anomaly determination based on temperature T(t) includes: Calculate the rate of temperature rise , where dT / dt is the derivative of temperature with respect to time; When T(t) is greater than the preset maximum temperature threshold ,or Greater than the preset maximum temperature rise rate threshold When an abnormal friction or localized damage is detected, an emergency shutdown is triggered.

[0041] In parallel with visual feature processing, the system also performs independent anomaly detection based on the temperature T(t) collected by infrared camera 6. Specifically, this includes: First, calculate the rate of temperature rise. , defined as the change in temperature per unit time, that is, the derivative of temperature with respect to time: In actual calculations, the temperature difference between adjacent sampling times can be obtained by dividing the sampling interval.

[0042] Then, the real-time temperature T(t) and the temperature rise rate are... Each value is compared to a preset threshold. The preset maximum temperature threshold is... and maximum temperature rise rate threshold For each pixel region in an infrared image, when T(t) > 0 in any region or the area > If such an anomaly is detected, the system will determine that there is abnormal friction or localized damage in the area. Abnormal friction usually stems from excessive slippage friction between the steel strand and the anchor, while localized damage may be caused by stress concentration or broken wires inside the steel strand. Once such anomalies are detected, the system will immediately trigger an emergency stop command, lock the current state of the jack, and prevent the accident from escalating.

[0043] In one specific implementation, a theoretical model comparison is included before standardizing the parameters in the state vector: Calculate theoretical elongation ,in Where E is the initial length of the steel strand, E is the elastic modulus, and A is the cross-sectional area; The actual elongation L(t) is compared with the theoretical elongation. Deviations are incorporated into the standardization process.

[0044] To further improve the accuracy of anomaly identification, a theoretical model comparison step can be introduced before standardizing the state vector.

[0045] First, based on the real-time tension F(t) and the known parameters of the steel strand, the theoretical elongation L is calculated. theory The calculation formula is: L theory =(F(t)× ) / (E×A) in, E is the initial length of the steel strand before tensioning, E is the elastic modulus of the steel strand material, and A is the nominal cross-sectional area of ​​the steel strand.

[0046] Then, the measured elongation L(t) and the theoretical elongation L are calculated. theory The deviation between the values ​​is treated as an independent anomaly, interacting with the components in the standardized vector. A large deviation indicates a significant deviation between the actual tensioning process and the ideal elastic model, which may be a sign of abnormal tension force transmission, local yielding of the steel strand, or anchor slippage. In the weighted fusion model, this deviation will affect the calculation result of the comprehensive safety evaluation index S(t).

[0047] In one specific implementation, the hierarchical closed-loop control includes: Set multiple threshold levels: α < β < γ; When S(t) < α, maintain the normal tensioning rate; when α ≤ S(t) < β, reduce the tensioning rate; when β ≤ S(t) < γ, pause tensioning and perform a check; when S(t) ≥ γ, immediately stop and lock the machine. The tensioning rate v(t) is adjusted by the following formula: ,in The initial tensioning rate is set, and k is the adjustment coefficient.

[0048] The closed-loop tensioning control module executes a hierarchical control strategy based on the comprehensive safety evaluation index S(t). First, three thresholds α, β, and γ are set, satisfying a strict increasing relationship of α < β < γ. Among them, α is the early warning threshold, β is the alarm threshold, and γ is the emergency shutdown threshold. The initial values ​​of each threshold are set according to historical data and engineering experience.

[0049] The specific control logic can be divided into the following four cases: Case 1: When S(t) < α, it indicates that the current tensioning state is in the safe range, the system maintains the normal tensioning rate, and the electric hydraulic jack 1 continues to operate according to the preset loading program.

[0050] Scenario 2: When α ≤ S(t) < β, it indicates a slight anomaly in the tension state. The system automatically reduces the tension rate and slows down the loading speed to observe the development trend of the anomaly. At this time, the tension rate v(t) is continuously adjusted according to the following formula: in, To initially set the tensioning rate, This is the adjustment coefficient. This formula makes the rate vary with... The increase is smoothed down, avoiding the impact of abrupt changes on the beam.

[0051] Scenario 3: When β≤S(t)<γ, it indicates that the tensioning state is moderately abnormal. The system suspends tensioning, maintains the current oil pressure and force value unchanged, and enters the verification state. At this time, the operator needs to check the equipment status, sensor readings and beam appearance. The system can only continue after the fault is eliminated.

[0052] Scenario 4: When S(t)≥γ, it indicates a serious abnormality in the tensioning state. The system immediately stops and locks, cuts off the hydraulic circuit or shuts down the servo motor, and keeps the electric hydraulic jack 1 in its current position without moving. At the same time, the emergency stop command triggered by visual and temperature-based single-source judgment also directly leads to this state.

[0053] In one specific implementation, a data-driven self-learning step is also included: Record historical data for each tensioning process to form a sample set Ω={(X(t),S(t))}; Update the weight parameters using the sample set Ω to And the thresholds α, β, and γ in hierarchical closed-loop control, to achieve adaptive optimization under different operating conditions.

[0054] To continuously improve the system's judgment accuracy and adaptability, this invention introduces a data-driven self-learning mechanism.

[0055] After each complete tensioning process, the system stores all state vectors X(t) and corresponding comprehensive safety evaluation index S(t) from that tensioning process as a record in the historical database. As the number of tensioning batches increases, the accumulated sample data in the database constitutes a sample set Ω. Ω={(X(t),S(t))} The self-learning unit periodically or irregularly extracts data from the sample set Ω to adjust the weight parameters in the weighted fusion model. to Furthermore, the thresholds α, β, and γ in the hierarchical closed-loop control are retrained and optimized. Retraining can employ machine learning methods such as regression analysis and neural networks, aiming to maximize the ability of S(t) to distinguish various anomalies, and adjusting the values ​​of weights and thresholds. After self-learning optimization, the system's anomaly recognition sensitivity, false alarm rate, and control response characteristics are gradually improved, achieving adaptive tension control under different working conditions, material properties, and environmental conditions.

[0056] This application also provides a prestressed slab beam intelligent tensioning system based on multi-source sensor information fusion, used to implement the above method, including: The platform 2 has a fixed end seat and a tensioning end seat at each end; An electric hydraulic jack 1 and a movable crossbeam 4 are sequentially arranged on the outside of the tensioning end seat; Prestressed steel bar 3 extends axially along the platform 2, with one end anchored to the fixed end seat and the other end passing through the tensioning end seat and the movable crossbeam 4 and then connected to the electric hydraulic jack 1. A stress sensor is installed in the force transmission path at the tension end of the prestressed steel bar 3 to collect the tension force F(t); A strain sensor is arranged on the prestressed steel bar 3 to collect strain ε(t); A displacement sensor is installed at the movable crossbeam 4 to collect the elongation L(t) of the steel strand; An industrial camera 5 is mounted on the side of the platform 2 and is used to acquire images of the prestressed steel bars 3. Infrared camera 6 is positioned above the tensioning end to collect temperature T(t); The data acquisition and fusion module is electrically connected to the stress sensor, strain sensor, displacement sensor, industrial camera 5 and infrared camera 6. It is configured to perform state vector construction, standardization processing and weighted fusion evaluation to obtain a comprehensive safety evaluation index S(t), and to determine single-source anomalies based on the visual features extracted from the image of the industrial camera 5 and the temperature collected by the infrared camera 6. The closed-loop tensioning control module is connected to the data acquisition and fusion module and the electric hydraulic jack 1, and is used to control the action of the electric hydraulic jack 1 according to S(t) and the anomaly judgment result.

[0057] In one specific embodiment, the industrial camera 5 and the data acquisition and fusion module are configured as follows: The equivalent diameter d(t) is obtained from the image through edge detection and least-squares circle fitting; Calculate the rate of change of diameter and lateral offset ; when When determining slippage, The system detects eccentric tensioning and triggers an alarm. in The initial diameter, To preset the maximum diameter change rate threshold, This is the preset maximum offset threshold.

[0058] In one specific embodiment, the infrared camera 6 and the data acquisition and fusion module are configured as follows: Calculate the rate of temperature rise ; when or If abnormal friction or localized damage is detected, an emergency shutdown will be triggered. in To preset the maximum temperature threshold, This is the preset maximum temperature rise rate threshold.

[0059] In one specific implementation, the data acquisition and fusion module further includes a historical database and a self-learning unit. The self-learning unit is used to store a sample set Ω={(X(t),S(t))} of the historical tensioning process and to update the weights in the weighted fusion using the sample set Ω. to And the thresholds α, β, and γ in hierarchical closed-loop control.

[0060] In this embodiment, the specific implementation methods of each module in the system are described in detail.

[0061] The stress sensor is installed in the force transmission path at the tensioning end of the prestressed steel bar 3. In specific implementation, a through-type force sensor can be installed at the contact surface between the piston rod front end of the electric hydraulic jack 1 and the movable crossbeam 4, or a load sensor can be installed between the anchor and the end seat, so that the transmission path of the tension force F(t) passes completely through the sensor, ensuring that the measured value accurately reflects the real force on the steel strand.

[0062] Strain sensors, or strain gauges, are uniformly adhered to the surface of the prestressed steel reinforcement. During adhesion, the steel reinforcement surface must be ground and cleaned, and then fixed with specialized adhesive to ensure reliable strain transmission. Multiple strain gauges can be evenly distributed along the circumference and axial direction of the steel reinforcement to monitor for any localized uneven stress distribution.

[0063] The displacement sensor is installed at the movable crossbeam 4; specifically, a wire-type displacement sensor or a magnetostrictive displacement sensor can be used. The sensor body is fixed on the fixed reference of the base 2, and the wire end or magnetic ring end is connected to the movable crossbeam 4. As the movable crossbeam 4 moves, the elongation L(t) is output in real time.

[0064] Industrial camera 5 is positioned above the center of pedestal 2, with its lens facing the exposed section of the steel strand and the outline of the beam. In practice, a high-resolution industrial CMOS camera can be used, along with an appropriate lighting source, to ensure that the image clarity meets the accuracy requirements for edge detection and circle fitting. Industrial camera 5 is connected to the data acquisition and fusion module via a gigabit Ethernet port or an image acquisition card to transmit real-time image streams.

[0065] After acquiring images from industrial camera 5, the data acquisition and fusion module performs the aforementioned edge detection and least-squares circle fitting algorithms to obtain the equivalent diameter d(t) of the steel strand, and then calculates the diameter change rate σ. D and the lateral offset Δx; specifically, σ D =(d(t)-D0) / D0,Δ x This is determined by comparing the center coordinates of the fitted circle in the current image with the center coordinates of the fitted circle in the initial image laterally. When σ D Exceeding the threshold σ max When determining slippage, when Δ x Exceeding threshold X max When eccentric tension is detected, an alarm is triggered.

[0066] Infrared camera 6 is positioned above the tensioning end or on the side of the beam, its field of view covering the anchorage working area and the area near the steel strands. Infrared camera 6 can be an uncooled focal plane detector type thermal imager with a temperature resolution of no less than 0.05℃ and a frame rate sufficient for real-time monitoring. The data acquisition and fusion module calculates the temperature rise rate R of each pixel area based on the temperature field data transmitted from infrared camera 6. T =dT / dt, and continuously check whether there exists T(t)>T max Or RT >R max The area is designated as an anomaly. An emergency shutdown is triggered immediately upon detection of an anomaly.

[0067] The closed-loop tensioning control module is composed of an industrial controller or an embedded industrial control unit. It is connected to the solenoid valve or servo driver of the electric hydraulic jack 1 through analog output or digital communication. According to the received S(t) value and the abnormal judgment command, the control module outputs the corresponding control signal according to the aforementioned hierarchical control strategy to realize speed reduction, pause or stop.

[0068] The self-learning unit is a software submodule within the data acquisition and fusion module. It is linked to a historical database and periodically optimizes and updates the weights ω1 to ω5 and the thresholds α, β, and γ using newly added sample sets Ω. The updates to the weights and thresholds can be implemented based on algorithms such as artificial neural network training, support vector machine classification, or simple least squares fitting. The updated parameters automatically replace the old parameters and take effect in the next pull loop, thereby achieving continuous evolution of system performance.

[0069] After tensioning reaches the design requirements, the system enters the anchoring and subsequent testing phase. The control system maintains the designed tension level for a period to ensure the prestressed steel reinforcement force values ​​stabilize before anchoring is performed manually or automatically. After anchoring, displacement sensors and strain gauges continue to monitor the rebound amount and compare it with the designed rebound range. An industrial camera performs a secondary scan of the beam to identify any post-tensioning cracks, localized deformation, or abnormal exposed steel strands, and confirms that the temperature field has returned to normal with no sustained localized overheating. All data is automatically compiled to generate a tensioning process record, which can be directly used for prestressed quality acceptance and traceability analysis.

[0070] Example The following is in conjunction with the appendix Figure 1 and Figure 2 The present invention provides a detailed description of its specific application in the tensioning construction of a prestressed concrete slab beam.

[0071] Before the tensioning operation begins, the prestressed slab beam intelligent tensioning system of the present invention is installed on the tensioning platform 2. The platform 2 is provided with a fixed end seat and a tensioning end seat at both ends. The electric hydraulic jack 1 and the movable crossbeam 4 are arranged in sequence on the outside of the tensioning end seat. The prestressed steel bar 3 passes through the two end seats along the axial direction of the platform 2, with one end anchored to the fixed end seat and the other end passing through the tensioning end seat and the movable crossbeam 4 and then connected to the electric hydraulic jack 1.

[0072] The first step involves installing and calibrating all sensors. Stress sensors are fixed in the force transmission path near the tensioning and fixing ends of the prestressed steel strands 3 to measure the actual stress on the steel strands during tensioning. Strain gauges are evenly attached to the surface of the prestressed steel strands 3 to monitor local strain. Displacement sensors are installed at the movable crossbeam 4 to detect the total elongation of the steel strands in real time. An industrial camera 5 is fixed above the center of the platform 2, and its field of view is adjusted to cover the exposed section of the steel strands and the beam outline. An infrared camera 6 is positioned above the tensioning end to monitor the temperature distribution in the anchorage area and near the steel strands.

[0073] The second step is to initialize the data acquisition system; connect all the signals from the stress sensor, strain gauge, and displacement sensor to the data acquisition host, and connect the industrial camera 5 and infrared camera 6 to the monitoring unit through the corresponding interfaces; the control system performs zero-point calibration, data synchronization, frame rate setting, and trigger mode setting for each sensor to ensure that all data are acquired under the same time reference.

[0074] The third step is to input the tensioning conditions. Input the target tension force, theoretical elongation, graded tensioning steps, tensioning rate, and allowable error range specified in the design into the control system. The control system checks the effectiveness of each sensor channel and verifies the continuous stability of each real-time data channel. After confirming that there are no abnormalities, it enters the tensioning ready state.

[0075] Start the tensioning process, and the electric hydraulic jack 1 will perform graded loading according to the design requirements.

[0076] During each loading stage, stress sensors monitor the magnitude of the force F(t) on the steel strand in real time and compare it with the design tension force; strain gauges monitor the elongation of the steel strand and the micro-deformation of the beam in real time to obtain the strain ε(t), which is used to determine whether the tensioning is uniform and whether there is any local abnormal stress; displacement sensors detect the total elongation L(t) of the steel strand in real time and compare it with the theoretical elongation; industrial camera 5 captures images of the steel strand through image recognition algorithms and extracts visual features D(t), including diameter changes and lateral offset information, to identify abnormalities such as steel strand slippage, eccentric tensioning, and bending deformation; infrared camera 6 continuously monitors the temperature field of the anchorage area and the surface of the steel strand to obtain the temperature T(t). If local temperature rise occurs, it indicates that there may be risks such as excessive friction, stress concentration, and steel strand damage.

[0077] The control system constructs a state vector X(t) = [F(t), ε(t), L(t), D(t), T(t)] from the aforementioned multi-source data, and then performs multi-index standardization processing. The standardization formula is Z0. i =(X i -μ i ) / σ i , where μ iσ is the historical mean of the corresponding parameter. i The standard deviation is denoted as Z(t); after standardization, the standardized vector Z(t) = [Z F Z ε Z L Z D Z T ].

[0078] Based on this, a weighted fusion evaluation model is constructed to calculate the comprehensive safety evaluation index S(t)=ω1Z F +ω2Z ε +ω3Z L +ω4Z D +ω5Z T ω1 to ω5 are the weights of each indicator; these weights are derived from historical data training and can be updated adaptively.

[0079] While data fusion is underway, the system performs visual anomaly recognition in parallel. Images acquired by industrial camera 5 are first processed using the Canny edge detection operator to extract the steel strand contour boundary, and then the equivalent diameter d(t) of the steel strand is obtained using least-squares circle fitting. The diameter change rate σ is then calculated. D =(d(t)-D0) / D0 and the lateral offset Δ x When σ D The maximum diameter change rate threshold σ is greater than the preset threshold. max When Δ is reached, it is determined to be a slippage; when Δ x Greater than the preset maximum offset threshold X max At that time, it was determined to be eccentric tensioning.

[0080] The system also performs temperature anomaly detection in parallel. The temperature rise rate R is defined. T =dT / dt, when the temperature T(t) in any monitoring area is greater than T max Or the rate of temperature rise R T Greater than R max When this occurs, it is determined that there is abnormal friction or local damage in the area.

[0081] The closed-loop tension control module performs hierarchical control based on the comprehensive safety evaluation index S(t). The system presets multiple thresholds α<β<γ: when S(t)<α, normal tensioning occurs; when α≤S(t)<β, the tensioning rate is reduced; when β≤S(t)<γ, tensioning is paused and a check is performed; when S(t)≥γ, tensioning is immediately stopped and the current state is locked. The specific adjustment of the tensioning rate is achieved using the formula v(t)=v0·(1-k·S(t)), where v0 is the initial set tensioning rate and k is the adjustment coefficient. This formula ensures that the rate decreases smoothly as S(t) increases, avoiding abrupt changes that could impact the beam.

[0082] In addition, if the single-source anomaly determination detects issues such as slippage, eccentric tensioning, or abnormal temperature, the system will directly trigger an alarm or emergency shutdown command without waiting for the determination result of the fusion index S(t).

[0083] After each tensioning process, the system automatically stores the state vector X(t) and comprehensive safety evaluation index S(t) of this tensioning process into the historical database to form a sample set. The self-learning unit uses the newly added samples to update the weight parameters ω1 to ω5 and the thresholds α, β, and γ in the hierarchical closed-loop control, so that the system can achieve adaptive optimization under different working conditions.

[0084] After the tension reaches the design force value, the control system maintains the design tension level for a period of time to ensure the stability of the prestressed steel strand force value, and then the anchoring is carried out manually or automatically.

[0085] After anchoring is completed, displacement sensors and strain gauges continue to monitor and obtain the rebound amount of the prestressed steel strands. The rebound amount is then compared with the designed rebound range to determine whether the anchoring effect is qualified.

[0086] The industrial camera 5 performs a secondary scan of the beam and uses image recognition algorithms to check for defects such as cracks after tensioning, local deformation, or abnormal exposure of steel strands.

[0087] Infrared camera 6 confirms whether the beam temperature has returned to normal and checks for any persistent local overheating areas to eliminate potential hazards such as anchor slippage, friction, or internal damage to the steel strands.

[0088] All monitoring data are automatically generated into a complete tensioning process record, including tension force curves, elongation curves, temperature change curves, visual monitoring images, and comprehensive safety evaluation index change curves for each stage. This data can be directly used for prestressed quality acceptance and construction process traceability analysis.

[0089] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0090] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for intelligent tensioning control of prestressed slab beams based on multi-source sensor information fusion, characterized in that, include: Real-time acquisition of multi-source data during the tensioning process: tension force F(t) is obtained through stress sensor, strain ε(t) is obtained through strain gauge, elongation L(t) of steel strand is obtained through displacement sensor, images of steel strand are obtained through industrial camera and geometric features are extracted to form visual features D(t), and temperature T(t) of anchorage area and steel strand is obtained through infrared camera. The multi-source data is constructed as a state vector X(t) = [F(t), ε(t), L(t), D(t), T(t)]; The parameters in the state vector are standardized to obtain the standardized vector Z(t) = [Z F Z ε Z L Z D Z T ], where the standardized formula is Z i =(X i -μ i ) / σ i , Let μ be the i-th parameter in the state vector X(t). i σ is the historical mean of the parameter. i Standard deviation; A weighted fusion evaluation model is constructed to obtain comprehensive security evaluation indicators: S(t) = ω1Z F +ω2Z ε +ω3Z L +ω4Z D +ω5Z T ,in to Weights for each indicator; Based on the comprehensive safety evaluation index S(t), hierarchical closed-loop control is executed to adjust the tensioning operation. At the same time, single-source anomaly is determined according to the visual feature D(t) and temperature T(t), and an alarm or shutdown is triggered when an anomaly is determined.

2. The method of claim 1, wherein, Extracting visual features D(t) from steel strand images and performing anomaly detection includes: The outline boundary of the steel strand is extracted by the edge detection operator, and the equivalent diameter d(t) is obtained by least squares circle fitting. Computing the rate of change of diameter wherein is the initial diameter; Calculating the lateral offset of the profile center of the steel strand relative to the initial position ; The visual feature D(t) comprises the diameter variation rate and the lateral offset ; when Greater than the preset maximum diameter change rate threshold When it is determined that a slippage has occurred, Greater than the preset maximum offset threshold The system detects eccentric tensioning and triggers an alarm.

3. The method of claim 1, wherein, The determination of single-source anomalies based on temperature T(t) includes: Computing the temperature rise rate where dT / dt is the derivative of temperature with respect to time; When T(t) is greater than a preset maximum temperature threshold or a preset maximum temperature rise rate threshold abnormal friction or local damage occurs, and emergency stop is triggered.

4. The method of claim 1, wherein, Before standardizing the parameters in the state vector, a comparison with theoretical models is also included: Calculated theoretical elongation wherein is the initial length of the steel strand, E is the modulus of elasticity, and A is the cross-sectional area. The deviation of the actual elongation L(t) from the theoretical elongation is incorporated into the standardization process.

5. The method of claim 1, wherein, The hierarchical closed-loop control includes: Set multiple threshold levels: α < β < γ; When S(t) < α, maintain the normal tensioning rate; when α ≤ S(t) < β, reduce the tensioning rate; when β ≤ S(t) < γ, pause tensioning and perform a check; when S(t) ≥ γ, immediately stop and lock the machine. The tensioning rate v(t) is adjusted by the following formula: ,in The initial tensioning rate is set, and k is the adjustment coefficient.

6. The method of claim 1, wherein, It also includes data-driven self-learning steps: Record historical data for each tensioning process to form a sample set Ω={(X(t),S(t))}; updating the weight parameters using the sample set Ω to and threshold values α, β, γ in hierarchical closed-loop control to achieve adaptive optimization under different working conditions.

7. A prestressed plate girder intelligent tensioning system based on multi-source sensing information fusion, for implementing the method according to any one of claims 1 to 6, characterized in that, include: The pedestal has a fixed end seat and a tensioning end seat at each end; An electric hydraulic jack and a movable crossbeam are sequentially arranged on the outside of the tensioning end seat; Prestressed steel bars extend axially along the platform, with one end anchored to the fixed end seat and the other end passing through the tensioning end seat and the movable crossbeam before being connected to the electric hydraulic jack. A stress sensor is installed in the force transmission path at the tension end of the prestressed steel bar to collect the tension force F(t); A strain sensor is arranged on the prestressed steel bar to collect strain ε(t); A displacement sensor is installed at the movable crossbeam to collect the elongation L(t) of the steel strand; An industrial camera, mounted on the side of the pedestal, is used to acquire images of prestressed steel bars; An infrared camera, positioned above the tensioning end, is used to collect temperature T(t); The data acquisition and fusion module is electrically connected to the stress sensor, strain sensor, displacement sensor, industrial camera and infrared camera, and is configured to perform state vector construction, standardization processing and weighted fusion evaluation as described in claim 1 to obtain a comprehensive safety evaluation index S(t), and to determine single-source anomalies based on visual features extracted from the images of the industrial camera and the temperature collected by the infrared camera. The closed-loop tensioning control module is connected to the data acquisition and fusion module and the electro-hydraulic jack, and is used to control the action of the electro-hydraulic jack according to S(t) and the anomaly judgment result.

8. The system of claim 7, wherein, The industrial camera and the data acquisition and fusion module are configured as follows: The equivalent diameter d(t) is obtained from the image through edge detection and least-squares circle fitting; a rate of change of diameter and a lateral offset ; When slip is determined, when eccentric tension is determined, and an alarm is triggered. in The initial diameter, To preset the maximum diameter change rate threshold, This is the preset maximum offset threshold.

9. The system of claim 7, wherein, The infrared camera and the data acquisition and fusion module are configured as follows: Computing the temperature rise rate ; When or a friction anomaly or local damage is determined, triggering an emergency stop. wherein is a preset maximum temperature threshold, is a preset maximum temperature rise rate threshold.

10. The system of claim 7, wherein, The data acquisition and fusion module also includes a historical database and a self-learning unit. The self-learning unit is used to store a sample set Ω={(X(t),S(t))} of the historical tensioning process and to update the weights in the weighted fusion using the sample set Ω. to And the thresholds α, β, and γ in hierarchical closed-loop control.

Images