A small cross-section prestressed pile-plate assembly type node working state monitoring system and method based on artificial intelligence

Abstract

This invention provides an artificial intelligence-based monitoring system and method for the working status of prestressed pile-slab prefabricated joints with small cross-sections, relating to the field of structural monitoring technology. The system includes: a sensor module for real-time acquisition of working status data of the prestressed pile-slab prefabricated joint under test; a data processing and transmission module connected to the sensor module for preprocessing and transmitting the working status data; and a digital twin and AI analysis engine module connected to the data processing and transmission module for analyzing the preprocessed working status data to obtain corresponding health analysis results. This invention utilizes artificial intelligence algorithms to intelligently identify the complex loads borne by the joint, accurately diagnose early damage, and predict the remaining lifespan, thereby issuing effective early warnings before safety hazards lead to accidents.

Description

Technical Field

[0001] This invention relates to the field of structural monitoring technology, specifically to a working status monitoring system and method for prestressed square piles with cross-sectional side lengths less than or equal to 400 mm. By integrating digital twins and artificial intelligence algorithms, the system enables real-time load identification, damage diagnosis, remaining life prediction, and graded early warning of the nodes under complex combined loads (tension, bending, and shear). Background Technology

[0002] Currently, prestressed pile-slab prefabricated joints are a novel structural form with broad application prospects in waterway revetments, wharves, and other water transport engineering projects, especially for small-section prestressed square piles with a side length ≤400mm. These joints achieve a rapid and environmentally friendly connection between the pile and the base slab through components such as embedded sleeves, anchoring steel bars, and prestressed main reinforcement, solving the problems of low efficiency and environmental pollution associated with traditional pile head breaking processes. However, during long-term service, these joints are subjected to complex combined loads (tension, bending, and shear combined effects) generated by ship impacts, water flow changes, and uneven foundation settlement. The evolution of their mechanical properties and their safety status are core issues of concern to the engineering community.

[0003] Currently, safety assessments of such nodes mainly rely on: (1) indoor tests and finite element simulations during the design phase. This method is a "one-time" verification and cannot reflect the individual differences and performance evolution of thousands of nodes during long-term service in actual engineering projects. (2) regular manual inspections during the operation phase. This method is highly subjective, inefficient, and has a long cycle. It also cannot detect early microscopic damage inside the structure. Often, when problems are discovered, the damage has already developed to a serious stage, and the repair cost is high. (3) Although a few studies have used sensors such as strain gauges in the tests, they are limited to short-term data acquisition and have not formed a complete system that is oriented towards the entire life cycle and can perform intelligent diagnosis and early warning.

[0004] Of particular concern is the lack of sensor deployment schemes for specific structures of small-section prestressed pile-slab joints (such as end plates, embedded sleeves, anchoring reinforcements, and other key force-transmitting components) in existing monitoring systems. Furthermore, early warning thresholds are mostly based on experience or general specifications, without sufficient verification through full-scale physical model tests and numerical simulations. For example, it is known in the field that the tensile ultimate bearing capacity of this joint in full-scale tests is 2119.5 kN, the flexural ultimate bearing capacity is 276.9 kN·m, and the shear ultimate bearing capacity is 680 kN (derived from full-scale physical model tests), but these key thresholds have not yet been systematically incorporated into the monitoring and early warning system.

[0005] Therefore, there is an urgent need in this field for a system and method specifically designed for prestressed pile-slab assembly nodes with small cross-sections, capable of real-time online monitoring, and intelligent diagnosis and graded early warning based on test thresholds, in order to realize the transformation of operation and maintenance mode from "passive maintenance" to "proactive early warning" and ensure the long-term safe operation of critical infrastructure. Summary of the Invention

[0006] In view of this, the present invention provides a monitoring system and method for the working status of prestressed pile-slab assembly nodes with small cross-section based on artificial intelligence. The system uses artificial intelligence algorithms to intelligently identify the complex loads borne by the nodes, accurately diagnose early damage, and predict the remaining life, thereby issuing an effective early warning before safety hazards lead to accidents.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: An artificial intelligence-based monitoring system for the working status of prestressed pile-slab assembly joints with small cross-sections includes: The sensor module is installed on the prestressed pile-slab assembly node to be tested and is used to collect the working status data of the prestressed pile-slab assembly node in real time. Specifically: the sensor module is installed on the prestressed prefabricated node to be tested, and is used to collect the working status data of the prestressed prefabricated node in real time; the prestressed prefabricated node to be tested is a prefabricated connection node between a small cross-section prestressed square pile with a side length of less than or equal to 400mm and a base plate, and its structure includes end steel plates, embedded sleeves, upper anchoring bars, lower anchoring bars and prestressed main bars; the sensor module includes: multiple strain sensors, respectively arranged on the upper anchoring bars, the core area concrete of the node and the prestressed main bars; multiple stress sensors, arranged in the key stress concentration area where the pile top steel plate and the sleeve are connected; and at least one triaxial accelerometer, installed on a rigid structure near the node; A data processing and transmission module, which is connected to the sensor module, is used to preprocess and transmit the working status data; The digital twin and AI analysis engine module is connected to the data processing and transmission module and is used to analyze the preprocessed working status data to obtain the corresponding health analysis results; wherein the small cross-section prestressed pile-slab prefabricated node refers to the prefabricated node between the prestressed square pile and the base plate with a cross-sectional side length of no more than 400mm.

[0008] Preferably, the digital twin and AI analysis engine module includes: A node digital twin model library construction unit is connected to the data processing and transmission module. It is used to establish a parametric finite element model corresponding to the prestressed pile-slab assembly node to be tested, and to perform simulation analysis in combination with the working state data to obtain the corresponding mechanical response analysis results. The AI ​​load identification unit is connected to the node digital twin model library construction unit. It is used to perform load analysis based on the mechanical response analysis results and the preprocessed working state data, and to identify the axial tensile force, bending moment and shear force on the node.

[0009] Preferably, the digital twin and AI analysis engine module further includes: The AI ​​damage diagnosis unit is connected to the data processing and transmission module and the node digital twin model library construction unit. It is used to identify damage on the preprocessed working status data and the mechanical response analysis results, that is, to identify the width of concrete cracks, steel bar slippage and prestress relaxation at the nodes. The AI ​​remaining lifetime prediction unit is connected to the data processing and transmission module and the node digital twin model library construction unit, and is used to predict the remaining lifetime based on the mechanical response analysis results and the preprocessed working status data.

[0010] Preferred options also include: The early warning and visualization interaction module is connected to the digital twin and AI analysis engine module. It is used to receive health analysis results and issue early warning information based on the health analysis results and multiple actual classification thresholds.

[0011] Preferably, the early warning and visualization interaction module includes: The early warning analysis unit is connected to the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit. It is used to predetermine multiple actual grading thresholds and make a comprehensive decision based on the results output by the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit according to the multiple actual grading thresholds to obtain the corresponding comprehensive analysis results. A push unit, which is connected to the early warning analysis unit, is used to compare multiple actual classification thresholds with the comprehensive analysis results, and push classified early warning information based on the comparison results. A visual interactive display unit is connected to the push unit, the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit, and is used to display early warning information and the output results of the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit in real time.

[0012] Preferably, the early warning information may include tiered early warnings, specifically including: Level I concern threshold: Concrete crack width reaches 0.05mm or load duration reaches 80%; Level II warning threshold: Concrete crack width reaches 0.1mm or load duration reaches 90% or the remaining life prediction shows an accelerated downward trend; Level III alarm threshold: Concrete crack width reaches 0.2mm or the load exceeds the design value (any one of the following: tensile strength 2119.5kN, flexural strength 276.9kN·m, shear strength 680kN) or the remaining service life is predicted to be less than 5 years.

[0013] Preferably, the specific processing steps for obtaining multiple actual classification thresholds include: By conducting physical model tests and numerical simulation tests on prestressed prefabricated nodes, the corresponding mechanical performance test results of the nodes were obtained. Based on the mechanical performance test results of the nodes, several actual classification thresholds were determined, including the bearing capacity threshold of the nodes under axial tension, bending moment, shear force and their combination, load-strain response curves, damage evolution laws and failure modes. Mechanical response models of the nodes under elastic, cracking and ultimate states were constructed, and these models serve as the fundamental basis for load identification, damage diagnosis and life prediction.

[0014] This invention also utilizes the above system to provide an artificial intelligence-based method for monitoring the working status of prestressed prefabricated joints, comprising the following steps: Step 1: Sensor deployment, data acquisition and preprocessing The system collects real-time operational status data of the prestressed precast joint under test and preprocesses the data. Specifically, fiber optic strain gauges (6 measuring points) are attached to the upper anchoring steel bars of the prestressed pile-slab precast joint with small cross-section under test; fiber optic strain gauges (12 measuring points) are pre-embedded in the concrete of the core area of ​​the joint; strain gauges (2 measuring points) are set on the prestressed main reinforcement; a miniature MEMS stress sensor (6 measuring points) is welded to the connection area between the pile top steel plate and the sleeve; and a triaxial accelerometer (1 unit, providing 3 acceleration channels) is installed near the bottom plate, for a total of 29 physical quantity measuring points. Strain, stress, and acceleration data are collected in real-time at a sampling frequency of not less than 100Hz. The collected raw data is filtered, denoised, temperature compensated, and normalized to form a 29-dimensional time-series feature vector, which is then transmitted to the analysis engine via wired or wireless means. Step 2: Construct the parametric model corresponding to the prestressed prefabricated node to be tested using the node digital twin model library construction unit, and analyze it in combination with the preprocessed working state data to obtain the corresponding mechanical response analysis results. Specifically, based on the actual geometric dimensions, material parameters and boundary conditions of the node to be tested, establish a parametric finite element model in ABAQUS software, and use the real-time acquired strain and stress data as boundary conditions to perform simulation analysis, and output the full-field stress and strain distribution and damage state of the node under the current load. Step 3: The AI ​​load recognition unit, AI damage diagnosis unit, and AI remaining life prediction unit are combined with the mechanical response analysis results and the preprocessed working state data to obtain the corresponding load analysis results, damage diagnosis results, and remaining life prediction results. Specifically, the preprocessed 29-dimensional feature vector is input into the pre-trained AI load recognition model, which outputs the axial tensile force (kN), bending moment (kN·m), and shear force (kN) of the current node. The recognition error is controlled within 5% of the full scale for axial force, 8% of the full scale for bending moment, and 10% of the full scale for shear force. The mechanical response analysis results and load recognition results are input into the AI ​​damage diagnosis model, which outputs the crack width (mm), crack location (positioning error ≤ ±10cm), steel bar slippage state, and prestress relaxation degree of the node concrete. The damage classification accuracy is ≥95%. The remaining fatigue life (years) of the node is predicted using a deep survival analysis model, with a prediction deviation of ≤15%. Step 4: Perform a fusion analysis on the load analysis results, damage diagnosis results, and remaining life prediction results to obtain the corresponding comprehensive analysis results and multiple actual classification thresholds; A tiered early warning system is implemented by combining comprehensive analysis results with comparisons of multiple actual grading thresholds. Specifically, the output load value, crack width, and remaining life are compared with a preset threshold system to trigger a three-level early warning. If the crack width is ≥0.05mm but <0.1mm, or the load duration is ≥80% but <90%, Level I attention is triggered: a prompt will be displayed on the monitoring platform interface, and a periodic report will be generated without immediate on-site intervention; If the crack width is ≥0.1mm but <0.2mm, or the load duration is ≥90%, or the remaining life prediction shows an accelerated downward trend, a Level II warning will be triggered: the location will be automatically marked, a prompt message will be pushed to the engineer's mobile terminal, and an on-site inspection will be recommended. If the crack width is ≥0.2mm, or the load exceeds the design value (any one of the following: tensile strength ≥2119.5kN, bending strength ≥276.9kN·m, shear strength ≥680kN), or the remaining life prediction is <5 years, a Level III alarm will be triggered: multiple alarms will be issued in parallel via platform pop-up, SMS and telephone, and the emergency plan will be activated. Step 5: Report Generation Automatically generate monitoring reports that include load time history, damage evolution diagrams, and remaining life prediction curves.

[0015] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a monitoring system and method for the working status of small-section prestressed pile-slab prefabricated joints based on artificial intelligence, which has the following beneficial effects: (1) For the first time, a multi-type sensor layout scheme with 29 measuring points was proposed for the key force transmission path of prestressed pile-slab prefabricated joints with small cross-sections (≤400mm). Real-time and forward-looking: It transforms periodic and passive inspections into continuous and proactive monitoring, which can capture the emergence of instantaneous abnormal loads and damage, providing a data foundation for predictive maintenance.

[0016] (2) Based on full-scale physical model tests (no node failure at tensile strength of 1400kN, no node failure at bending strength of 276.9kN·m, and no node failure at shear strength of 680kN) and finite element numerical simulations (tensile strength of 2350kN, bending strength of 315.7kN·m, and shear strength of 1251kN), scientific multi-level early warning thresholds were determined, and the depth and accuracy of diagnosis were improved. By integrating physical-driven (digital twin) and data-driven (AI model) methods, internal and early microscopic damage that cannot be detected by manual inspection and traditional threshold alarms can be diagnosed, which greatly improves the depth and accuracy of diagnosis.

[0017] (3) Intelligent decision support: It provides a digital twin and AI multi-task model to realize integrated intelligent analysis of load identification, damage diagnosis and life prediction. The system not only informs whether a problem has occurred, but also accurately answers the nature of the problem, the location of the problem, the severity of the problem and the remaining service life of the component, providing quantitative and scientific decision-making basis for operation and maintenance management.

[0018] (4) Full life cycle management: The system covers the entire process from the construction and operation of the node to its decommissioning. It can accumulate valuable full life cycle performance data, which is of great value for optimizing the design, construction and operation and maintenance of similar nodes in the future.

[0019] (5) The AI ​​load identification model, damage diagnosis model and remaining life prediction model constructed in this invention have all been trained and verified by large-scale finite element simulation and full-scale test data. Their load inversion error is small, the damage identification accuracy is high (≥95%), and the life prediction deviation is controllable (≤15%), which ensures the scientific nature and reliability of system diagnosis and early warning.

[0020] (6) This invention defines a three-level early warning mechanism (Level I attention, Level II early warning, and Level III alarm) based on the fusion of multiple parameters such as concrete crack width, load over-limit ratio and remaining life. The three-level early warning mechanism is linked to specific test thresholds to achieve tiered safety management. It also clarifies the quantitative thresholds and qualitative response procedures for each level, realizing tiered and intelligent safety management from risk warning to emergency response. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0022] Figure 1 A structural principle block diagram of an artificial intelligence-based prestressed pile-slab assembly node working status monitoring system provided by the present invention; Figure 2 This is a structural schematic diagram of a prestressed prefabricated node provided in an embodiment of the present invention; Figure 3 This is a typical layout diagram of the sensor module provided in an embodiment of the present invention. Detailed Implementation

[0023] 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.

[0024] See Figure 1 As shown, this embodiment of the invention discloses a monitoring system for the working status of small-section prestressed pile-slab prefabricated joints based on artificial intelligence, comprising: The sensor module is installed on the prestressed pile-slab assembly node to be tested and is used to collect the working status data of the prestressed pile-slab assembly node in real time. A data processing and transmission module, which is connected to the sensor module, is used to preprocess and transmit the working status data; The digital twin and AI analysis engine module is connected to the data processing and transmission module and is used to analyze the preprocessed working status data to obtain corresponding health analysis results.

[0025] Specifically, the structure of the prestressed pile-slab prefabricated joint to be tested can be found in [reference needed]. Figure 2 As shown, the sensor module is deployed on the critical force transmission path of the prestressed pile-slab assembly node to be tested, and is used to collect the working status data of the prestressed pile-slab assembly node in real time. The specific deployment method is as follows: Fiber grating strain gauges are attached to the upper anchoring steel bars, micro MEMS stress sensors are welded to the surface of the pile top steel plate, fiber grating strain gauges are pre-embedded in the concrete of the joint cast-in-place zone, and triaxial accelerometers are installed near the bottom plate.

[0026] To fully cover the key force transmission paths of the nodes, typical sensor deployment methods can be found in [reference needed]. Figure 3 As shown, this embodiment of the invention determines the key force transmission path and mechanical response characteristics of the prestressed pile-slab joint based on physical model tests and numerical simulation experiments. A total of 29 physical measurement points are arranged on a single node, forming a 29-dimensional time-series input feature vector for each inference used in subsequent analyses, specifically including: (1) 20 strain monitoring points: distributed in the upper anchoring steel bars (6 points), the core concrete of the node (12 points) and the prestressed main reinforcement (2 points) respectively, to capture the strain field of the whole domain caused by tension, compression, bending and shear. Among them, the monitoring points of the upper anchoring steel bars are used to monitor the stress change of the steel bars caused by tension and bending moment. (2) Six stress monitoring points: set up in the key stress concentration area where the steel plate at the top of the pile connects to the sleeve, directly monitor the interface contact stress, and are used to monitor the stress concentration in the core connection area of ​​the node. (3) Three dynamic acceleration channels: derived from a three-axis accelerometer.

[0027] The in-depth function and installation of the triaxial accelerometer are as follows: The triaxial accelerometer is fixedly installed on a rigid structure near the node. Its core functions are: to capture transient impact events, such as ship collisions and abnormal wave forces, and to provide dynamic load information that may be missed by strain sensors, for monitoring the node vibration response and assisting in the identification of impact loads; to provide a structural dynamic fingerprint, whose output triaxial acceleration time history data is used to identify load types and changes in local structural stiffness; and to assist in modal analysis and state verification: by long-term monitoring of vibration response, the changes in structural frequency and damping ratio are analyzed, providing a verification basis for the overall stiffness degradation of the node that is independent of strain monitoring.

[0028] All 29 channels of data from the sensors are synchronously acquired, processed, and digitized by the data acquisition and transmission module, ultimately forming a real-time data stream that is uploaded to the cloud platform for subsequent processing by the analysis engine.

[0029] In one specific embodiment, the digital twin and AI analysis engine module includes: A node digital twin model library construction unit, connected to the data processing and transmission module, is used to establish a parametric finite element model corresponding to the prestressed pile-slab assembly node to be tested, and to perform simulation analysis in conjunction with the working state data to obtain the corresponding mechanical response analysis results. The node digital twin model library construction unit adopts modeling parameters consistent with experimental data, and uses ABAQUS software to establish a parametric finite element model for each node. The model considers material nonlinearity (elastic-plastic damage of C30 / C60 concrete, ideal elastic-plasticity of steel reinforcement) and complex contact relationships. The AI ​​load identification unit is connected to the node digital twin model library construction unit and is used to perform load analysis based on the mechanical response analysis results and the preprocessed working state data. The load analysis index is the load duration rate.

[0030] In one specific embodiment, the digital twin and AI analysis engine module further includes: The AI ​​damage diagnosis unit is connected to the data processing and transmission module and the node digital twin model library construction unit. It is used to identify damage from the preprocessed working status data and the mechanical response analysis results. The index for evaluating damage is concrete microcracks. The AI ​​remaining lifetime prediction unit is connected to the data processing and transmission module and the node digital twin model library construction unit, and is used to predict the remaining lifetime based on the mechanical response analysis results and the preprocessed working status data.

[0031] Specifically, the AI ​​load recognition unit, AI damage diagnosis unit, and AI remaining life prediction unit all have built-in corresponding AI load recognition models and AI damage diagnosis models. The corresponding model training and testing processes are as follows: Training data comes from two main sources: 1) Finite element simulation data: Based on the validated digital twin model, a total of 3,000 sets of combined tensile, bending, and shear loads covering elastic to failure states are applied, and the global strain, stress, and acceleration response are output to form a simulation dataset.

[0032] 2) Indoor full-scale test data: Based on 12 sets of nodal tests with different loading paths (monotonous, cyclic, fatigue), synchronous data were collected through fiber optic gratings, strain gauges and triaxial accelerometers to form a test dataset.

[0033] The input features of the above data are uniformly defined as a 29-dimensional time-series vector, including 20 strain measurement points, 6 stress measurement points, and 3-axis acceleration data. Output labels are defined for different models: the load identification model corresponds to the true values ​​of axial force, bending moment, and shear force; the damage diagnosis model corresponds to classification labels such as no damage, concrete microcracks, steel bar slippage, and prestress relaxation. The dataset is divided into training, validation, and test sets in an 8:1:1 ratio.

[0034] 3) The model's performance has been rigorously validated: After the above training and testing, the average absolute error of the AI ​​load recognition model on the test set is: axial force ≤ 5% of full scale, bending moment ≤ 8% of full scale, and shear force ≤ 10% of full scale; the AI ​​damage diagnosis model has a classification accuracy of ≥ 95% for early damage and a damage spatial positioning error ≤ ± 10 cm; the prediction deviation (root mean square error) of the remaining life prediction model is controlled within 15% relative to the actual fatigue life.

[0035] In one specific embodiment, it also includes: The early warning and visualization interaction module is connected to the digital twin and AI analysis engine module, and is used to receive health analysis results and issue early warning information based on the health analysis results.

[0036] In one specific embodiment, the early warning and visualization interaction module includes: An early warning analysis unit, connected to the AI ​​load identification unit, AI damage diagnosis unit, and AI remaining life prediction unit, is used to predetermine multiple actual grading thresholds and make comprehensive decisions based on the results output by the AI ​​load identification unit, AI damage diagnosis unit, and AI remaining life prediction unit to obtain corresponding comprehensive analysis results. The AI ​​load identification unit can be implemented using a Transformer architecture, the AI ​​damage diagnosis unit can be implemented using a graph neural network, and the AI ​​remaining life prediction unit can be implemented using a deep survival analysis model. A push unit, which is connected to the early warning analysis unit, is used to compare multiple actual classification thresholds with the comprehensive analysis results, and push classified early warning information based on the comparison results. A visual interactive display unit is connected to the push unit, the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit, and is used to display early warning information and the output results of the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit in real time.

[0037] Specifically, the implementation process of the early warning analysis unit and the push unit includes: (1) Early warning analysis unit: The early warning mechanism is triggered when the load duration reaches 80% or the concrete microcracks are between 0.05mm and 0.1mm. When the load sustainment rate reaches 90%, or the concrete microcracks are between 0.1mm and 0.2mm, or the remaining life prediction shows an accelerated downward trend, the system can send a prompt message to the staff and perform relevant inspection tasks when any of the above conditions are met. When the concrete microcrack width is ≥0.2mm, or the load exceeds the design value, or the remaining life prediction is less than 5 years (health level <70%), the system will issue multiple alarms in parallel through platform pop-ups, SMS and telephone calls when any of the above conditions are met, and immediately activate the emergency plan. (2) Push unit: When the load duration rate, concrete microcracks and remaining life all meet the above threshold requirements, the above load duration rate, concrete microcracks and remaining life are integrated and analyzed to output the corresponding comprehensive analysis results and the corresponding multiple actual grading thresholds, wherein the multiple actual grading thresholds may specifically include the first actual grading threshold and the second actual grading threshold. When the comprehensive analysis result is less than the first actual grading threshold, no warning push will be sent. Instead, a prompt will be displayed on the platform interface and a periodic report will be generated without immediate on-site intervention. When the comprehensive analysis result is greater than or equal to the first actual grading threshold but less than the second actual grading threshold, the area will be automatically marked and staff will be prompted to conduct an on-site inspection. When the comprehensive analysis result is greater than or equal to the second actual grading threshold, multiple alarms will be issued in parallel via platform pop-ups, SMS, and telephone, and the emergency plan will be activated immediately.

[0038] In a specific embodiment, the specific processing procedure for obtaining multiple actual classification thresholds includes: By conducting physical model tests and numerical simulation tests on prestressed prefabricated nodes, the corresponding mechanical performance test results of the nodes were obtained. Based on the mechanical performance test results of the nodes, several actual classification thresholds were determined, including the bearing capacity threshold of the nodes under axial tension, bending moment, shear force and their combination, load-strain response curves, damage evolution laws and failure modes. Mechanical response models of the nodes under elastic, cracking and ultimate states were constructed, and these models serve as the fundamental basis for load identification, damage diagnosis and life prediction.

[0039] This invention also provides a method for monitoring the working status of small-section prestressed pile-slab prefabricated joints based on artificial intelligence as described in any of the above embodiments, comprising the following steps: Step 1: Complete the sensor deployment, collect the working status data of the prestressed prefabricated node to be tested in real time, and preprocess the working status data; Step 2: Construct the parameter model corresponding to the prestressed prefabricated node to be tested using the node digital twin model library construction unit, and analyze it in combination with the preprocessed working state data to obtain the corresponding mechanical response analysis results; Step 3: The AI ​​load identification unit, AI damage diagnosis unit, and AI remaining life prediction unit are combined with the mechanical response analysis results and the preprocessed working state data to obtain the corresponding load analysis results, damage diagnosis results, and remaining life prediction results. Step 4: Perform a fusion analysis on the load analysis results, damage diagnosis results, and remaining life prediction results to obtain the corresponding comprehensive analysis results and multiple actual grading thresholds. Combine the comprehensive analysis results with the comparison results of multiple actual grading thresholds to push graded early warnings. Step 5: Based on the results obtained in Steps 2-4, automatically generate a monitoring report containing load time history, damage evolution diagram, and remaining life prediction curve.

[0040] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0041] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An artificial intelligence-based small cross-section prestressed pile-plate fabricated node working condition monitoring system, characterized in that, include: The sensor module is installed on the prestressed pile-slab assembly node to be tested and is used to collect the working status data of the prestressed pile-slab assembly node in real time. A data processing and transmission module, which is connected to the sensor module, is used to preprocess and transmit the working status data; The digital twin and AI analysis engine module is connected to the data processing and transmission module and is used to analyze the preprocessed working status data to obtain corresponding health analysis results.

2. The working state monitoring system for small cross-section prestressed pile-plate fabricated node based on artificial intelligence according to claim 1, characterized in that, The digital twin and AI analysis engine module includes: A node digital twin model library construction unit is connected to the data processing and transmission module. It is used to establish a parametric finite element model corresponding to the prestressed pile-slab assembly node to be tested, and to perform simulation analysis in combination with the working state data to obtain the corresponding mechanical response analysis results. The AI ​​load identification unit is connected to the node digital twin model library construction unit and is used to perform load analysis based on the mechanical response analysis results and the preprocessed working state data.

3. The working state monitoring system for small cross-section prestressed pile-plate fabricated node based on artificial intelligence according to claim 2, characterized in that, The digital twin and AI analysis engine module also includes: The AI ​​damage diagnosis unit is connected to the data processing and transmission module and the node digital twin model library construction unit, and is used to identify damage from the preprocessed working status data and the mechanical response analysis results. The AI ​​remaining lifetime prediction unit is connected to the data processing and transmission module and the node digital twin model library construction unit, and is used to predict the remaining lifetime based on the mechanical response analysis results and the preprocessed working status data.

4. The working state monitoring system for small cross-section prestressed pile-plate fabricated node based on artificial intelligence according to claim 3, characterized in that, Also includes: The early warning and visualization interaction module is connected to the digital twin and AI analysis engine module, and is used to receive health analysis results and issue early warning information based on the health analysis results.

5. The working state monitoring system for small cross-section prestressed pile-plate fabricated node based on artificial intelligence according to claim 4, characterized in that, The early warning and visualization interaction module includes: The early warning analysis unit is connected to the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit. It is used to predetermine multiple actual grading thresholds and make a comprehensive decision based on the results output by the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit according to the multiple actual grading thresholds to obtain the corresponding comprehensive analysis results. A push unit, which is connected to the early warning analysis unit, is used to compare multiple actual classification thresholds with the comprehensive analysis results, and push classified early warning information based on the comparison results. A visual interactive display unit is connected to the push unit, the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit, and is used to display early warning information and the output results of the AI ​​load identification unit, the AI ​​damage diagnosis unit, and the AI ​​remaining life prediction unit in real time.

6. The working state monitoring system for small cross-section prestressed pile-plate fabricated node based on artificial intelligence according to claim 5, characterized in that, The specific processing steps to obtain multiple actual classification thresholds include: By conducting physical model tests and numerical simulation tests on prestressed prefabricated nodes, the corresponding test results of node mechanical properties were obtained, and multiple actual grading thresholds were determined based on the test results of node mechanical properties.

7. A method for monitoring the working status of prestressed pile-slab assembly joints based on artificial intelligence as described in any one of claims 1-5, characterized in that, Includes the following steps: Step 1: Complete the sensor deployment, collect the working status data of the prestressed prefabricated node to be tested in real time, and preprocess the working status data; Step 2: Construct the parameter model corresponding to the prestressed prefabricated node to be tested using the node digital twin model library construction unit, and analyze it in combination with the preprocessed working state data to obtain the corresponding mechanical response analysis results; Step 3: The AI ​​load identification unit, AI damage diagnosis unit, and AI remaining life prediction unit are combined with the mechanical response analysis results and the preprocessed working state data to obtain the corresponding load analysis results, damage diagnosis results, and remaining life prediction results. Step 4: Perform a fusion analysis on the load analysis results, damage diagnosis results, and remaining life prediction results to obtain the corresponding comprehensive analysis results and multiple actual grading thresholds. Combine the comprehensive analysis results with the comparison results of multiple actual grading thresholds to push graded early warnings. Step 5: Based on the results obtained in Steps 2-4, automatically generate a monitoring report containing load time history, damage evolution diagram, and remaining life prediction curve.

Images