Multi-modal fusion and quantum optimization concrete filled steel tube arch bridge detection system and method

Abstract

The concrete-filled steel tube arch bridge detection system and method based on multi-modal fusion and quantum optimization are used for bridge structure health monitoring. The system comprises a multi-modal data acquisition module, a quantum optimization data fusion module, a defect self-growth simulation damage prediction module, a low-power-consumption intelligent sensor network and a quantum safety remote monitoring module. The multi-modal data acquisition module is fused with an ultrasonic radar, a millimeter wave radar, a hyperspectral image, a LiDAR and a wireless sensor, and bridge surface and internal damage detection is realized; the data weight is dynamically adjusted by adopting a quantum optimization algorithm, and the fusion precision is improved; in combination with a graph neural network and deformable grid calculation, crack propagation and fatigue damage development are predicted, the power consumption of the sensor is reduced through an energy collection technology, and the safety of remote monitoring data is ensured by adopting a quantum safety remote monitoring module. Compared with the prior art, the method improves the detection precision, prediction capability and data safety, and is suitable for health monitoring of various bridges and infrastructures.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge structural health monitoring, and particularly relates to a detection system and method for a concrete-filled steel tube arch bridge with multimodal fusion and quantum optimization. Background Art

[0002] Due to its advantages such as high bearing capacity, excellent seismic performance, and convenient construction, concrete-filled steel tube arch bridges are widely used in the construction of highway, railway, and cross-river and cross-sea bridges. However, during long-term service, concrete-filled steel tube arch bridges will be affected by factors such as environmental erosion, temperature changes, and fatigue loads, which may lead to structural problems such as steel tube corrosion, concrete spalling, weld cracking, and fatigue damage. In order to ensure bridge safety, a series of bridge health monitoring technologies have been developed in the industry, mainly including the following categories:

[0003] 1. Traditional manual inspection method

[0004] Traditional bridge inspection mainly relies on manual inspection, and combines simple tools (such as crack gauges, rebound hammers, ultrasonic detectors, etc.) to regularly check the surface damage of the bridge. Although this method can find obvious surface damage to a certain extent, it has the following disadvantages:

[0005] Low efficiency: The manual inspection period is long, it is difficult to achieve high-frequency monitoring, and potential safety hazards cannot be discovered in time.

[0006] Affected by human factors: The inspection results depend on the experience of the inspectors, the misjudgment rate is relatively high, and there may be significant differences in the results between different inspectors.

[0007] Unable to detect internal damage: Manual inspection cannot detect structural defects such as weld fatigue and corrosion expansion inside the steel tube, and it is difficult to comprehensively evaluate the health status of the bridge.

[0008] 2. Structural sensor monitoring technology

[0009] At present, some bridges use fixed sensors for long-term monitoring. For example, strain gauges, accelerometers, and displacement sensors are used for bridge vibration and deformation monitoring; fiber Bragg grating (FBG) sensors monitor strain and temperature changes; ultrasonic / impact echo detection is used to identify internal cracks or debonding in concrete; infrared thermal imaging is used to detect damage caused by temperature anomalies. However, this method also has the following problems:

[0010] Single data: A single sensor can only provide partial structural information, it is difficult to form an overall health assessment, and it cannot comprehensively reflect the health status of the bridge.

[0011] Great influence of environmental interference: Environmental factors such as temperature, humidity, and noise may affect the accuracy of the sensors, resulting in false alarms or missed alarms, and reducing the reliability of the monitoring.

[0012] High sensor maintenance cost: Sensors require external power supply, consume a large amount of power during long-term operation, are prone to damage, and have high replacement costs, increasing the maintenance difficulty and cost.

[0013] 3. UAV + AI Image Analysis and Detection

[0014] In recent years, unmanned aerial vehicles (UAVs) combined with computer vision and deep learning technologies have been used for the automatic identification of damages such as cracks, spalling, and corrosion on the surface of bridges. For example, UAVs are equipped with high-definition cameras to obtain bridge images, and AI algorithms such as YOLO and ResNet are combined for crack detection; Light Detection and Ranging (LiDAR) combined with point cloud analysis technology is used to obtain the deformation of bridges. Although this method has made significant progress in surface damage detection, it still has the following limitations:

[0015] Limited to surface defect detection: AI vision technology can only identify external cracks and cannot detect hidden damages such as weld fatigue and corrosion inside steel pipes, making it impossible to comprehensively evaluate the health status of bridges.

[0016] Affected by environmental light: Changes in rain, fog, and lighting conditions may lead to a decrease in image recognition accuracy, affecting the accuracy and reliability of detection.

[0017] 4. Finite Element Analysis (FEA) + Damage Prediction

[0018] Finite Element Analysis (FEA) is used to simulate the response of bridge structures. Combining historical data can predict damage trends, such as analyzing the stress distribution of bridges under load; establishing a fatigue damage model in combination with historical data to predict the service life of bridges. However, this method also has the following problems:

[0019] Dependence on a large number of known parameters: The FEA model requires accurate material parameters and load conditions of the bridge. However, in actual engineering, material aging and construction errors lead to large parameter errors, affecting the prediction accuracy.

[0020] Inability to update the model in real time: Damage prediction requires long-term data accumulation, and the model is not flexible to adjust, making it difficult to adapt to the complex changes during the long-term service of bridges and unable to reflect the health status of bridges in real time.

[0021] In summary, the existing bridge health monitoring technologies have the following main problems:

[0022] Insufficient data fusion ability: The traditional sensor data fusion in the existing technical solutions uses fixed weights or simple weighted averages, cannot adapt to environmental changes, has limited fusion accuracy, and is difficult to comprehensively evaluate the health status of bridges.

[0023] Internal damage is difficult to detect: In the existing technical solutions, drones + AI vision detection are used, which can only detect surface cracks and cannot obtain hidden damages such as internal welds and corrosion of steel pipes, making it impossible to comprehensively evaluate the health status of bridges.

[0024] Limited damage prediction ability: The existing technical solutions rely on FEA and time series analysis (such as LSTM), which cannot simulate the actual crack propagation and fatigue accumulation processes, resulting in large prediction errors and making it difficult to early warn of potential safety hazards.

[0025] High sensor power consumption: The sensors in the existing technical solutions need to be powered by wire or battery, making remote bridge maintenance difficult, the sensors are easily damaged, and the long-term monitoring cost is high, making it difficult to achieve long-term unattended monitoring.

[0026] Low remote data security: In the existing technical solutions, data transmission uses AES / RSA encryption. Quantum computing may crack the existing encryption algorithms, and the monitoring data faces security risks, making it impossible to ensure the integrity and immutability of the data. Summary of the Invention

[0027] In view of the problems existing in the above-mentioned prior art, the present invention provides a detection system for concrete-filled steel tubular arch bridges with adaptive multimodal fusion and quantum optimization, aiming to realize real-time monitoring, damage identification, life prediction and remote data security management of concrete-filled steel tubular arch bridges by integrating multimodal sensing technology, quantum optimization calculation, artificial intelligence deep learning, intelligent wireless sensor network and quantum secure communication technology, effectively solving the problems existing in the prior art.

[0028] To achieve the above object, the specific scheme of the present invention is as follows:

[0029] The detection system for concrete-filled steel tubular arch bridges with adaptive multimodal fusion and quantum optimization includes the following modules:

[0030] The multimodal data acquisition module is used to obtain damage data on the surface and inside of the bridge, and the damage data includes crack, corrosion, fatigue damage and deformation data;

[0031] The quantum optimization data fusion module is used to fuse the data obtained by the multimodal data acquisition module, and the quantum optimization data fusion module dynamically adjusts the sensor data weights by using a quantum optimization algorithm;

[0032] The defect self-growth simulation damage prediction module is used to analyze the data fused by the quantum optimization data fusion module and predict the development trend of bridge damage;

[0033] The low-power intelligent sensor network is used to reduce the power consumption of the multimodal data acquisition module;

[0034] A quantum-secure encryption remote monitoring module is used to ensure the security and immutability of the data generated by the multimodal data acquisition module and the defect self-growth simulation damage prediction module during remote transmission.

[0035] Further, the multimodal data acquisition module includes:

[0036] A variable-frequency ultrasonic radar for detecting fatigue, debonding, and porosity defects in the internal welds of steel pipes;

[0037] A millimeter-wave radar for evaluating the internal corrosion of steel pipes;

[0038] A hyperspectral imaging sensor for analyzing concrete carbonation and crack propagation;

[0039] A lidar carried by a drone for obtaining three-dimensional point cloud data of the bridge surface;

[0040] Wireless sensors, including strain gauges, accelerometers, displacement sensors, and fiber Bragg gratings, for real-time monitoring of the strain and dynamic response of the bridge;

[0041] A high-definition camera carried by a drone for collecting crack and deformation image data of the bridge surface.

[0042] Further, the quantum-optimized data fusion module includes:

[0043] A quantum-optimized algorithm unit for fusing and optimizing the wireless sensor data in the multimodal data acquisition module using a quantum-optimized algorithm;

[0044] A data weight adjustment unit for constructing a dynamic data weight adjustment model;

[0045] A Bayesian optimization unit for automatically adjusting the data fusion strategy according to environmental conditions such as temperature, humidity, and wind speed;

[0046] An environment adaptive algorithm unit for ensuring that the system can adapt to different types of bridges and environmental conditions.

[0047] Further, the defect self-growth simulation damage prediction module uses a graph neural network and a long short-term memory network. The defect self-growth simulation damage prediction module includes:

[0048] A graph neural network unit for simulating the crack propagation path;

[0049] A deformable grid calculation unit for simulating the fatigue damage development path;

[0050] A finite element analysis unit, which is used to predict the remaining life of a bridge by combining the outputs of a graph neural network unit and a deformable mesh calculation unit as well as historical data; the historical data includes the crack positions, lengths, widths, fatigue damage degrees, propagation paths, strains, displacements, accelerations, environmental conditions, and bridge load data detected in the past;

[0051] A self-supervised learning unit, which is used to optimize the damage prediction model under unlabeled conditions. The unlabeled conditions refer to learning and optimizing through the internal structure and features of data in the case of no explicit labeled damage data.

[0052] Further, the low-power intelligent sensor network includes:

[0053] An energy harvesting unit, which is used to harvest energy from bridge vibrations, solar energy, or temperature difference energy;

[0054] An edge computing unit, which is used to perform data preprocessing at the sensor end, reduce the amount of data transmission, and reduce power consumption;

[0055] A dynamic adaptive sampling unit, which is used to automatically adjust the data acquisition frequency according to the bridge health condition.

[0056] Further, the quantum-secure encryption remote monitoring module includes:

[0057] A quantum key distribution unit, which is used to generate and distribute quantum keys;

[0058] A blockchain evidence storage unit, which is used to store the encrypted monitoring data and verify the integrity of the data;

[0059] A 5G and satellite communication unit, which is used to support the remote transmission of the encrypted monitoring data and expert diagnosis.

[0060] An intelligent detection method for concrete-filled steel tube arch bridges based on adaptive multimodal data fusion and quantum-optimized damage prediction, including the following steps:

[0061] Step 1, Multimodal data acquisition: Use a variable-frequency ultrasonic radar to detect fatigue, debonding, and pore defects inside the steel tube; use a millimeter-wave radar to evaluate the corrosion inside the steel tube; use a hyperspectral imaging sensor to analyze concrete carbonation and crack propagation; use a lidar carried by a drone to obtain the three-dimensional point cloud data of the bridge surface; use wireless sensors to monitor the strain and dynamic response of the bridge in real time; use a high-definition camera carried by a drone to collect the crack and deformation image data of the bridge surface;

[0062] Step 2, Quantum Optimization Data Fusion: Use a quantum optimization algorithm to dynamically adjust the data weights of each wireless sensor in the multi-modal data acquisition module; automatically adjust the data fusion strategy according to environmental conditions through the Bayesian optimization method to generate an optimized multi-modal data set;

[0063] Step 3, Defect Self-Growth Simulation Damage Prediction: Based on the optimized multi-modal data set generated in Step 2, use a graph neural network and a deformable grid to establish a three-dimensional self-growth model of bridge cracks and fatigue damage; combine finite element analysis to evaluate the impact of different damages on the bearing capacity of the bridge structure; use self-supervised learning to optimize the damage prediction model under unlabeled conditions;

[0064] Step 4, Life Prediction and Maintenance Suggestions: Based on the three-dimensional self-growth model of bridge cracks and fatigue damage established in Step 3, perform time series prediction through a long short-term memory network, analyze historical data of crack positions, lengths, widths, fatigue damage degrees, propagation paths, strains, displacements, accelerations, environmental conditions, and bridge loads detected in the past, and predict the remaining service life of the bridge; according to the prediction results of the long short-term memory network, use an AI intelligent decision-making model to generate maintenance suggestions, including whether emergency maintenance is required, recommended construction materials, reinforcement plans, and health trend predictions for the next 3-5 years;

[0065] Step 5, Low-Power Intelligent Sensor Network Management: To ensure the continuity and efficiency of multi-modal data acquisition in Step 1, use energy harvesting technology to collect energy from bridge vibrations or solar energy to provide stable energy support for the wireless sensor network; perform data preprocessing at the sensor end to reduce the amount of data transmitted and improve the system response speed; automatically adjust the data acquisition frequency according to the maintenance suggestions generated in Step 4 and the bridge health status, optimize the energy consumption management of the sensor network, and ensure the timely acquisition and transmission of key data;

[0066] Step 6, Quantum-Secure Encryption Remote Monitoring: Use quantum key distribution to generate and distribute quantum keys; use blockchain forensics technology to store and verify the integrity of monitoring data, and support the remote transmission of encrypted monitoring data through 5G and satellite communication technologies to expert diagnosis networks and long short-term memory networks.

[0067] An application of the described detection system in a structural health monitoring system for a concrete-filled steel tube arch bridge, where the concrete-filled steel tube arch bridge includes railway bridges, long-span arch bridges, mountain bridges, and cross-sea bridges.

[0068] An application of the described detection system in a structural health monitoring system for other civil engineering structures, where the other civil engineering structures include prestressed concrete bridges, steel structure bridges, tunnels, dams, and high-rise buildings.

[0069] Advantages of the present invention

[0070] The present invention provides an intelligent detection system for concrete-filled steel tubular arch bridges based on adaptive multimodal data fusion and quantum optimization damage prediction. Compared with the prior art, it has the following remarkable advantages:

[0071] 1. Higher detection accuracy: A variety of sensors such as variable-frequency ultrasonic radar, millimeter-wave radar, hyperspectral imaging, lidar carried by unmanned aerial vehicles, and wireless sensors are used to obtain comprehensive health data on the surface and inside of the bridge from different dimensions, breaking through the limitations of traditional single-sensor detection. The data fusion error is reduced by 30%, and the damage detection accuracy is improved to over 98%, providing more comprehensive and accurate data support for bridge health monitoring. The QAOA quantum optimization algorithm is introduced to automatically adjust the weights of different sensor data, further optimizing the accuracy of data fusion, enabling the system to more accurately identify and locate bridge damage, and improving the reliability of detection results. It is also equipped with an adaptive environment recognition algorithm, enabling the detection system to adaptively optimize under different climate, temperature, and humidity conditions, ensuring that the stability and accuracy of detection are not affected by environmental changes, and enhancing the environmental adaptability of the system.

[0072] 2. More accurate damage prediction: Innovatively, the combination of GNN graph neural network and DMC deformable grid calculation is adopted to simulate complex damage processes such as crack propagation in bridge structures, steel pipe corrosion development, and concrete spalling, establishing a defect self-growth model, providing a simulation basis closer to the actual situation for damage prediction. Combining with FEA finite element analysis, the structural response of the bridge under long-term load, environmental erosion, temperature and humidity changes, etc. is simulated, comprehensively considering the influence of various factors on bridge damage, further improving the accuracy of damage prediction, reducing the prediction error to ±5%, and being able to predict the damage trend 1 - 3 years in advance, providing a strong basis for optimizing the bridge maintenance plan. Unsupervised learning is also used to automatically optimize the damage prediction model, enabling it to adapt to different bridge structures and service environments, and realizing self-optimization and improvement of the model without a large amount of labeled data, enhancing the versatility and adaptability of the system.

[0073] 3. Lower power consumption of the monitoring system: By adopting energy harvesting technology, natural energy such as bridge vibration, solar energy, wind energy, and temperature difference energy is used to power the sensors, realizing self-power supply of the sensors, solving the problems of high power consumption and high maintenance cost caused by the traditional sensors relying on fixed power supplies or batteries, and reducing the long-term operation cost of the sensors. Edge computing is introduced to perform preliminary data processing at the sensor end, and data is uploaded only when key events occur, reducing unnecessary data transmission volume, lowering power consumption during data transmission, and improving the energy efficiency of the system. The DAS dynamic adaptive sampling technology is also used to dynamically adjust the sensor data sampling frequency according to the bridge health condition. On the premise of ensuring the monitoring effect, the power consumption of the sensors is further reduced, the service life of the sensors is extended, the system operation life is extended by more than 3 times, which is suitable for long-term unattended remote bridge monitoring and improves the sustainability of the monitoring system.

[0074] 4. Higher data security: Quantum key distribution technology is adopted to generate highly secure quantum keys for data encryption transmission. Even in the era of quantum computing, it can effectively resist the potential threats of quantum computing to traditional encryption algorithms, ensure the security of bridge monitoring data during transmission, and prevent data from being stolen or tampered with. Combined with blockchain evidence storage technology, all detection data is stored in an immutable blockchain database to ensure the integrity and traceability of the data, providing a strong guarantee for the authenticity and reliability of the data and enhancing the credibility of the monitoring system. 5G and satellite communication technologies are also used to support remote bridge condition monitoring, realizing high-security and low-latency data transmission globally, ensuring that remote experts can obtain monitoring data in a timely manner for diagnosis and decision-making, improving the feasibility and practicality of remote monitoring, and the data transmission security level is increased to 99.999%, effectively preventing problems such as malicious attacks, data loss, or being tampered with.

[0075] 5. Wider applicability: The modular design concept is adopted, enabling the system to be flexibly applicable to different types of bridges, including highway bridges, railway bridges, long-span arch bridges, cross-sea bridges, mountain bridges, etc., enhancing the versatility and scalability of the system. It is not only limited to bridge health monitoring but can also be extended to other civil infrastructure fields, such as tunnel health monitoring for detecting structural deformation, water seepage, etc., dam structural safety monitoring for real-time evaluation of dam stress and leakage, and wind power tower detection for predicting wind turbine structural fatigue damage, etc. It has broad application prospects and market potential, and can provide effective technical support for the intelligent monitoring and management of various infrastructures, improving the overall intelligent level of the infrastructures.

[0076] In summary, the present invention has achieved a revolutionary breakthrough in the field of bridge health monitoring technology. Through a series of innovative technologies such as multi-modal sensor fusion, quantum optimization, defect self-growth simulation, low-power sensor network, and quantum-secure communication, the detection accuracy, damage prediction ability, energy efficiency of the monitoring system, and data security have been significantly improved. At the same time, it has wide applicability and scalability, and can be widely applied to fields such as bridge safety monitoring and intelligent management of urban infrastructure, providing key technical support for future intelligent transportation and intelligent maintenance of infrastructure, and having important social and economic value. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 It is a module diagram of the detection system for concrete-filled steel tubular arch bridges with adaptive multi-modal fusion and quantum optimization of the present invention.

[0078] Figure 2 For Figure 1 It is a flowchart of the quantum optimization data fusion module in

[0079] Figure 3 For Figure 1 It is a flowchart of the defect self-growth simulation damage prediction module in DETAILED DESCRIPTION OF THE INVENTION

[0080] The present invention will be further explained and described below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the specific embodiments are not used to limit the scope of the rights of the present invention.

[0081] As Figures 1 to 3 shown, the detection system for concrete-filled steel tubular arch bridges with adaptive multi-modal fusion and quantum optimization provided by this specific embodiment includes a multi-modal data acquisition module, a quantum optimization data fusion module, a defect self-growth simulation damage prediction module, a low-power intelligent sensor network, and a quantum-secure encryption remote monitoring module.

[0082] The multi-modal data acquisition module is used to obtain damage data on the surface and inside of the bridge, and the damage data includes crack, corrosion, fatigue damage, and deformation data;

[0083] The multi-modal data acquisition module includes:

[0084] A variable-frequency ultrasonic radar for detecting fatigue, debonding, and porosity defects in the internal welds of steel pipes; the parameter conditions of the variable-frequency ultrasonic radar are: the frequency range is set to 50 kHz - 500 kHz, and the detection depth ≤ 500 mm.

[0085] A millimeter-wave radar for evaluating the internal corrosion of steel pipes through the outer wall of the steel pipes; the parameter conditions of the millimeter-wave radar are: the frequency range is set to 24 GHz - 77 GHz, and the accuracy ≤ 0.1 mm.

[0086] Hyperspectral imaging sensors are used to analyze concrete carbonation and crack propagation; the parameter conditions for hyperspectral imaging are as follows: the spectral range is 400 - 2500 nm, and the recognition accuracy is ≤ 2 nm.

[0087] LiDAR carried by drones is used to obtain three-dimensional point cloud data of the bridge surface;

[0088] Wireless sensors, including strain gauges, accelerometers, displacement sensors, and fiber Bragg gratings, are used to monitor the strain and dynamic response of the bridge in real time; high-definition cameras carried by drones are used to collect image data of cracks and deformations on the bridge surface.

[0089] The quantum optimization data fusion module is used to perform fusion processing on the data obtained by the multimodal data acquisition module. The quantum optimization data fusion module uses a quantum optimization algorithm to dynamically adjust the weights of sensor data.

[0090] The quantum optimization data fusion module includes a quantum optimization algorithm unit, a data weight adjustment unit, a Bayesian optimization unit, and an environment adaptive algorithm unit;

[0091] The quantum optimization algorithm unit uses a quantum optimization algorithm (hereinafter referred to as the QAOA algorithm) to optimize the weight allocation of each sensor data in the multimodal data fusion process, thereby achieving the accuracy and robustness of data fusion.

[0092] The data weight adjustment unit includes:

[0093] A Convolutional Neural Network (CNN) subunit is used to extract the spatial features of multimodal sensor data;

[0094] A Transformer subunit is used to process the time series features of multimodal sensor data;

[0095] A data weight dynamic adjustment model combines the outputs of the CNN subunit and the Transformer subunit to dynamically adjust the weights of sensor data;

[0096] The Bayesian optimization unit is used to automatically adjust the data fusion strategy according to environmental conditions, improve the accuracy of data fusion, and reduce the error by more than 30%;

[0097] The environment adaptive algorithm unit is used to make the system adapt to different bridge types and environmental conditions, including changes in temperature, humidity, and wind speed.

[0098] The quantum optimization algorithm unit and the Bayesian optimization unit act together in the data fusion process. The quantum optimization algorithm unit optimizes the fusion result by dynamically adjusting the data weights, while the Bayesian optimization unit adjusts the fusion strategy according to the environmental conditions to ensure the best fusion effect in different environments. This collaborative working method can significantly improve the adaptability and accuracy of the system.

[0099] A defect self-growth simulation damage prediction module. In the defect self-growth simulation damage prediction module, a graph neural network and a long short-term memory network are adopted, and the error is reduced to ±5%. It is used to analyze the data after fusion by the quantum optimization data fusion module and predict the development trend of bridge damage. The defect self-growth simulation damage prediction module is based on the graph neural network and deformable grid calculation, simulates the crack propagation and fatigue damage development path, and combines finite element analysis to predict the remaining life of the bridge.

[0100] The defect self-growth simulation damage prediction module includes:

[0101] A graph neural network unit for simulating the crack propagation path; that is, according to the geometric structure and material properties of the bridge, predicting the propagation direction and speed of the crack.

[0102] A deformable grid calculation unit for simulating the fatigue damage development path, that is, according to the fatigue characteristics of the material, predicting the propagation path and rate of the fatigue damage.

[0103] A finite element analysis unit for combining the outputs of the graph neural network unit and the deformable grid calculation unit and historical data to predict the remaining life of the bridge; the historical data includes the crack positions, lengths, widths, fatigue damage degrees, propagation paths, strains, displacements, accelerations, environmental conditions and bridge load data detected in the past.

[0104] A self-supervised learning unit for optimizing the damage prediction model under unlabeled conditions. The unlabeled conditions refer to learning and optimizing through the internal structure and characteristics of the data in the absence of clearly labeled damage data, improving the accuracy and robustness of the model.

[0105] A low-power intelligent sensor network for reducing the power consumption of wireless sensors in the multi-modal data acquisition module. The low-power intelligent sensor network combines energy harvesting technology and edge computing to optimize the data transmission power consumption.

[0106] The low-power intelligent sensor network includes:

[0107] An energy harvesting unit for harvesting energy from bridge vibration, solar energy or temperature difference to achieve self-power supply of the sensor, and reducing the sensor power consumption by 60%.

[0108] An edge computing unit for pre - processing data at the sensor end, reducing the data transmission volume and frequency, lowering power consumption, uploading data only when abnormal conditions occur, and reducing bandwidth occupancy;

[0109] A dynamic adaptive sampling unit for automatically adjusting the data acquisition frequency according to the bridge health condition, further optimizing power management, and improving data acquisition efficiency.

[0110] A quantum - secure encrypted remote monitoring module for ensuring the security and non - tamperability of the data generated by the multi - modal data acquisition module and the defect self - growing simulation damage prediction module during remote transmission. The quantum - secure encrypted remote monitoring module adopts quantum key distribution and blockchain evidence - storing technologies to make the data transmission security level reach 99.999%.

[0111] The quantum - secure encrypted remote monitoring module includes:

[0112] A quantum key distribution unit for generating and distributing quantum keys to ensure data transmission security;

[0113] A blockchain evidence - storing unit for storing the encrypted monitoring data and verifying the integrity of the data to ensure data non - tampering and traceability;

[0114] A 5G and satellite communication unit for supporting the remote transmission of encrypted monitoring data and expert diagnosis, ensuring efficient and secure data transmission in various environments.

[0115] In the adaptive multi - modal fusion and quantum - optimized concrete - filled steel tubular arch bridge detection system of this embodiment, the multi - modal data acquisition module, the quantum - optimized data fusion module, the defect self - growing simulation damage prediction module, the low - power intelligent sensor network, and the quantum - secure encrypted remote monitoring module communicate and cooperate through standardized interfaces. Each module has a unified data input and output format, ensuring flexible combination and replacement between different modules. The functions of each module can be independently debugged and verified, enabling the system to be flexibly configured in different application scenarios to meet different monitoring requirements. Specifically as follows:

[0116] The multi - modal data acquisition module can select different sensor combinations according to the needs of different bridges or infrastructures. In the concrete - filled steel tubular arch bridge, ultrasonic radar and millimeter - wave radar are used for detection; while in tunnel monitoring, fiber optic sensors and wireless sensors are selected.

[0117] The quantum - optimized data fusion module dynamically adjusts the weights of the data of each sensor through quantum - optimized algorithms and configures according to the application requirements of different types of sensors. For example, for the high - precision ultrasonic radar data, the quantum - optimized algorithm can automatically give a higher weight to improve the fusion accuracy.

[0118] When the monitoring object expands from a concrete-filled steel tubular arch bridge to other types of infrastructure (such as long-span steel structure bridges, tunnels, etc.), it can adapt to new application scenarios by adding or replacing modules.

[0119] According to the requirements of different structural types, the detection system of this embodiment can select or add specific sensors. For example, for tunnel monitoring, it is selected to enhance the data acquisition ability by adding a fiber Bragg grating sensor module; for cross-sea bridges, it is selected to add environmental monitoring modules (such as meteorological sensors, marine salt spray sensors, etc.) to improve the comprehensiveness of monitoring.

[0120] The quantum optimization data fusion module can be customized according to different bridge structures or monitoring scenarios. For example, for bridge monitoring in high-temperature environments, the environmental adaptability module in the data fusion algorithm can be adjusted to improve the data processing ability in extreme climates.

[0121] Users can select and combine different modules according to specific application scenarios, and even allow users to develop customized modules for certain specific scenarios. This flexible configuration ability enables the system to widely adapt to the health monitoring needs of various infrastructures.

[0122] An intelligent detection method for concrete-filled steel tubular arch bridges based on adaptive multimodal data fusion and quantum optimization damage prediction includes the following steps:

[0123] Step 1, Multimodal data acquisition: Use a variable-frequency ultrasonic radar to detect fatigue, debonding, and pore defects in the internal welds of the steel pipe; use a millimeter-wave radar to evaluate the internal corrosion of the steel pipe; use a hyperspectral imaging sensor to analyze the carbonation and crack propagation of the concrete; use a lidar carried by a drone to obtain the three-dimensional point cloud data of the bridge surface; use wireless sensors to continuously monitor the strain and dynamic response of the bridge; use a high-definition camera carried by a drone to collect the crack and deformation image data of the bridge surface;

[0124] Step 2, Quantum optimization data fusion: Adopt a quantum optimization algorithm to dynamically adjust the data weights of each wireless sensor in the multimodal data acquisition module to improve the accuracy and reliability of the data; use the Bayesian optimization method to automatically adjust the data fusion strategy according to the environmental conditions to generate an optimized multimodal data set, providing high-quality data input for subsequent damage prediction;

[0125] In the process of multimodal data fusion, by combining the sensor data of the variable-frequency ultrasonic radar, millimeter-wave radar, and hyperspectral imaging sensor in Step 1 for weighted averaging, the contribution amount of each sensor data is dynamically adjusted according to its accuracy and reliability. The QAOA algorithm is used to optimize the weighted coefficients of the sensor data to improve the overall data fusion accuracy.

[0126] The specific implementation method is as follows:

[0127] (1) Definition of the objective function

[0128] The objective function is used to measure the fusion accuracy after weighting different sensor data. Set the error metric function after data fusion, and the formula is as follows:

[0129]

[0130] In the formula, w represents the weighting coefficient of each sensor data; E(w) represents the fusion error.

[0131] (2) Quantum circuit design

[0132] Use qubits to represent the weighting coefficients of sensor data. The QAOA algorithm adjusts the states of these qubits through quantum gate operations to find the optimal combination of weighting coefficients, thereby minimizing the objective function.

[0133] Initialization stage: Initialize the qubits through an initial quantum state (such as a uniform superposition state).

[0134] Quantum circuit design: Through a series of quantum gate operations (such as Hadamard gates, rotation gates, and controlled-phase gates), realize the interaction between qubits to adjust the weights of each sensor data during the quantum computing process.

[0135] (3) Optimization process

[0136] The core of the QAOA algorithm is to repeatedly execute the quantum circuit, measure the states of the qubits after each execution, and update the parameters in the quantum circuit according to the measurement results. Each round of update is based on the measurement results of the current quantum state, and the parameters of the quantum circuit are adjusted through classical computing methods until the objective function converges to the minimum value.

[0137] (4) Parameter setting and tuning

[0138] When implementing the QAOA algorithm, the key parameters include:

[0139] Number of qubits: Each weighting coefficient of sensor data corresponds to a qubit, usually set to qubits, where is the number of sensors.

[0140] Number of gates: Controls the depth of the QAOA circuit, usually set to layers, and each layer includes Hadamard transformation and controlled-phase gates.

[0141] Optimizer selection: Use classical optimization algorithms (such as gradient descent, genetic algorithms, etc.) to adjust the parameters in the quantum circuit to minimize the objective function.

[0142] In practical applications, the parameters of QAOA (such as the number of qubits, the number of quantum gates, etc.) will be adjusted according to the specific requirements of bridge health monitoring.

[0143] (5) Optimization process of the QAOA algorithm

[0144] Through quantum computing, QAOA can find an optimized weight combination to minimize the error of data fusion. The specific steps are as follows:

[0145] Initialize the weighting coefficients of sensor data to their initial values.

[0146] Adjust the weights of each sensor data through a quantum circuit, and continuously optimize the objective function using QAOA.

[0147] Measure the qubits and feedback the results through classical computing methods to update the parameters in the quantum circuit.

[0148] Repeat the optimization process until the error reaches the minimum value.

[0149] Through the above process, QAOA can dynamically adjust the weights of different sensor data, improve the accuracy of multi-modal data fusion, and thus ensure a comprehensive and accurate assessment of the health status of the concrete-filled steel tube arch bridge by the detection system.

[0150] (6) Combination of the QAOA algorithm and Bayesian optimization

[0151] Combine the QAOA algorithm and Bayesian optimization technology to dynamically adjust the data fusion strategy according to environmental conditions (such as temperature, humidity, wind speed, etc.). In actual operation, Bayesian optimization can automatically adjust the parameters in the QAOA algorithm by modeling environmental variables to optimize the data fusion process. For example, in an environment with high humidity, it may be necessary to increase the weight of the ultrasonic radar, while in an environment with low temperature, the weight of the lidar may be appropriately increased.

[0152] (7) Implementation effects and advantages

[0153] By introducing the QAOA algorithm, the system can achieve more accurate and dynamic data fusion, solving the limitations brought by fixed weights in traditional data weighting methods. The optimized fusion data enables the damage detection accuracy to reach over 98%, reducing the error by 30%, being able to better adapt to environmental changes, and improving the overall performance of bridge health monitoring.

[0154] Step 3, Defect Self-Growth Simulation Damage Prediction: Based on the optimized multi-modal dataset generated in Step 2, use graph neural networks and deformable meshes to establish a three-dimensional self-growth model of bridge cracks and fatigue damage to simulate the crack propagation path and fatigue damage development process in the concrete-filled steel tube arch bridge structure; and combine it with the finite element analysis model for fusion to evaluate the impact of different damages on the bearing capacity of the bridge structure, so as to achieve high-precision structural life prediction and damage development trend analysis. Use self-supervised learning to optimize the damage prediction model under unlabeled conditions to further improve the accuracy and reliability of the prediction, as follows:

[0155] I. Role of Graph Neural Network (GNN) in Crack Propagation Simulation

[0156] (1) Input Data Structure

[0157] Model the bridge structure as a graph structure: Nodes represent key structural points (such as connection nodes, stress concentration areas), and edges represent structural connection relationships (such as steel bar connections, welds, stress paths between concrete blocks).

[0158] The node attributes in the graph include: strain, stress, material type, historical damage data.

[0159] Edge attributes include: material connection strength, historical crack information, etc.

[0160] (2) Simulation Mechanism

[0161] GNN learns the influence relationship between nodes on the graph through the Message Passing mechanism;

[0162] Simulate the crack development trend generated by stress-strain changes on the structural nodes;

[0163] Predict the possible propagation path and speed of cracks through multiple rounds of iteration, and identify potential risk areas.

[0164] II. Role of Deformable Mesh Computation (DMC) in Fatigue Development Simulation

[0165] (1) Mesh Initialization

[0166] Establish an initial mesh based on the bridge finite element model, with the node density encrypted in key areas (such as stress concentration areas);

[0167] The initial mesh is distributed based on material properties and load history;

[0168] (2) Dynamic Evolution Mechanism

[0169] Generate mesh deformation areas in fatigue-sensitive areas according to the crack propagation path predicted by GNN;

[0170] Dynamically adjust the shape and size of grid cells to more precisely simulate the stress concentration phenomenon at the crack tip;

[0171] Adjust the grid density and calculation accuracy considering the stress intensity factor (SIF) and fatigue crack growth rate (e.g., based on Paris' law) at the crack tip.

[0172] III. Integration Method of GNN and DMC

[0173] Step 1: GNN conducts topological learning on the bridge structure and outputs possible crack propagation paths (such as node-edge path sequences);

[0174] Step 2: Use the output result of GNN as guiding information to dynamically adjust the grid density and local deformation range of DMC;

[0175] Step 3: DMC simulates the three-dimensional morphological changes of crack fatigue damage according to the local stress state;

[0176] Step 4: The output results of GNN and DMC are used as the input of the damage evolution of the structural model and fed back to the FEA module for full-structure stress recalculation;

[0177] IV. Fusion Mechanism with Finite Element Analysis (FEA)

[0178] Initial bridge model establishment: Construct a finite element model based on the actual bridge structure design parameters (materials, dimensions, boundary conditions);

[0179] (1) Coupled data input:

[0180] GNN output: Potential crack paths and time series predictions;

[0181] DMC output: Local grid deformation area and fatigue evolution model;

[0182] (2) Joint analysis:

[0183] Input the above outputs as the "damage field" into FEA;

[0184] FEA iteratively solves the overall stress field, strain field, and displacement field of the bridge under the influence of damage;

[0185] (3) Remaining life assessment:

[0186] Output the response of the bridge structure under future loads through FEA, and evaluate the remaining life in combination with damage factors (such as Miner's fatigue accumulation);

[0187] Predict the time node when the structure reaches the limit state to provide a basis for the maintenance plan.

[0188] V. Summary of Technical Advantages

[0189] High-precision simulation: The GNN captures topological features, the DMC realizes local micro-scale simulation, and the FEA is responsible for the overall response analysis. The three form a closed-loop coupled simulation;

[0190] Strong prediction ability: It can identify the initial position of cracks and the future development path in advance;

[0191] Strong system adaptability: It can adapt to bridge structures with different materials, structural forms, and load conditions;

[0192] Optimize the maintenance strategy: Provide a more realistic input of the structural evolution trend for the AI decision-making system.

[0193] Step 4, Life prediction and maintenance suggestions: Based on the three-dimensional self-growing model of bridge cracks and fatigue damage established in Step 3, perform time-series prediction through a long short-term memory network, analyze the historical data of the crack positions, lengths, widths, fatigue damage degrees, propagation paths, strains, displacements, accelerations, environmental conditions, and bridge loads detected in the past, and predict the remaining service life of the bridge; According to the prediction results of the long short-term memory network, use the AI intelligent decision-making model to generate maintenance suggestions, including whether emergency maintenance is required, the recommended construction materials, reinforcement plans, and the health trend prediction for the next 3-5 years;

[0194] Step 5, Low-power intelligent sensor network management: To ensure the continuity and efficiency of multi-modal data collection in Step 1, use energy harvesting technology to collect energy from bridge vibrations or solar energy to provide stable energy support for the wireless sensor network; Perform data preprocessing at the sensor end to reduce the amount of data transmitted and improve the system response speed; Automatically adjust the data collection frequency according to the maintenance suggestions generated in Step 4 and the bridge health status, optimize the energy consumption management of the sensor network, and ensure the timely collection and transmission of key data;

[0195] Step 6, Quantum-secure encrypted remote monitoring: To ensure the security and integrity of the data collected and processed in Steps 1 to 4, use quantum key distribution to generate and distribute quantum keys to encrypt the monitoring data; Use blockchain evidence storage technology to store and verify the integrity of the monitoring data to ensure the immutability of the data during transmission and storage; Support the remote transmission of the encrypted monitoring data through 5G and satellite communication technologies, enabling experts to perform real-time diagnosis and analysis, and further improving the intelligent level of the bridge detection system.

[0196] An application of the above detection system in a structural health monitoring system for a concrete-filled steel tube arch bridge, where the concrete-filled steel tube arch bridge includes railway bridges, long-span arch bridges, mountain bridges, and sea-crossing bridges.

[0197] The application of the above detection system in the structural health monitoring systems of other civil engineering structures, where the other civil engineering structures include prestressed concrete bridges, steel structure bridges, tunnels, dams, and high-rise buildings.

[0198] To ensure that the detection system can adapt to various infrastructure environments, the detection system of this specific embodiment is experimentally verified in multiple scenarios as follows:

[0199] (1) Application verification of concrete-filled steel tube arch bridges

[0200] Application scenario: In actual concrete-filled steel tube arch bridges, a multi-modal data acquisition module is adopted, and surface and internal damage data of the bridge are collected by means of ultrasonic radar, millimeter-wave radar, and wireless sensors.

[0201] Verification method: The system automatically adjusts the weights of the data of each sensor through the quantum optimization data fusion module, and combines graph neural network and finite element analysis to predict the propagation path of cracks and the remaining life of the bridge. The experimental results show that the damage prediction accuracy of the detection system in this embodiment reaches more than 98%, and it can effectively identify potential cracks and fatigue damage.

[0202] Modular expansion: The detection system of this embodiment can quickly adapt to the specific requirements of different bridges by adjusting the combination of sensor modules. For example, the detection ability for different materials can be optimized by replacing or adjusting the types of hyperspectral imaging sensors.

[0203] (2) Structural health monitoring of tunnels

[0204] Application scenario: In the structural health monitoring of tunnels, a multi-modal data acquisition scheme combining fiber optic sensors and wireless sensors is adopted to obtain strain, displacement, and temperature data of the tunnel.

[0205] Verification method: Through the quantum optimization data fusion module, the data of fiber optic sensors and wireless sensors are optimized to ensure the accuracy of the data. The self-growing simulated damage prediction module combines graph neural network and DMC model to predict the propagation path of tunnel cracks and fatigue damage.

[0206] Modular expansion: During the application process of the system, the accuracy of tunnel monitoring is effectively improved by adding specific environmental monitoring modules (such as humidity sensors and gas detection sensors).

[0207] (3) Application verification of long-span steel structure bridges

[0208] Application scenario: In the monitoring of long-span steel structure bridges, the system uses a combination of lidar and millimeter-wave radar for surface and internal structure monitoring, and at the same time uses wireless sensors to monitor the vibration and strain of the bridge in real time.

[0209] Verification method: Through the quantum-secure encryption remote monitoring module, ensure the secure and reliable remote transmission process of all data, and perform dynamic weighted optimization processing of the data through the quantum optimization algorithm to ensure high-precision damage detection.

[0210] Modular expansion: The system can easily adapt to the health monitoring needs of various steel structures by adjusting the data acquisition module and the fusion module according to different steel structure bridge designs.

[0211] The detection system of this embodiment has successfully achieved high scalability and adaptability of the system through modular design. The detection system can flexibly select, combine, and replace different functional modules according to the types and requirements of different infrastructures to ensure accurate and efficient monitoring services in various monitoring scenarios.

Claims

1. A detection system for concrete-filled steel tubular arch bridges with adaptive multimodal fusion and quantum optimization, characterized in that, It includes the following modules: A multimodal data acquisition module, which is used to obtain damage data on the surface and inside of the bridge. The damage data includes crack, corrosion, fatigue damage, and deformation data; A quantum optimization data fusion module, which is used to perform fusion processing on the data obtained by the multimodal data acquisition module. The quantum optimization data fusion module uses a quantum optimization algorithm to dynamically adjust the weights of sensor data; A defect self-growth simulation damage prediction module, which is used to analyze the data fused by the quantum optimization data fusion module and predict the development trend of bridge damage; A low-power intelligent sensor network, which is used to reduce the power consumption of the multimodal data acquisition module; A quantum-secure encryption remote monitoring module, which is used to ensure the security and non-tamperability of the data generated by the multimodal data acquisition module and the defect self-growth simulation damage prediction module during remote transmission.

2. The detection system according to claim 1, wherein The multimodal data acquisition module includes: A variable-frequency ultrasonic radar, which is used to detect fatigue, debonding, and pore defects in the internal welds of steel pipes; A millimeter-wave radar, which is used to evaluate the corrosion condition inside steel pipes; A hyperspectral imaging sensor, which is used to analyze concrete carbonation and crack propagation; A lidar carried by a drone, which is used to obtain the three-dimensional point cloud data of the bridge surface; Wireless sensors, including strain gauges, accelerometers, displacement sensors, and fiber Bragg gratings, which are used to monitor the strain and dynamic response of the bridge in real time; A high-definition camera carried by a drone, which is used to collect crack and deformation image data of the bridge surface.

3. The detection system according to claim 1, wherein The quantum optimization data fusion module includes: A quantum optimization algorithm unit, which is used to adopt a quantum optimization algorithm to fuse and optimize the wireless sensor data in the multimodal data acquisition module; A data weight adjustment unit, which is used to construct a dynamic data weight adjustment model; A Bayesian optimization unit, which is used to automatically adjust the data fusion strategy according to environmental conditions such as temperature, humidity, and wind speed; An environment adaptive algorithm unit, which is used to ensure that the system can adapt to different types of bridges and environmental conditions.

4. The detection system according to claim 1, wherein The defect self-growth simulation damage prediction module uses a graph neural network and a long short-term memory network. The defect self-growth simulation damage prediction module includes: A graph neural network unit, which is used to simulate the crack propagation path; A deformable mesh calculation unit, which is used to simulate the fatigue damage development path; A finite element analysis unit, which is used to combine the outputs of the graph neural network unit and the deformable mesh calculation unit and historical data to predict the remaining life of the bridge. The historical data includes the crack positions, lengths, widths, fatigue damage degrees, propagation paths, strain, displacement, acceleration, environmental conditions, and bridge load data detected in the past; A self-supervised learning unit, which is used to optimize the damage prediction model under unlabeled conditions. The unlabeled conditions refer to learning and optimizing through the internal structure and characteristics of the data in the case of no clear labeled damage data.

5. The detection system according to claim 1, characterized in that, The low-power intelligent sensor network includes: An energy harvesting unit, which is used to harvest energy from bridge vibration, solar energy, or temperature difference energy; An edge computing unit, which is used to perform data preprocessing at the sensor end to reduce the data transmission volume and power consumption; A dynamic adaptive sampling unit, which is used to automatically adjust the data acquisition frequency according to the bridge health condition.

6. The detection system according to claim 1, wherein, The quantum-secure encrypted remote monitoring module includes: a quantum key distribution unit for generating and distributing quantum keys; a blockchain evidence storage unit for storing the encrypted monitoring data and verifying the integrity of the data; a 5G and satellite communication unit for supporting the remote transmission of the encrypted monitoring data and expert diagnosis.

7. An intelligent detection method for concrete-filled steel tubular arch bridges based on adaptive multi-modal data fusion and quantum optimization damage prediction, characterized in that, It includes the following steps: Step 1, multi-modal data acquisition: Use a variable-frequency ultrasonic radar to detect fatigue, debonding, and porosity defects in the internal welds of steel pipes; use a millimeter-wave radar to evaluate the internal corrosion of steel pipes; use a hyperspectral imaging sensor to analyze concrete carbonation and crack propagation; use a lidar carried by a drone to obtain the three-dimensional point cloud data of the bridge surface; use wireless sensors to monitor the strain and dynamic response of the bridge in real time; use a high-definition camera carried by a drone to collect the crack and deformation image data of the bridge surface; Step 2, quantum-optimized data fusion: Adopt a quantum optimization algorithm to dynamically adjust the data weights of each wireless sensor in the multi-modal data acquisition module; automatically adjust the data fusion strategy according to the environmental conditions through the Bayesian optimization method to generate an optimized multi-modal data set; Step 3, defect self-growth simulation damage prediction: Based on the optimized multi-modal data set generated in Step 2, use a graph neural network and a deformable grid to establish a three-dimensional self-growth model of bridge cracks and fatigue damage; combine finite element analysis to evaluate the impact of different damages on the bearing capacity of the bridge structure; adopt self-supervised learning to optimize the damage prediction model under unlabeled conditions; Step 4, life prediction and maintenance suggestions: Based on the three-dimensional self-growth model of bridge cracks and fatigue damage established in Step 3, perform time-series prediction through a long short-term memory network, analyze the historical data of the crack position, length, width, fatigue damage degree, propagation path, strain, displacement, acceleration, environmental conditions, and bridge load detected in the past, and predict the remaining service life of the bridge; according to the prediction results of the long short-term memory network, adopt an AI intelligent decision-making model to generate maintenance suggestions, including whether emergency maintenance is required, the recommended construction materials, reinforcement plans, and the health trend prediction for the next 3-5 years; Step 5, low-power intelligent sensor network management: Use energy harvesting technology to collect energy from bridge vibrations or solar energy for data preprocessing at the sensor end in Step 1, and automatically adjust the data acquisition frequency and optimize the energy consumption management of the sensor network according to the maintenance suggestions generated in Step 4 and the bridge health status; Step 6, quantum-secure encrypted remote monitoring: Use quantum key distribution to generate and distribute quantum keys; use blockchain evidence storage technology to store and verify the integrity of the monitoring data, and support the remote transmission of the encrypted monitoring data and expert diagnosis networks and long short-term memory networks through 5G and satellite communication technologies.

8. The application of the detection system according to claim 1 in the structural health monitoring system of a concrete-filled steel tube arch bridge, wherein the concrete-filled steel tube arch bridge includes railway bridges, long-span arch bridges, mountain bridges, and cross-sea bridges.

9. Application of the detection system according to claim 1 in the structural health monitoring system of other civil engineering structures, wherein the other civil engineering structures include prestressed concrete bridges, steel structure bridges, tunnels, dams and high-rise buildings.

Images