Cable-stayed bridge pylon digital monitoring method and related equipment

Abstract

The application discloses a cable-stayed bridge pylon digital monitoring method and related equipment, and relates to the technical field of bridge monitoring. The method comprises the following steps: acquiring initial data, wherein the initial data comprises pylon environment data and three-dimensional shape information; positioning a skeleton position based on magnetic coding data and the three-dimensional shape information; adjusting a template posture based on the skeleton position, so that the template posture matches the skeleton position, and acquiring correction data in the adjustment process; monitoring a pouring process based on an intelligent hydraulic control system to obtain dynamic data; based on a concrete surface image, analyzing feature points of the concrete surface image, calculating a deviation error, and determining deviation data; integrating multi-source data to monitor a pylon structure, and predicting a health state of the pylon structure, wherein the multi-source data comprises the initial data, the correction data, the dynamic data, and the deviation data.

Description

Technical Field

[0001] This application relates to the field of bridge monitoring technology, and in particular to a digital monitoring method and related equipment for cable-stayed bridge towers. Background Technology

[0002] Existing monitoring technologies for cable-stayed bridge towers typically rely on regular manual inspections and localized sensor monitoring. This traditional method suffers from limitations such as limited monitoring range, incomplete data collection, and poor real-time performance, making it impossible to comprehensively and accurately assess the overall structural health of the towers.

[0003] Furthermore, due to limitations in monitoring technology, existing monitoring methods cannot provide dynamic monitoring of the tower during the pouring and construction process, which may result in potential structural deviations not being corrected in a timely manner.

[0004] With the increasing demands for structural safety in modern bridge engineering, there is an urgent need for a digital monitoring method that can monitor the health status of cable-stayed bridge towers in real time and comprehensively. Summary of the Invention

[0005] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. This summary section is not intended to limit the key and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.

[0006] Firstly, this application provides a digital monitoring method for cable-stayed bridge towers, including:

[0007] Acquire initial data, which includes Sota environmental data and three-dimensional morphological information;

[0008] The skeleton position is located based on the magnetic encoding data and the aforementioned three-dimensional morphological information.

[0009] Adjust the template posture based on the above skeleton position to match the template posture with the above skeleton position, and obtain the correction data during the adjustment process;

[0010] The pouring process is monitored using an intelligent hydraulic control system to obtain dynamic data;

[0011] Based on the concrete surface image, the feature points of the concrete surface image are analyzed, the deviation error is calculated, and the deviation data is determined.

[0012] The system integrates multi-source data to monitor the cable tower structure and predict its health status. The multi-source data includes the initial data, the correction data, the dynamic data, and the offset data.

[0013] In one feasible implementation, the aforementioned environmental data of the tower includes a sensor data matrix containing time series data, which includes temperature data, humidity data, stress data, and vibration data.

[0014] The aforementioned three-dimensional morphological information is point cloud data transformed from a set of three-dimensional coordinates. This set of three-dimensional coordinates is a collection of three-dimensional coordinates of the cable tower nodes obtained by scanning the cable tower with a lidar.

[0015] In one feasible implementation, it further includes:

[0016] Based on the coordinate information collected by the magnetic encoder, the above magnetic encoding data is determined;

[0017] The skeleton positions are calculated based on the spatial location and weight of the encoding points.

[0018] In one feasible implementation, the above-mentioned adjustment of the template pose based on the skeleton position to match the template pose with the skeleton position includes:

[0019] The target attitude adjustment amount is determined based on minimizing the error between the template position and the aforementioned skeleton position;

[0020] Based on the aforementioned target attitude adjustment amount, a deep learning model is used to optimize the algorithm to generate an attitude adjustment signal;

[0021] The attitude adjustment process of the template is controlled based on the attitude adjustment signal to match the template attitude with the skeleton position.

[0022] In one feasible implementation, it further includes:

[0023] The aforementioned intelligent hydraulic control system includes a hydraulic pump, a hydraulic cylinder, and a control system;

[0024] The aforementioned control system monitors the operating status of the hydraulic pump and the hydraulic cylinder, and dynamically adjusts the operating parameters of the hydraulic pump based on a PID control algorithm.

[0025] In one feasible implementation, the above-mentioned analysis of feature points in the concrete surface image and calculation of the offset error based on the concrete surface image includes:

[0026] The above concrete surface image is preprocessed by denoising and illumination correction.

[0027] Feature points of the above concrete surface image were extracted using a convolutional neural network.

[0028] The aforementioned feature points are compared with the preset positions of the template, and the aforementioned offset error is calculated.

[0029] In one feasible implementation, the above-mentioned integration of multi-source data monitoring of the cable tower structure and prediction of the health status of the cable tower structure includes:

[0030] Based on the above multi-source data, the tilt state and damage level of the tower column were assessed using big data analysis technology.

[0031] The aforementioned multi-source data, tilt status, and damage level are input into the time series analysis prediction model to obtain the health status of the tower structure.

[0032] Secondly, this application proposes a digital monitoring device for cable-stayed bridge towers, comprising:

[0033] The acquisition unit is used to acquire initial data, which includes the Sota environment data and three-dimensional morphological information.

[0034] The positioning unit is used to locate the skeleton position based on the magnetic encoding data and the above-mentioned three-dimensional morphological information;

[0035] The matching unit is used to adjust the template posture based on the skeleton position so that the template posture matches the skeleton position, and to acquire correction data during the adjustment process.

[0036] The monitoring unit is used to monitor the pouring process based on the intelligent hydraulic control system and obtain dynamic data.

[0037] The recognition unit is used to analyze the feature points of the concrete surface image based on the concrete surface image, calculate the offset error, and determine the offset data.

[0038] The prediction unit is used to integrate multi-source data to monitor the cable tower structure and predict the health status of the cable tower structure. The multi-source data includes the initial data, the correction data, the dynamic data, and the offset data.

[0039] Thirdly, an electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program stored in the memory to implement the steps of the digital monitoring method for cable-stayed bridge towers as described in any of the first aspects above.

[0040] Fourthly, this application also proposes a computer-readable storage medium storing a computer program thereon, characterized in that: when the computer program is executed by a processor, it implements the digital monitoring method for cable-stayed bridge towers according to any one of the first aspects.

[0041] In summary, this application's embodiments, through the combination of LiDAR scanning, magnetic encoders, and multiple types of environmental sensors, can acquire the three-dimensional morphology of the cable tower, the position of the frame, the posture of the formwork, and environmental data (such as temperature, humidity, stress, and vibration data) during the pouring process. This multi-source data acquisition method allows the monitoring system to comprehensively understand the structural state and construction progress of the cable tower from multiple dimensions, providing a more detailed data foundation than a single sensor system. This application employs a technology combining magnetic encoders and LiDAR, which can accurately determine the position of the cable tower frame and automatically optimize the formwork posture using deep learning models and reinforcement learning algorithms. This intelligent correction method can ensure precise matching between the formwork and the frame by minimizing the error between them, thereby improving the accuracy of concrete pouring and avoiding structural problems caused by formwork misalignment. During the concrete pouring process, an intelligent hydraulic control system is used to monitor and adjust the formwork position in real time. This system, based on a PID control algorithm, can dynamically adjust the operating parameters of the hydraulic pump to ensure that the formwork remains in the optimal position throughout the pouring process. This not only improves the pouring accuracy but also avoids structural deviations caused by changes in the formwork position during construction. After concrete pouring, a high-precision camera mounted on a drone acquires images of the concrete surface, and a convolutional neural network (CNN) is used to analyze the feature points of the concrete surface to detect surface deviations after pouring. This application can compare the deviation between the actual surface and the preset template in real time to ensure pouring quality, and enables remote analysis and storage of data through cloud data processing. By integrating multi-source data (initial data, correction data, dynamic data, and deviation data) during the construction process, this application can use big data analysis technology to assess the overall health status of the tower, including the tilt angle and degree of damage of the tower columns. Simultaneously, by using a time series analysis model to predict the long-term structural health status of the tower, it helps to identify potential structural problems in advance and carry out preventative maintenance. The intelligent system of this application can automatically monitor and adjust the pouring process, reducing errors from manual operation. All data is uploaded to the cloud in real time via wireless transmission, and construction personnel can view the health status and construction progress of the tower in real time through a remote monitoring platform, significantly improving construction efficiency and safety.

[0042] The digital monitoring method for cable-stayed bridge towers proposed in this invention, along with other advantages, objectives, and features of this invention, will be partly apparent from the following description and partly understood by those skilled in the art through study and practice of this invention. Attached Figure Description

[0043] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit this specification. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0044] Figure 1 A schematic flowchart of a digital monitoring method for cable-stayed bridge towers provided in this application embodiment;

[0045] Figure 2 A schematic diagram of a digital monitoring device for cable-stayed bridge towers provided in this application embodiment;

[0046] Figure 3 This is a schematic diagram of a digital monitoring electronic device for cable-stayed bridge towers, provided as an embodiment of this application. Detailed Implementation

[0047] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus. The technical solutions of the embodiments of this application will now be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0048] Please see Figure 1 This is a schematic diagram of a digital monitoring method for cable-stayed bridge towers provided in an embodiment of this application, which specifically includes:

[0049] S110. Obtain initial data, wherein the initial data includes the Sota environment data and three-dimensional morphological information;

[0050] For example, in the construction or monitoring of cable-stayed bridge towers, initial data needs to be acquired first. This initial data includes environmental data and three-dimensional morphological information of the tower. Environmental data is collected in real time by sensors deployed at various key locations on the tower, collecting environmental parameters such as temperature, humidity, stress, and vibration. All of this data is collected over time. This data not only reflects the external environmental conditions of the tower but also helps analyze the stress changes and responses of the structure under different conditions. Three-dimensional morphological information is obtained using lidar scanning technology, which acquires the geometric information of the tower by emitting laser beams and receiving reflected signals. The scan results are presented in the form of a point cloud, i.e., a set of three-dimensional coordinates of the tower structure. This three-dimensional point cloud data provides the precise geometry of the tower and lays the foundation for calculating the skeleton position and adjusting the template orientation in subsequent steps.

[0051] S120. Locate the skeleton position based on magnetic encoding data and the above-mentioned three-dimensional morphological information;

[0052] For example, the spatial position of the internal skeleton of the cable tower is precisely located using 3D morphological information obtained from magnetic encoders and LiDAR. Magnetic encoders are installed on key nodes of the cable tower skeleton to detect the precise positions of these nodes. The magnetic encoders generate coordinate information representing the spatial position of each node in the skeleton by monitoring changes in the magnetic field. Point cloud data generated by the LiDAR provides a reference for the overall 3D shape of the skeleton. Combining the encoded point data collected by the magnetic encoders, the center position and spatial orientation of the skeleton can be located through weighted averaging or other computational models. Accurate positioning of the skeleton is crucial for subsequent template correction.

[0053] S130. Adjust the template posture based on the above skeleton position so that the template posture matches the above skeleton position, and obtain the correction data during the adjustment process;

[0054] For example, the template is a temporary frame used to define the shape of the poured concrete. It must be precisely aligned with the skeleton to ensure the accuracy of the pouring process. In this step, the template's posture is adjusted to match the positioned skeleton, and correction data is acquired in the process. The position and angle of the template must be finely adjusted to maintain consistency with the skeleton. Using deep learning algorithms and posture optimization models, the template's posture is continuously adjusted based on the initial error between the skeleton and the template until the error is minimized. This process relies on the precise position data provided by the magnetic encoder and LiDAR. During the posture adjustment process, the template's position information and the error after each adjustment are recorded in real time. This data is stored as correction data for use in subsequent steps.

[0055] S140. The pouring process is monitored based on an intelligent hydraulic control system to obtain dynamic data;

[0056] During concrete pouring, the system uses an intelligent hydraulic control system to dynamically monitor the formwork's posture, ensuring its stability throughout the process. The hydraulic system consists of a hydraulic pump, hydraulic cylinders, and a control system. By monitoring the deviation between the actual and expected positions of the formwork, the system can adjust the hydraulic pump's operating parameters (such as pressure and flow rate) in real time to maintain formwork stability. The system relies on a PID (proportional, integral, derivative) control algorithm for adjustment. By comparing the deviation between the actual and target positions of the formwork, it continuously adjusts the operation of the hydraulic cylinders to ensure the formwork position matches the predetermined target. Dynamic monitoring data records every adjustment during the pouring process, forming a real-time dynamic dataset.

[0057] S150. Based on the concrete surface image, analyze the feature points of the concrete surface image, calculate the deviation error, and determine the deviation data.

[0058] For example, after concrete pouring is completed, image processing techniques are used to analyze the concrete surface to detect any pouring errors. Images of the poured concrete surface are acquired using drones or other high-precision camera equipment. To improve image quality, the images undergo preprocessing steps such as noise reduction and illumination correction before being input into the deep learning model to ensure the clarity of feature points. Feature points are extracted from the concrete surface images using a convolutional neural network (CNN). CNNs can automatically identify key points on the concrete surface, such as edges, cracks, or uneven areas. The actual feature points in the image are compared with the preset positions on the template to calculate the offset error generated during concrete pouring. This error data can be used to evaluate construction quality and provide a reference for subsequent adjustments.

[0059] S160. Integrate multi-source data to monitor the tower structure and predict the health status of the tower structure, wherein the multi-source data includes initial data, correction data, dynamic data and offset data.

[0060] For example, this step integrates all the data generated in the preceding steps and analyzes and predicts the health status of the cable tower based on this multi-source data. Through big data analytics, this data from different sources is integrated to form a comprehensive structural condition monitoring system. The integrated data not only reflects the current geometry and construction status of the cable tower but also includes information on potential deviations during the pouring process. Based on the integrated data, big data analytics and time series forecasting models enable long-term monitoring and prediction of the cable tower's health status. The model can assess the cable tower's tilt, stress concentration areas, and potential damage locations, helping to predict future structural problems and thus providing data support for maintenance and repair.

[0061] In summary, this application's embodiments, through the combination of LiDAR scanning, magnetic encoders, and multiple types of environmental sensors, can acquire the three-dimensional morphology of the cable tower, the position of the frame, the posture of the formwork, and environmental data (such as temperature, humidity, stress, and vibration data) during the pouring process. This multi-source data acquisition method allows the monitoring system to comprehensively understand the structural state and construction progress of the cable tower from multiple dimensions, providing a more detailed data foundation than a single sensor system. This application employs a technology combining magnetic encoders and LiDAR, which can accurately determine the position of the cable tower frame and automatically optimize the formwork posture using deep learning models and reinforcement learning algorithms. This intelligent correction method can ensure precise matching between the formwork and the frame by minimizing the error between them, thereby improving the accuracy of concrete pouring and avoiding structural problems caused by formwork misalignment. During the concrete pouring process, an intelligent hydraulic control system is used to monitor and adjust the formwork position in real time. This system, based on a PID control algorithm, can dynamically adjust the operating parameters of the hydraulic pump to ensure that the formwork remains in the optimal position throughout the pouring process. This not only improves the pouring accuracy but also avoids structural deviations caused by changes in the formwork position during construction. After concrete pouring, a high-precision camera mounted on a drone acquires images of the concrete surface, and a convolutional neural network (CNN) is used to analyze the feature points of the concrete surface to detect surface deviations after pouring. This application can compare the deviation between the actual surface and the preset template in real time to ensure pouring quality, and enables remote analysis and storage of data through cloud data processing. By integrating multi-source data (initial data, correction data, dynamic data, and deviation data) during the construction process, this application can use big data analysis technology to assess the overall health status of the tower, including the tilt angle and degree of damage of the tower columns. Simultaneously, by using a time series analysis model to predict the long-term structural health status of the tower, it helps to identify potential structural problems in advance and carry out preventative maintenance. The intelligent system of this application can automatically monitor and adjust the pouring process, reducing errors from manual operation. All data is uploaded to the cloud in real time via wireless transmission, and construction personnel can view the health status and construction progress of the tower in real time through a remote monitoring platform, significantly improving construction efficiency and safety.

[0062] In some examples, the aforementioned Sota environmental data includes a time-series sensor data matrix containing temperature, humidity, stress, and vibration data.

[0063] The aforementioned three-dimensional morphological information is point cloud data transformed from a set of three-dimensional coordinates. This set of three-dimensional coordinates is a collection of three-dimensional coordinates of the cable tower nodes obtained by scanning the cable tower with a lidar.

[0064] For example, a lidar system obtains the three-dimensional coordinate information of key nodes of a cable tower by emitting a laser beam and detecting its reflected signal. Each node of the cable tower is represented as a set of discrete point cloud data, and the three-dimensional coordinates of each point are represented as T. i，j，k (x, y, z), where i, j, and k are the indices of the Sota node, and (x, y, z) are the coordinates of the node in three-dimensional space.

[0065] Each beam of light emitted by the lidar calculates the distance using the time-of-flight principle, samples the distances of multiple reflection points, and obtains a set of three-dimensional point cloud data for the entire cable tower. In order to reconstruct the three-dimensional shape of the cable tower more accurately, the lidar will repeatedly scan multiple angles and use the iterative closest point algorithm (ICP) to perform multiple alignments, ultimately obtaining an accurate three-dimensional shape.

[0066] Multiple sensors were deployed at different locations on the tower to monitor various environmental parameters. The data matrix generated by the sensors can be represented as follows:

[0067] S(t) = [T t H t E t V t ]

[0068] Among them, T t Here is the temperature data at time t; H t For humidity data; E t For stress data; V t t represents vibration data; t represents the time parameter.

[0069] The data collected by the sensors is considered a function of time; this data updates continuously over time, therefore each set of data can be viewed as a time-series multidimensional dataset. All sensor data and the 3D morphological data generated by the LiDAR are uploaded to the cloud in real time via a wireless network, forming a preliminary dataset. This dataset is represented as follows:

[0070] D1={S(t),T i，j，k (x,y,z)}

[0071] This set includes the sensor environment data matrix S(t) and the three-dimensional spatial coordinate point set T. i，j，k (x, y, z) is a multidimensional dataset composed of sensor data and three-dimensional morphological data, containing information on the tower's geometry and structural health. This dataset is the core input for subsequent steps, used for tasks such as template calibration and hydraulic system control.

[0072] In some examples, it also includes:

[0073] Based on the coordinate information collected by the magnetic encoder, the above magnetic encoding data is determined;

[0074] The skeleton positions are calculated based on the spatial location and weight of the encoding points.

[0075] For example, a magnetic encoder is a sensor capable of detecting changes in magnetic field strength and converting them into an electrical signal output. Magnetic encoders are installed at key locations on the rigid frame (such as template interfaces, important nodes, etc.). The magnetic encoders, installed at key locations on the rigid frame, collect coordinate information from multiple points on the frame.

[0076] The "skeleton" typically refers to the internal supporting structure in a bridge. In the construction of cable-stayed bridge towers, the skeleton generally refers to the structural frame or stiffening skeleton used to support and fix the tower. These skeletons are responsible for bearing the loads during construction and the bridge's service life, ensuring that the overall shape and stiffness of the tower meet design requirements. Spatial coordinate data of the skeleton is acquired using magnetic encoders and lidar for precise positioning to ensure it aligns with the design specifications. The "formwork" is a temporary structure used to define the shape of the concrete during pouring. The formwork is the external frame installed around the skeleton to ensure that the concrete solidifies according to the predetermined shape and dimensions during pouring. The installation posture of the formwork must be precise to ensure that the structural shape after concrete pouring matches the design drawings. Precise positioning of the skeleton ensures structural stability, while formwork alignment ensures the accuracy of concrete pouring; the two work closely together during construction.

[0077] The core objective of the skeleton localization section is to use a magnetic encoder combined with the three-dimensional morphological data T from a lidar system. i，j，k (x, y, z) ) The framework of the tower is precisely located. The coordinate points acquired by the magnetic encoder are discrete, denoted as P. n (x n y n , z n ), where n is the number of the encoding point. To locate the stiffness skeleton, a weighting coefficient α is used. n By taking a weighted average of these discrete points, we obtain the skeleton localization model, which is represented as:

[0078]

[0079] Where M(x, y, z) is the center position of the skeleton; α n The weight coefficient for each coding point; N is the total number of coding points; P n (x n y n , z n ) represents the position of the nth encoding point.

[0080] Weighting coefficient α nThe accuracy of data measured by a magnetic encoder may vary at each point, therefore different weighting coefficients α are used. n To characterize the contribution of each coding point to skeleton localization, the weighting coefficients satisfy the following conditions:

[0081]

[0082] The above conditions ensure that the final skeleton position M(x, y, z) is the weighted average of all encoded points, and that the spatial position P of each encoded point is obtained. n (x n y n , z n ) and weight α n The skeleton position M(x, y, z) can then be calculated precisely.

[0083] In some examples, the template pose is adjusted based on the skeleton position to match the skeleton position, including:

[0084] The target attitude adjustment amount is determined based on minimizing the error between the template position and the aforementioned skeleton position;

[0085] Based on the target attitude adjustment amount, a deep learning model optimization algorithm is used to generate an attitude adjustment signal;

[0086] The attitude adjustment process of the template is controlled based on the attitude adjustment signal to match the template attitude with the skeleton position.

[0087] For example, the direction and magnitude of adjustment are determined by calculating the error between the current position of the template and the target position of the skeleton. This error is obtained by comparing the actual position of the template in 3D space with the designed position of the skeleton. Minimizing this error yields the adjustment direction and magnitude of the template's posture, i.e., the target posture adjustment amount. A deep learning optimization algorithm is used to process the target posture adjustment amount calculated in the previous step. Based on the current posture of the template and the theoretical position of the skeleton, the deep learning model generates a posture adjustment signal. This signal guides the control system on how to adjust the template's posture to reduce the error. Based on the adjustment signal generated in the previous step, the control system fine-tunes and optimizes the template's posture according to this signal. This process is a dynamic feedback control; the system continuously monitors the template's position and further optimizes based on the feedback signal until the template posture and skeleton position are precisely matched.

[0088] For example, template pose correction is achieved by minimizing the error between the template position and the skeleton positioning. The following objective function describes the deviation between the template pose and the skeleton position:

[0089]

[0090] Among them, A opt P is the optimal pose matrix for the template; n M represents the location of the encoding point. n (A) represents the theoretical position of the skeleton in pose A; N represents the number of encoding points; ||·|| represents the Euclidean distance.

[0091] The template pose matrix A describes the template's rotation and translation in space. By adjusting A, the template's position is kept consistent with the rigid frame. The optimal template installation pose is found by minimizing the error between the actual position of the template and the theoretical position of the frame. This process is implemented using a deep learning model, which continuously adjusts the template's pose until it reaches the optimum. This includes:

[0092] The input to a deep learning model is the current pose of the template and the theoretical position of the skeleton, and the output is a new pose adjustment signal. By calculating the error each time and feeding it back to the model, the model learns through continuous training how to adjust the position of the template more accurately to reduce the error.

[0093] Input the current template pose A, the theoretical skeleton position M(x, y, z), and the encoding point P. n (x n y n , z n ).

[0094] The output deep learning model predicts a new pose adjustment vector ΔA, which serves as the model's recommended next pose adjustment operation.

[0095] Error calculation: The error is the distance difference between the current position of the template and the theoretical position of the skeleton, expressed as:

[0096]

[0097] The goal of deep learning models is to continuously reduce the error E(A) by adjusting the pose.

[0098] First, prepare the training dataset, which includes different template poses A and their corresponding error values. Through a large number of different pose adjustments and error feedback after each adjustment, the training dataset is formed.

[0099] Deep learning models typically use multi-layer neural networks (e.g., convolutional neural networks or fully connected networks) to process these inputs and outputs. The model architecture can be understood as follows: input layer, which receives the pose A and skeleton position M(x, y, z) of the current template; hidden layer, which learns how to adjust the proposal based on the input pose through several layers of non-linear activation functions; output layer, which outputs a new pose adjustment vector ΔA.

[0100] During the training phase, a large amount of input data is provided, namely different template poses and their corresponding error values. The model continuously adjusts the network weights and parameters through the backpropagation algorithm, so that the predicted pose adjustment ΔA can better reduce the error between the template and the skeleton.

[0101] In practical applications, the trained deep learning model takes the current template pose A and skeleton position M(x, y, z) as input in each iteration and outputs a new pose adjustment ΔA. The system updates the template pose based on this adjustment and then recalculates the error; this process iterates until the error converges to its minimum, yielding the optimal template pose A. opt .

[0102] During template correction, reinforcement learning algorithms are used to optimize the template's mounting posture. Reinforcement learning continuously adjusts the strategy by interacting with the environment and calculating immediate rewards at each step.

[0103] The objective function L(θ) for reinforcement learning is expressed as:

[0104]

[0105] Where s is the current state (the position and orientation of the template); a is the current action (adjusting the orientation of the template); r t γ is the immediate reward (representing the change in error after each posture adjustment); γ is the discount factor, used to represent the importance of future rewards; Q(s, a; θ) is the state-action value function, representing the expected reward obtained by choosing a certain action in the current state.

[0106] The reinforcement learning process is as follows:

[0107] In state s, at each step, the system decides whether to adjust the pose based on the current template pose and skeleton position.

[0108] Action a: The system can choose to rotate or translate the template to adjust its posture.

[0109] Reward r t After each adjustment, the system calculates the error between the template and the skeleton. If the error decreases, the system receives a positive reward; otherwise, it receives a negative reward.

[0110] The strategy is updated by continuously adjusting the attitude. The system will find the optimal installation attitude until the objective function L(θ) converges.

[0111] After skeleton localization and template correction are completed, the data will be uploaded to the cloud to form a new dataset D2, which contains the skeleton position M(x, y, z) and the optimal pose matrix A of the template. optAnd combined with dataset D1 from step S110, it is represented as:

[0112] D2={M(x,y,z),A opt D1}

[0113] Based on magnetic encoder and LiDAR point cloud data, the position of the stiffening frame is determined using a weighted average method. Deep learning and reinforcement learning algorithms are then employed to adjust the template pose to minimize deviation from the frame, ultimately obtaining the optimal template mounting pose matrix. Aopt Dataset D2 It contains all the data for skeleton localization and template correction, providing an important foundation for subsequent steps.

[0114] In some examples, it also includes:

[0115] The aforementioned intelligent hydraulic control system includes a hydraulic pump, a hydraulic cylinder, and a control system;

[0116] The aforementioned control system monitors the operating status of the hydraulic pump and the hydraulic cylinder, and dynamically adjusts the operating parameters of the hydraulic pump based on a PID control algorithm.

[0117] For example, hydraulic pumps and hydraulic cylinders are core components in hydraulic control systems. In the construction of cable-stayed bridge towers, the specific applications of hydraulic pumps and hydraulic cylinders are as follows:

[0118] When the hydraulic pump operates, it generates high-pressure oil, which is delivered to the hydraulic cylinder through pipelines. The piston rod inside the cylinder extends under the pressure of the high-pressure oil, pushing the formwork upwards. Once the formwork reaches the predetermined position, the control system detects the position change via a displacement sensor and adjusts the hydraulic pump output, stopping the hydraulic cylinder's movement. The control system then instructs the hydraulic pump to reverse its operation, causing the piston rod inside the cylinder to retract during the oil return process, lowering the formwork. After descending to the next pouring section, the formwork is fixed again, and concrete pouring begins.

[0119] Hydraulic pumps are typically located in hydraulic pump stations at construction sites, serving as the power center of the hydraulic system. The pumps deliver pressurized fluid to hydraulic cylinders via pipelines. The hydraulic cylinders are usually installed near the formwork or frame, responsible for controlling the position and displacement of the formwork and frame. The piston of the hydraulic cylinder, based on the pressure provided by the hydraulic pump, pushes the formwork or frame to a predetermined position.

[0120] The intelligent hydraulic control system, based on the skeleton and template correction data from step S130, dynamically adjusts the hydraulic pump's operating parameters (such as pressure and displacement, control force of the hydraulic control system, actual template displacement, and actual skeleton displacement) during the pouring process through a feedback control system. The hydraulic control system relies on a PID control algorithm to ensure the stability of the concrete pouring process, enabling precise concrete shaping. This is achieved by comparing the actual displacement X(t) with the reference displacement X.ref The output of the hydraulic system is dynamically adjusted based on the error.

[0121] The basic equations for hydraulic control are as follows:

[0122]

[0123] Where F(t) is the control force applied by the hydraulic system at time t; X(t) is the actual displacement of the template or frame position in the current hydraulic system; X ref K represents the system's reference displacement, i.e., the ideal position that the template or skeleton should reach; p K represents the proportional gain, indicating the effect of the current displacement error on the control force. d The differential gain represents the effect of the rate of change of displacement error on the control force; displacement error X(t) - X ref This is the difference between the system's current displacement and the reference displacement, representing the deviation between the actual and ideal state. The goal of the control system is to minimize this error to zero; the proportional control term K... p (X(t)-X ref Adjust directly based on the current error. Proportional gain K p The differential control term determines the strength of the system's response to errors; the larger the value, the faster the system response. The differential control term takes into account the rate of change of the error, helping the system to anticipate future error trends and make adjustments in advance. This helps to reduce system oscillations caused by rapid changes in error.

[0124] At each time t, the system collects the actual displacement X(t) of the template or skeleton and the control force F(t) of the hydraulic system in real time through sensors, and uploads this data to the cloud. At this time, the dataset will be updated to:

[0125] D3 = {F(t), X(t), D2}

[0126] Data set D3 includes: F(t) is the control force of the hydraulic control system; X(t) is the actual displacement of the template or skeleton; D2 is the data set generated in step S130, including data on skeleton positioning and template correction.

[0127] Throughout the pouring process, the system continuously adjusts the parameters of the hydraulic pump based on the current template position X(t) and the frame position. The data after each adjustment is fed back to the control system in real time to ensure that the next adjustment is more accurate. In this way, the system can dynamically respond to environmental changes and uncertainties in the pouring process, ensuring that the positions of the template and frame remain stable.

[0128] In some examples, the above-mentioned analysis of feature points in concrete surface images to calculate offset errors, based on concrete surface images, includes:

[0129] The above concrete surface image is preprocessed by denoising and illumination correction.

[0130] Feature points of the above concrete surface image were extracted using a convolutional neural network.

[0131] The aforementioned feature points are compared with the preset positions of the template, and the aforementioned offset error is calculated.

[0132] For example, a drone uses a high-precision camera to acquire images of the concrete surface after pouring. The image data includes the surface texture, shape, and other visual features. These images are preprocessed (e.g., denoising, illumination correction) and then input into a deep learning model for further analysis.

[0133] The following image processing techniques are used in the preprocessing process:

[0134] Using nonlocal means or bilateral filtering to remove noise while preserving edge details, as shown below:

[0135]

[0136] Where I(p) is the gray value of pixel p; ω(p) is the weight, representing the similarity between neighboring pixels and the current pixel; and ω(p) is the neighborhood range.

[0137] Use histogram equalization or adaptive histogram equalization to adjust the contrast of an image.

[0138] After image preprocessing, convolutional neural networks (CNNs) are used to analyze feature points in the image and perform feature point detection. CNNs are deep learning models commonly used for image classification and feature extraction. Their structure consists of convolutional layers, pooling layers, and fully connected layers. After convolution and pooling operations, the network outputs a set of feature points {f}. k These feature points represent important locations on the concrete surface, such as surface elevation differences and edges.

[0139] Feature points f extracted by a deep learning model k , and the template reference position X ref A comparison is performed to detect the deviation between the actual poured surface and the reference template. The deviation is calculated using the following formula:

[0140] AP = P actual -P ref

[0141] Among them, P actual This represents the location of the actual surface feature points obtained through feature point detection.

[0142] After the deviation data calculation is completed, the system uploads this data to the cloud and updates dataset D4.

[0143] D4 = {ΔP, D3}

[0144] Data set D4 contains the deviation ΔP, the offset between the actual surface and the reference template; and D3 from the previous step, which contains the hydraulic system data, skeleton positioning, and template correction data.

[0145] For example, adjusting the template installation posture involves adjusting the initial installation position and orientation of the template to match the skeleton, with a focus on the template installation posture A. opt Actual formwork displacement refers to the actual displacement of the formwork during the pouring process caused by external forces (hydraulic control), denoted by X(t). The key is to dynamically adjust the formwork's position through the control system. Formwork reference position refers to the design reference position P of the formwork. ref It is used to compare with the actual poured surface to detect deviations.

[0146] In some examples, the above-mentioned integration of multi-source data to monitor cable tower structures and predict their health status includes:

[0147] Based on the above multi-source data, the tilt state and damage level of the tower column were assessed using big data analysis technology.

[0148] The aforementioned multi-source data, tilt status, and damage level are input into the time series analysis prediction model to obtain the health status of the tower structure.

[0149] For example, the tilt state of the tower column is assessed by analyzing its lateral displacement at different time points. The tilt angle reflects whether the tower column has undergone structural deformation or instability. The formula for the tilt angle is as follows:

[0150]

[0151] Where θ is the tilt angle of the tower column, Δy is the difference in lateral displacement between the top and bottom of the tower column, representing the distance the tower column deviates from the vertical direction, and L is the height of the tower column; Δy is the lateral displacement data of the top of the tower column relative to the bottom collected by the sensor. The change in lateral displacement may be caused by environmental influences (such as wind, earthquake) or structural stress; the vertical height L of the tower column is used to normalize the lateral displacement to ensure the accuracy of the tilt angle calculation.

[0152] The core of long-term structural health assessment of tower columns lies in analyzing changes in structural stress. The degree of damage is expressed by the following formula:

[0153]

[0154] Here, Dd represents the overall damage level, used to assess the health status of the tower column. By analyzing stress changes at multiple stress monitoring points, the potential damage level of the tower column can be estimated. A large Dd value indicates that the structure may have severe stress anomalies, and its health status needs further evaluation; Δσ sen For the stress change at the sen-th monitoring point, the difference between the actual stress value collected by the stress sensor and the reference stress at different monitoring points on the tower column is considered. Stress change is one of the early signals of structural damage, which may be caused by external loads, material aging, etc.; σ ref The reference stress value represents the ideal stress state set during the design of the tower column, indicating the load that the tower column should bear under normal operating conditions; SEN is the total number of sensor monitoring points.

[0155] By integrating previously generated multi-source data, the tower's tilt status, and the degree of damage, this information is input into a time-series analysis predictive model to assess and predict the future structural health of the tower. Time-series analysis is used to detect the tower's changing trends over a period of time and predict potential future structural problems.

[0156] Long Short-Term Memory (LSTM) networks were used as the time series forecasting model. LSTMs can process data with time dependencies, taking a combination of multi-source data, skew state, and impairment level as input, and outputting a prediction of future health status. The LSTM model takes a combination of multi-source data, skew state, and impairment level as input, and outputs a prediction of future health status.

[0157] The input is represented as (D4(T), Dd(t), θ(t)), and the output is represented as:

[0158] The LSTM model processes the input through a recursive structure, as follows:

[0159] h t =LSTM(h t-1 ,D4(t),Dd(t),θ(t))

[0160] Among them, h t The hidden state at time t represents the system's current health information; h t-1 D4(t) represents the hidden state at the previous time step; D4(t) represents the multi-source input data at the current time step.

[0161]

[0162] In the formula, To predict future health status.

[0163] The health status of a cable tower typically includes assessments of various aspects, such as overall structural stability, stress distribution, tilt state, and degree of material damage. In practical applications, the health status of cable towers can be categorized into the following common situations:

[0164] Under normal conditions, the tower's tilt angle, stress distribution, and material condition are all within the design range, with no obvious signs of damage. The tilt angle is close to zero, the lateral displacement is very small, and the structure remains vertical. The degree of damage is close to zero. The material stress is stable, and the stress changes at all monitoring points are small. The tower is in good condition, can normally bear the design load, and has good structural stability. The system does not require maintenance measures.

[0165] Localized damage occurs when stress concentrations or micro-cracks appear in certain parts of the cable tower, leading to localized damage. Common causes include localized material aging due to long-term use, stress concentration points, or environmental stress. Even with minimal changes in the tilt angle, abnormal stress changes at local monitoring points may indicate minor structural shifts. Significantly increased damage, with large stress variations in specific areas potentially exceeding the material's normal load-bearing capacity, necessitates targeted maintenance measures, such as repairing damaged areas, strengthening local structures, and continuous monitoring of the affected area to prevent damage spread.

[0166] Severe damage or structural instability indicates a significant deterioration in the structural health of the tower, with clear risks of material failure, stress overload, or tilting instability. The tilt angle is extremely large, potentially exceeding design limits, posing a risk of structural collapse. The damage is extremely severe, with drastic stress changes at multiple monitoring points, possibly accompanied by significant material fracture, cracking, or fatigue. In such cases, immediate cessation of structural use and emergency reinforcement or reconstruction are necessary. Severe damage or instability signals the end of the structure's lifespan, and failure to repair in a timely manner could lead to catastrophic consequences.

[0167] Please see Figure 2 The diagram below illustrates the structure of a digital monitoring device for cable-stayed bridge towers, as provided in this embodiment of the application. The device includes:

[0168] Acquisition unit 21 is used to acquire initial data, wherein the initial data includes soda environment data and three-dimensional morphological information;

[0169] Positioning unit 22 is used to locate the skeleton position based on magnetic encoding data and the above-mentioned three-dimensional morphological information;

[0170] Matching unit 23 is used to adjust the template posture based on the skeleton position so that the template posture matches the skeleton position, and to obtain correction data during the adjustment process;

[0171] Monitoring unit 24 is used to monitor the pouring process based on the intelligent hydraulic control system and obtain dynamic data;

[0172] The identification unit 25 is used to analyze the feature points of the concrete surface image based on the concrete surface image, calculate the deviation error, and determine the deviation data.

[0173] The prediction unit 26 is used to integrate multi-source data to monitor the cable tower structure and predict the health status of the cable tower structure. The multi-source data includes the initial data, the correction data, the dynamic data, and the offset data.

[0174] Please see Figure 3 This application also provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor. When the processor 320 executes the computer program 311, it implements the steps of any of the above-described methods for digital monitoring of cable-stayed bridge towers.

[0175] Since the electronic device described in this embodiment is the device used to implement the digital monitoring device for cable-stayed bridge towers in this application embodiment, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the method described in this application embodiment. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. Any device used by those skilled in the art to implement the method in this application embodiment is within the scope of protection of this application.

[0176] In practice, when the computer program 311 is executed by the processor, it can implement any of the embodiments corresponding to the first aspect.

[0177] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0178] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-readable program code.

[0179] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0180] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0181] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0182] This application also provides a computer program product, which includes computer software instructions that, when executed on a processing device, cause the processing device to perform... Figure 1 The flowchart of the digital monitoring method for cable-stayed bridge towers in the corresponding embodiment.

[0183] A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0184] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0185] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between devices or units, and may be electrical, mechanical, or other forms.

[0186] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0187] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0188] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0189] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

[0190] Although preferred embodiments have been described in this specification, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of this specification. Clearly, those skilled in the art can make various alterations and modifications to this specification without departing from its spirit and scope. Thus, if such modifications and modifications fall within the scope of the claims and their equivalents, this specification also intends to include such modifications and modifications.

Claims

1. A digital monitoring method for cable-stayed bridge towers, characterized in that, The method includes: Acquire initial data, which includes cable tower environmental data and three-dimensional morphological information; the cable tower environmental data includes a sensor data matrix containing time series, the sensor data matrix including temperature data, humidity data, stress data and vibration data; the three-dimensional morphological information is point cloud data transformed from a three-dimensional coordinate set, the three-dimensional coordinate set being a set of three-dimensional coordinates of cable tower nodes obtained by scanning the cable tower with lidar; The skeleton position is located based on the magnetic encoding data and the three-dimensional morphological information. The template pose is adjusted based on the skeleton position to match the template pose with the skeleton position, and correction data is obtained during the adjustment process. The step of adjusting the template pose based on the skeleton position to match the template pose with the skeleton position includes: The target attitude adjustment amount is determined based on minimizing the error between the template position and the skeleton position; Based on the target attitude adjustment amount, a deep learning model optimization algorithm is used to generate an attitude adjustment signal; The template posture adjustment process is controlled based on the posture adjustment signal to match the template posture with the skeleton position; The pouring process is monitored using an intelligent hydraulic control system to obtain dynamic data; The intelligent hydraulic control system includes a hydraulic pump, a hydraulic cylinder, and a control system. The control system monitors the working status of the hydraulic pump and the hydraulic cylinder, and dynamically adjusts the working parameters of the hydraulic pump based on a PID control algorithm. Based on the concrete surface image, the feature points of the concrete surface image are analyzed, the offset error is calculated, and the offset data is determined. The process of analyzing feature points of the concrete surface image and calculating the offset error based on the concrete surface image includes: Preprocessing operations such as denoising and illumination correction are performed on the concrete surface image; Feature points of the concrete surface image were extracted using a convolutional neural network. The feature points are compared with the preset positions of the template to calculate the offset error; Integrating multi-source data to monitor the cable tower structure and predict the health status of the cable tower structure, wherein the multi-source data includes the initial data, the correction data, the dynamic data, and the offset data; The method of integrating multi-source data to monitor the cable tower structure and predict its health status includes: Based on the multi-source data, the tilt state and damage level of the tower column were assessed using big data analytics. The multi-source data, the tilt state, and the degree of damage are input into a time series analysis prediction model to obtain the health status of the tower structure. Also includes: The magnetic encoded data is determined based on the coordinate information collected by the magnetic encoder; The skeleton position is calculated based on the spatial location and weight of the encoding points.

2. A digital monitoring device for cable-stayed bridge towers, used to execute the digital monitoring method for cable-stayed bridge towers as described in claim 1, characterized in that, include: An acquisition unit is used to acquire initial data, wherein the initial data includes the tower environment data and three-dimensional morphological information; The positioning unit is used to locate the skeleton position based on the magnetic encoding data and the three-dimensional morphological information; The matching unit is used to adjust the template pose based on the skeleton position so that the template pose matches the skeleton position, and to acquire correction data during the adjustment process; The monitoring unit is used to monitor the pouring process based on the intelligent hydraulic control system and obtain dynamic data. The identification unit is used to analyze the feature points of the concrete surface image based on the concrete surface image, calculate the deviation error, and determine the deviation data. The prediction unit is used to integrate multi-source data to monitor the cable tower structure and predict the health status of the cable tower structure, wherein the multi-source data includes the initial data, the correction data, the dynamic data, and the offset data.

3. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program stored in the memory to implement the steps of the digital monitoring method for cable-stayed bridge towers as described in claim 1.

4. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the digital monitoring method for cable-stayed bridge towers as described in claim 1.

Images