A coordinated control method and system of multi-scale hybrid strategy for bridge damping
Through the coordinated control method of multi-scale hybrid strategy, combined with graph neural network and multiple data processing technologies, the damper configuration is optimized, and the problems of insufficient frequency range coverage and difficulty in dynamic adjustment in the existing technology are solved, and the full frequency range of the bridge is realized Effective control of vibration improves the safety and durability of the bridge structure.
Patent Information
- Application Number
- CN202411288880.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-14
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-09-14
AI Technical Summary
The existing bridge vibration damping technology is difficult to cover vibrations in the entire frequency range and cannot be dynamically adjusted, resulting in unsatisfactory control effect and safety risks.
The coordinated control method of multi-scale hybrid strategy is adopted to fuse sensor data through graph neural networks to perform multi-source vibration data matrix processing, combining wavelet transformation, empirical modal decomposition, Vigner quasi-probability distribution method, core principal component analysis and density space Cluster, identify damping requirements, and optimize damper configuration using simulated annealing and genetic algorithms to achieve coordinated control of vibrations in different frequency ranges.
Effectively cover the full frequency range vibration of all parts of the bridge, improve vibration control effect, and enhance the safety and durability of the bridge structure.
Smart Images

Figure CN118797322B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of bridge engineering, and in particular to a coordinated control method and system of a multi-scale hybrid strategy for bridge damping. Background Art
[0002] With the continuous development of bridge engineering technology and the increase of traffic load, the safety and durability of bridge structures have become the focus of attention in the engineering community. Among them, bridge vibration control, as an important means to ensure the safety of bridge structures, has been widely studied.
[0003] However, current technologies still have some problems with bridge vibration damping: a single damping technology is difficult to cover the full-frequency range of vibrations in different parts of the bridge structure, resulting in unsatisfactory control effects; in addition, existing damping control strategies usually rely on dampers with fixed parameters and cannot be dynamically adjusted and optimized according to real-time vibration data, thus failing to achieve effective control of complex vibration environments.
[0004] The existing technology has tried to solve these problems by increasing the number of dampers or improving the performance of a single damper, but the effect is limited. The reason is that the traditional damping control method mainly relies on passive dampers, which makes it difficult to effectively control multiple frequency components. In addition, since different parts of the bridge (such as the main beam, piers, and bridge deck) have different vibration characteristics, the existing method is difficult to synchronously and accurately control the vibration of all parts. Therefore, the above problems make the bridge structure face greater safety risks in long-term operation, and there is an urgent need for a bridge vibration damping control method that can dynamically adapt to different parts and frequency ranges. Summary of the invention
[0005] The purpose of the present invention is to provide a coordinated control method and system of a multi-scale hybrid strategy for bridge damping to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:
[0006] In a first aspect, the present application provides a coordinated control method of a multi-scale hybrid strategy for bridge damping, comprising:
[0007] Obtain vibration signals from different parts of the bridge. According to the principle of bridge vibration distribution, use sensors to collect vibration signals, and perform fusion processing through the graph neural network algorithm to obtain a multi-source vibration data matrix. Different parts include the main beam, piers and bridge deck;
[0008] Based on the multi-source vibration data matrix, after the first decomposition process, the vibration signal components of each scale are obtained. After the second decomposition process, the signal is decomposed into multiple intrinsic mode functions to obtain the refined frequency components.
[0009] According to the frequency components obtained by multi-scale decomposition, the Wigner quasi-probability distribution method is used for frequency domain analysis. The kernel principal component analysis is applied to reduce the dimension of the frequency domain analysis results, and the instantaneous frequency and amplitude of the intrinsic mode function are analyzed to obtain the frequency domain distribution of vibration energy at each scale. The Hilbert-Huang transform method is used to analyze the instantaneous frequency and amplitude of the intrinsic mode function to obtain the vibration energy distribution.
[0010] The density-based spatial clustering algorithm divides the frequency range of vibration energy distribution into several clusters. Each cluster corresponds to a main vibration frequency range. The distribution of the main vibration energy in different frequency ranges is obtained, and then the damping requirements of each frequency range are determined.
[0011] According to various damping requirements, the optimal damper configuration scheme is searched in the parameter space after being processed by the simulated annealing algorithm, with the goal of minimizing the vibration energy in each frequency range. The genetic algorithm is then used for refined search to achieve the optimal solution for the coordinated control of the multi-scale hybrid strategy for vibrations in different frequency ranges.
[0012] Preferably, the vibration signals of different parts of the bridge are obtained by using sensors to collect vibration signals according to the vibration distribution principle of the bridge, and the multi-source vibration data matrix is obtained by fusion processing through a graph neural network algorithm, which includes:
[0013] Using accelerometers and strain gauges, capture the first original vibration signal data of the deformation of the main beam structure and the dynamic response due to vibration, using displacement sensors and accelerometers, monitor the displacement of the pier and the second original vibration signal data with vibration characteristics, using accelerometers and strain gauges, monitor the third original vibration signal data of the dynamic response caused by the vehicle load and environmental vibration on the bridge deck, wherein each data point in the first original vibration signal data, the second original vibration signal data and the third original vibration signal data corresponds to time and position;
[0014] According to the first original vibration signal data, the second original vibration signal data and the third original vibration signal data, after topological modeling processing, sensor data nodes are constructed, each sensor is regarded as a node, and a graph structure of a bridge is formed to obtain a graph structure including the position, type and vibration signal data of each node;
[0015] Based on the constructed graph structure and graph neural network algorithm, the nodes are embedded. After node feature extraction, the vibration signal data of each node is converted into a feature vector to obtain a node embedding vector containing spatiotemporal features.
[0016] According to the node embedding vector, the message passing mechanism in the graph neural network is used to perform multiple iterative calculations. After the hierarchical transmission processing of the graph network, the embedding vector of each node is fused with the embedding vectors of the adjacent nodes. Through the neighborhood aggregation function processing, the multi-source vibration data matrix is obtained.
[0017] Preferably, the multi-source vibration data matrix is subjected to a first decomposition process to obtain vibration signal components of each scale, and the signal is decomposed into multiple intrinsic mode functions through a second decomposition process to obtain refined frequency components, including:
[0018] The multi-source vibration data matrix is processed by discrete wavelet transform, and the mother wavelet suitable for the characteristics of bridge vibration signals is selected. Each column vector in the multi-source vibration data matrix is decomposed, and the signal is decomposed into detail coefficients and approximation coefficients at different scales to obtain vibration signal components at each scale, where the vibration signal components include different vibration characteristic data of the main beam, pier and bridge deck;
[0019] Based on the vibration signal components of each scale, these components are reorganized into a multi-scale signal component matrix through matrix construction processing;
[0020] Traverse each signal component in the multi-scale signal component matrix, and select all the maximum and minimum values in the vibration signal components of each scale, construct the upper envelope and the lower envelope through the maximum and minimum points respectively, and calculate the mean of the upper envelope and the lower envelope; subtract the mean value from the signal to obtain the preliminary intrinsic mode function, iterate repeatedly until the preset conditions of the intrinsic mode function are met, separate the intrinsic mode function from the signal, and use the remaining part as a new signal. Circular calculation until there is no intrinsic mode function in the remaining signal, and obtain the refined frequency components, where the refined frequency components include the vibration frequency components of the main beam, piers and bridge deck, and the calculation formulas of the upper envelope and the lower envelope are as follows:
[0021]
[0022] Where u(t) is the upper envelope, l(t) is the lower envelope, and m(t) is the mean.
[0023] Preferably, the frequency components obtained by multi-scale decomposition are subjected to frequency domain analysis using the Wigner quasi-probability distribution method, and the kernel principal component analysis is applied to reduce the dimension of the frequency domain analysis results, and the instantaneous frequency and amplitude of the intrinsic mode function are analyzed to obtain the frequency domain distribution of the vibration energy of each scale, and the Hilbert-Huang transform method is applied to analyze the instantaneous frequency and amplitude of the intrinsic mode function to obtain the vibration energy distribution, which includes:
[0024] Based on the frequency components obtained by matrix decomposition of multi-source vibration data, the time-frequency energy distribution of the main beam vibration signal, the time-frequency energy distribution of the pier vibration signal and the time-frequency energy distribution of the bridge deck vibration signal are obtained through the Wigner quasi-probability distribution method, and are recorded as the main beam time-frequency distribution, the pier time-frequency distribution and the bridge deck time-frequency distribution respectively. The calculation formula of the Wigner quasi-probability distribution method is as follows;
[0025]
[0026] In the formula, W in it is the Wigner-Ville distribution, x is the time signal, t is the time variable, f is a frequency variable, x(t) represents the signal value at time t, Expressed as the complex conjugate of the signal at time t, is expressed as an integral variable, representing the time delay, represents the exponential term, indicating the frequency f and time delay The complex exponential function of represents the integration of time delay;
[0027] According to the time-frequency distribution of the main beam, the time-frequency distribution of the bridge pier and the bridge deck, the dimension is reduced by kernel principal component analysis to obtain the frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the bridge deck after dimension reduction;
[0028] Based on the frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the frequency domain distribution data of the bridge deck, the Hilbert-Huang transform method is used to analyze the instantaneous frequency and amplitude of the intrinsic mode function, and the instantaneous frequency and amplitude data of the main beam, the instantaneous frequency and amplitude data of the bridge pier and the instantaneous frequency and amplitude data of the bridge deck are obtained;
[0029] According to the instantaneous frequency and amplitude data of the main beam, the instantaneous frequency and amplitude data of the bridge piers and the bridge deck, the instantaneous energy of the main beam, the bridge piers and the bridge deck is calculated, and the frequency range of the instantaneous frequency corresponding to the main beam, the bridge piers and the bridge deck is divided into several frequency bands. The instantaneous energy in each frequency band is integrated to obtain the total energy in the frequency band, that is, the total energy of the main beam, the total energy of the bridge piers and the total energy of the bridge deck. The concentration of each total energy is analyzed to determine the main concentrated frequency range, identify the frequency band with higher energy, determine the vibration characteristics of the frequency range, and then obtain the vibration energy distribution.
[0030] Preferably, the method searches for the best damper configuration scheme in the parameter space according to each damping requirement through simulated annealing algorithm, with the goal of minimizing the vibration energy in each frequency range, including:
[0031] According to the vibration energy distribution of the main beam, piers and bridge deck, the damping requirements of the main beam, piers and bridge deck in different frequency ranges are calculated, where the damping requirements are proportional to the energy;
[0032] According to the damping requirements and actual needs, the value range of the damper parameters is set, the damper parameters are randomly generated in each frequency range, and the initial damper configuration scheme is randomly generated in the parameter space;
[0033] Based on the initial damper configuration scheme and the vibration energy distribution of each bridge part, the simulated annealing algorithm is used to search within the initial damper configuration scheme. In each iteration, a new configuration scheme is generated according to the current configuration scheme, and its energy objective function value is calculated. After multiple iterations, until the damper configuration scheme with the minimum vibration energy is found, it is recorded as the optimal damper configuration scheme searched.
[0034] In a second aspect, the present application also provides a coordinated control system of a multi-scale hybrid strategy for bridge damping, comprising:
[0035] Acquisition module: used to obtain vibration signals from different parts of the bridge. According to the vibration distribution principle of the bridge, the vibration signals are collected by sensors and fused through the graph neural network algorithm to obtain a multi-source vibration data matrix. The different parts include the main beam, piers and bridge deck;
[0036] Decomposition module: Based on the multi-source vibration data matrix, after the first decomposition process, the vibration signal components of each scale are obtained, and after the second decomposition process, the signal is decomposed into multiple intrinsic mode functions to obtain the refined frequency components;
[0037] Analysis module: It is used to perform frequency domain analysis using Wigner quasi-probability distribution method based on the frequency components obtained by multi-scale decomposition, apply kernel principal component analysis to reduce the dimension of the frequency domain analysis results, analyze the instantaneous frequency and amplitude of the intrinsic mode function, obtain the frequency domain distribution of vibration energy at each scale, and apply Hilbert-Huang transform method to analyze the instantaneous frequency and amplitude of the intrinsic mode function to obtain the vibration energy distribution;
[0038] Determination module: It is used to divide the frequency range of vibration energy distribution into several clusters based on density-based spatial clustering algorithm. Each cluster corresponds to a main vibration frequency range, and the distribution of main vibration energy in different frequency ranges is obtained, so as to determine the damping requirements of each frequency range.
[0039] Search module: It is used to search for the best damper configuration scheme in the parameter space according to various damping requirements after processing by simulated annealing algorithm, with the goal of minimizing the vibration energy in each frequency range, and apply genetic algorithm for refined search to achieve the optimal solution for coordinated control of multi-scale hybrid strategies for vibrations in different frequency ranges.
[0040] The beneficial effects of the present invention are:
[0041] Based on the principle of bridge vibration distribution, the present invention obtains vibration signals from different parts of the bridge, uses a variety of sensors such as accelerometers, displacement sensors and strain gauges for data collection, and fuses multi-source sensor data through graph neural networks to construct a multi-source vibration data matrix containing spatiotemporal characteristics. In the data processing stage, wavelet transform and empirical mode decomposition are used to perform multi-scale decomposition of the vibration signal to obtain refined frequency components; then, the frequency domain analysis is performed using the Wigner quasi-probability distribution method and kernel principal component analysis to determine the main vibration energy distribution in each frequency range; and the density-based spatial clustering algorithm is used to divide the vibration energy distribution into frequency ranges to identify the damping requirements of each frequency range; finally, the simulated annealing algorithm and genetic algorithm are used to optimize the damper configuration scheme, dynamically adjust the damper parameters, and realize the coordinated control of the multi-scale hybrid strategy for vibrations in different frequency ranges; therefore, the present method can effectively cover the full frequency range vibrations of various parts of the bridge, improve the vibration control effect, and enhance the safety and durability of the bridge structure.
[0042] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.
[0044] Figure 1 It is a schematic flow chart of a coordinated control method of a multi-scale hybrid strategy for bridge damping described in an embodiment of the present invention;
[0045] Figure 2 Schematic diagram of the coordinated control system structure of the multi-scale hybrid strategy for bridge damping described in an embodiment of the present invention.
[0046] In the figure: 701, acquisition module; 7011, monitoring unit; 7012, first processing unit; 7013, second processing unit; 7014, calculation fusion unit; 702, decomposition module; 7021, acquisition unit; 7022, third processing unit; 7023, loop unit; 703, analysis module; 7031, fourth processing unit; 7032, dimensionality reduction unit; 7033, analysis unit; 7034, calculation energy unit; 704, determination module; 705, search module; 7051, calculation demand unit; 7052, generation unit; 7053, iteration unit. DETAILED DESCRIPTION
[0047] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0048] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.
[0049] Embodiment 1:
[0050] This embodiment provides a coordinated control method of a multi-scale hybrid strategy for bridge damping.
[0051] See also Figure 1 , the figure shows that the method includes step S100, step S200, step S300, step S400 and step S500.
[0052] S100. Obtain vibration signals from different parts of the bridge. According to the vibration distribution principle of the bridge, use sensors to collect vibration signals, and perform fusion processing through a graph neural network algorithm to obtain a multi-source vibration data matrix, in which different parts include the main beam, piers and bridge deck.
[0053] It can be understood that the step S100 includes S101, S102, S103 and S104, wherein:
[0054] S101, using an accelerometer and a strain gauge to capture the first original vibration signal data of the deformation of the main beam structure and the dynamic response due to vibration, using a displacement sensor and an accelerometer to monitor the displacement of the bridge pier and the second original vibration signal data with vibration characteristics, and using an accelerometer and a strain gauge to monitor the third original vibration signal data of the dynamic response caused by the vehicle load on the bridge deck and the environmental vibration, wherein each data point in the first original vibration signal data, the second original vibration signal data and the third original vibration signal data corresponds to time and position;
[0055] It should be noted that the main beam is one of the main load-bearing structures of the bridge, bearing huge vehicle loads and environmental vibrations. Therefore, accelerometers and strain gauges are installed on the main beam to monitor its structural deformation and dynamic response. The accelerometer can record the vibration acceleration of the main beam at different frequencies, while the strain gauge can record the strain changes of the main beam. By synchronously recording the acceleration and strain data, the first original vibration signal data of the main beam under vibration can be obtained. Each point of the data corresponds to a specific time and position, thereby recording in detail the dynamic characteristics of the main beam at different times; the pier is a vertical structure supporting the bridge, and its stability is crucial to the overall safety of the bridge. The displacement sensor can accurately measure the displacement changes of the pier, especially in large vibration events such as earthquakes. In the device, the accelerometer records the acceleration response of the bridge pier under the action of vibration. These data combined can reflect the vibration characteristics and displacement changes of the bridge pier, and generate the second original vibration signal data. These data also correspond to specific time and position, ensuring the comprehensiveness and accuracy of the monitoring; the bridge deck is the part that directly bears the vehicle load, and its vibration characteristics have an important impact on the overall performance of the bridge. The accelerometer records the acceleration response of the bridge deck under vehicle load and environmental vibration, and the strain gauge records the strain changes of the bridge deck. Through the data of these sensors, we can understand in detail the dynamic response of the bridge deck under actual working conditions, and generate the third original vibration signal data. Each data point corresponds to a specific time and position, ensuring the integrity and accuracy of the data.
[0056] It is understandable that, combined with actual application situations, the arrangement of sensors and the data acquisition process need to fully consider the specific structure and environmental conditions of the bridge. For example, in a long-span bridge, the vibration signals of the main beam and piers may be significantly affected by wind loads, so special attention needs to be paid to the wind vibration effect when arranging sensors; at the same time, the dynamic changes of vehicle loads will also produce different vibration responses on the bridge deck, so the acquisition frequency and sensitivity of the sensor need to adapt to this dynamic change. In addition, during the data processing process, data denoising and correction need to be considered to improve the quality and reliability of the data.
[0057] S102, constructing sensor data nodes through topological modeling processing according to the first original vibration signal data, the second original vibration signal data, and the third original vibration signal data, treating each sensor as a node, and forming a graph structure of a bridge, thereby obtaining a graph structure including the position, type, and vibration signal data of each node;
[0058] Specifically, according to the first original vibration signal data, the second original vibration signal data and the third original vibration signal data, the accelerometers and strain gauges on the main beam, the displacement sensors and accelerometers on the piers, and the accelerometers and strain gauges on the bridge deck are respectively defined as independent nodes, each node records its specific position, sensor type and corresponding vibration signal data, and these nodes are connected according to the physical structure of the bridge and the sensor layout, wherein the nodes on the main beam, piers and bridge deck establish adjacency relationships according to the actual spatial distribution and physical connection methods, forming a topological graph reflecting the overall structure of the bridge, and the connection between each node represents the interaction and connection between different parts of the bridge structure, and the vibration signal data of each node is integrated into the graph structure through topological modeling processing.
[0059] In this step, the vibration data of the node includes not only the time series data, but also the relative position of the node in the topological graph and its connection relationship with adjacent nodes. Finally, a graph structure containing node position, type and vibration signal data is formed, which can fully reflect the dynamic characteristics of the bridge and provide a basis for subsequent graph neural network processing.
[0060] It should be noted that the graph structure accurately reflects the physical and geometric properties of the bridge, can comprehensively capture the dynamic responses of different parts of the bridge, integrate the data of different types of sensors into a unified graph structure, and provide a systematic data framework to facilitate subsequent graph neural network processing. The graph structure contains time series data and spatial relationships, and can dynamically reflect the vibration characteristics of the bridge under different conditions. Through precise node definitions and connection relationships, the accuracy and reliability of the data are ensured, providing a solid foundation for subsequent vibration signal analysis and vibration energy distribution.
[0061] S103, embedding the nodes based on the constructed graph structure and graph neural network algorithm, extracting node features, converting the vibration signal data of each node into a feature vector, and obtaining a node embedding vector containing spatiotemporal features;
[0062] S104. According to the node embedding vector, multiple iterative calculations are performed using the message passing mechanism in the graph neural network. After hierarchical transmission processing of the graph network, the embedding vector of each node is fused with the embedding vector of the adjacent node, and processed through the neighborhood aggregation function to obtain a multi-source vibration data matrix.
[0063] It should be noted that, according to the constructed graph structure, the vibration signal data of each node is feature extracted. Specifically, the graph convolution layer in the graph neural network is used to represent the vibration signal data of each node as a feature vector. After multi-layer graph convolution processing, the embedding vector of each node is obtained. These embedding vectors not only contain the vibration signal characteristics of the node itself, but also combine the spatiotemporal information of adjacent nodes. The result of this step is a node embedding vector containing spatiotemporal features. According to the node embedding vector, the message passing mechanism in the graph neural network is used to perform multiple iterative calculations. The message passing mechanism fuses the embedding vector of each node with the embedding vector of the adjacent node through the neighborhood aggregation function (such as sum, average, maximum value, etc.), and updates the feature vector of the node. Among them, the message passing mechanism can be expressed as:
[0064]
[0065] In the formula, Aggregate represents the aggregation function, For Node i The aggregated value or feature, Represents a collection containing all nodes i The values or features of adjacent nodes ,in For Node i 's neighborhood.
[0066] Through multiple iterative calculations, the embedded vectors of all nodes are fused and processed, and finally a multi-source vibration data matrix is obtained. This matrix contains the comprehensive characteristics of vibration signals of different parts of the bridge (main beam, piers and bridge deck) and takes into account the fusion of spatiotemporal information.
[0067] Therefore, in this step, the graph neural network algorithm is used to extract and embed the features of the vibration signal data of each node, which integrates the spatiotemporal characteristics of the nodes and improves the accuracy of data representation. The message passing mechanism is used to integrate the feature information of different nodes and their adjacent nodes, and the spatiotemporal dynamic characteristics of the bridge structure are captured. Through multiple iterative calculations, a multi-source vibration data matrix is constructed, which provides a high-quality data foundation for subsequent vibration signal analysis.
[0068] S200, based on the multi-source vibration data matrix, after the first decomposition process, the vibration signal components of each scale are obtained, and after the second decomposition process, the signal is decomposed into multiple intrinsic mode functions to obtain refined frequency components.
[0069] It can be understood that step S200 includes S201, S202 and S203, wherein:
[0070] S201, processing the multi-source vibration data matrix by discrete wavelet transform, selecting a mother wavelet suitable for the characteristics of the bridge vibration signal, decomposing each column vector in the multi-source vibration data matrix, decomposing the signal into detail coefficients and approximation coefficients at different scales, and obtaining vibration signal components at each scale, wherein the vibration signal components include different vibration characteristic data of the main beam, the pier and the bridge deck;
[0071] It should be noted that according to the characteristics of the bridge vibration signal, a suitable mother wavelet is selected, where the selection of the mother wavelet affects the decomposition accuracy and the extraction effect of the signal features. Each column vector in the multi-source vibration data matrix is decomposed by wavelet to obtain vibration signal components of various scales, including different vibration characteristic data of the main beam, pier and bridge deck. Through discrete wavelet transform, multi-scale decomposition of vibration signals is achieved, vibration characteristics in different frequency ranges are captured, and the accuracy of signal analysis is improved.
[0072] S202, based on the vibration signal components of each scale, these components are reorganized into a multi-scale signal component matrix through matrix construction processing;
[0073] S203, traverse each signal component in the multi-scale signal component matrix, and select all the maximum and minimum values in the vibration signal components of each scale, respectively construct the upper envelope and the lower envelope through the maximum and minimum points, and calculate the mean of the upper envelope and the lower envelope; subtract the mean from the signal to obtain a preliminary intrinsic mode function, iterate repeatedly until the preset conditions of the intrinsic mode function are met, separate the intrinsic mode function from the signal, and use the remaining part as a new signal, and calculate cyclically until there is no intrinsic mode function in the remaining signal, and obtain the refined frequency components, wherein the refined frequency components include the vibration frequency components of the main beam, the pier and the bridge deck, and the calculation formulas of the upper envelope and the lower envelope are as follows:
[0074]
[0075] Where u(t) is the upper envelope, l(t) is the lower envelope, and m(t) is the mean.
[0076] It should be noted that the detail coefficients and approximation coefficients of each scale are reorganized into a multi-scale signal component matrix S, and each signal component in the matrix S is traversed to extract the characteristics of each vibration signal.
[0077] In step S203, for each signal component, the upper envelope and the lower envelope are constructed by the maximum point and the minimum point, respectively, and then the mean m(t) is calculated and subtracted from the signal to obtain the preliminary intrinsic mode function h(t). The above steps are repeated until the preset conditions of the intrinsic mode function are met (such as the mean is close to zero and the standard deviation is small). After that, the intrinsic mode function is separated from the signal, and the remaining part is used as a new signal. The calculation is repeated until there is no intrinsic mode function in the remaining signal, and the refined frequency components are obtained. These components include the vibration frequency components of the main beam, piers and bridge deck. Therefore, when monitoring long-span bridges, the use of wavelet transform can effectively separate the vibration characteristics in different frequency ranges, helping to identify local damage or abnormalities of the bridge. The empirical mode decomposition further refines these features and provides detailed frequency component information, which helps to accurately locate and evaluate the health status of the structure. It can be understood that the signal components in the multi-source vibration data matrix are uniformly processed to construct a multi-scale signal component matrix containing rich time-frequency information, which provides a high-quality data basis for subsequent vibration signal analysis. This method has important application value in actual bridge monitoring and can be used to monitor and evaluate the structural health status of bridges in real time, improving bridge safety and maintenance efficiency.
[0078] S300. Based on the frequency components obtained by multi-scale decomposition, the Wigner quasi-probability distribution method is used to perform frequency domain analysis. The kernel principal component analysis is applied to reduce the dimension of the frequency domain analysis results, and the instantaneous frequency and amplitude of the intrinsic mode function are analyzed to obtain the frequency domain distribution of vibration energy at each scale. The Hilbert-Huang transform method is applied to analyze the instantaneous frequency and amplitude of the intrinsic mode function to obtain the vibration energy distribution.
[0079] It can be understood that step S300 includes S301, S302, S303 and S304, wherein:
[0080] S301. Based on the frequency components obtained by matrix decomposition of multi-source vibration data, the time-frequency energy distribution of the main beam vibration signal, the time-frequency energy distribution of the pier vibration signal and the time-frequency energy distribution of the bridge deck vibration signal are obtained by Wigner quasi-probability distribution method, and they are recorded as main beam time-frequency distribution, pier time-frequency distribution and bridge deck time-frequency distribution respectively. The calculation formula of the Wigner quasi-probability distribution method is as follows;
[0081]
[0082] In the formula, W in it is the Wigner-Ville distribution, x is the time signal, t is the time variable, f is a frequency variable, x(t) represents the signal value at time t, Expressed as the complex conjugate of the signal at time t, is expressed as an integral variable, representing the time delay, represents the exponential term, indicating the frequency f and time delay The complex exponential function of represents the integration of time delay;
[0083] S302, performing dimensionality reduction by kernel principal component analysis according to the time-frequency distribution of the main beam, the time-frequency distribution of the bridge pier and the time-frequency distribution of the bridge deck, and obtaining frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the bridge deck after dimensionality reduction;
[0084] It should be noted that according to the distribution characteristics of the data, a suitable kernel function is selected. Common kernel functions include Gaussian kernel, linear kernel and polynomial kernel. In the dimensionality reduction processing of bridge vibration signals, Gaussian kernel is often selected because it can capture nonlinear characteristics. The kernel matrix is constructed using the selected kernel function, and the kernel matrix K is centralized to ensure the stability and accuracy of the dimensionality reduction result. The eigenvalue decomposition of the centralized kernel matrix K′ is performed to obtain eigenvalues and eigenvectors. The eigenvectors corresponding to the first d largest eigenvalues are selected for dimensionality reduction. The original high-dimensional data is mapped to a new low-dimensional space to obtain the reduced-dimensional data. After the KPCA dimensionality reduction processing, the frequency domain distribution data of the main beam, pier and bridge deck are obtained. These data retain the main time-frequency characteristics, but the dimension is significantly reduced, which is convenient for subsequent processing and analysis. Therefore, KPCA has good anti-noise performance, can filter out noise and irrelevant information in the original data, and improve the reliability of data analysis.
[0085] S303, based on the frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the frequency domain distribution data of the bridge deck, using the Hilbert-Huang transform method, perform instantaneous frequency and amplitude analysis on the intrinsic mode function to obtain instantaneous frequency and amplitude data of the main beam, the instantaneous frequency and amplitude data of the bridge pier and the instantaneous frequency and amplitude data of the bridge deck;
[0086] S304. Calculate the instantaneous energy of the main beam, bridge piers and bridge deck according to the instantaneous frequency and amplitude data of the main beam, bridge piers and bridge deck, and divide the frequency range of the instantaneous frequency corresponding to the main beam, bridge piers and bridge deck into several frequency bands. Integrate the instantaneous energy in each frequency band to obtain the total energy in the frequency band, i.e. the total energy of the main beam, the total energy of the piers and the total energy of the bridge deck. Analyze the concentration of each total energy, determine the main concentrated frequency range, identify the frequency band with higher energy, determine the vibration characteristics of the frequency range, and then obtain the vibration energy distribution.
[0087] It should be noted that, first, the IMFs of each bridge are extracted from the frequency domain distribution data of the main beam, pier and bridge deck after dimensionality reduction. These IMFs represent the intrinsic characteristics of the signal and can reflect the local frequency and amplitude changes of the signal. Hilbert transform is performed on each IMF to obtain the corresponding analytical signal:
[0088]
[0089] Where x(t) is the original signal, x^(t) is its Hilbert transform, H(t) is the analytical signal, and j is the imaginary unit.
[0090] The analytical signal can be used to calculate the instantaneous amplitude A(t) and the instantaneous phase φ(t). The instantaneous frequency is obtained based on the instantaneous amplitude and the instantaneous phase. The instantaneous frequency and amplitude data of the main beam, bridge piers and bridge deck at different time points are obtained through the above calculations. Then, the instantaneous energy is calculated using the instantaneous frequency and amplitude data, and the frequency range is divided into several frequency bands based on the instantaneous frequency data of the main beam, bridge piers and bridge deck. The instantaneous energy in each frequency band is integrated to obtain the total energy of the frequency band:
[0091]
[0092] In the formula, is the total energy, f1 and f2 are the lower and upper limits of the frequency band respectively, E ( t ) is about time t The energy density function is expressed as t energy distribution.
[0093] By analyzing the total energy of each frequency band, determining which frequency bands the energy is mainly concentrated in, and identifying the frequency bands with high energy, we can understand which frequency ranges of vibration have the greatest impact on the bridge structure. According to the energy concentration, we can determine the vibration characteristics of the frequency range, understand the vibration behavior of different parts (main beams, piers, bridge decks) at different frequencies, and combine the analysis results of each part to obtain the vibration energy distribution of the entire bridge, providing a basis for bridge health monitoring and maintenance decisions.
[0094] Therefore, through the Hilbert-Huang transform, the instantaneous frequency and amplitude of the signal can be extracted in detail, the subtle changes in the signal can be captured, the instantaneous energy can be divided into frequency bands and integrated, and the frequency bands in which the vibration energy is concentrated can be accurately identified, which helps to understand the impact of vibration on the structure. It not only improves the accuracy and efficiency of vibration signal processing, but also provides an important technical means for the health monitoring of bridge structures.
[0095] S400, a density-based spatial clustering algorithm is used to divide the frequency range of the vibration energy distribution into several clusters, each cluster corresponds to a main vibration frequency range, and the distribution of the main vibration energy in different frequency ranges is obtained, thereby determining the damping requirements of each frequency range.
[0096] S500, according to each damping requirement, after processing by simulated annealing algorithm, searches for the best damper configuration scheme in parameter space, with the goal of minimizing the vibration energy in each frequency range, and applies genetic algorithm for refined search, to achieve the optimal solution of coordinated control of multi-scale hybrid strategy for vibrations in different frequency ranges.
[0097] It can be understood that step S500 includes S501, S502 and S503, wherein:
[0098] S501, calculating the damping requirements of the main beam, the bridge piers and the bridge deck in different frequency ranges according to the vibration energy distribution of the main beam, the bridge piers and the bridge deck, wherein the damping requirement is proportional to the energy;
[0099] S502, setting the value range of the damper parameters according to the damping requirements and actual requirements, randomly generating the damper parameters in each frequency range, and randomly generating an initial damper configuration scheme in the parameter space;
[0100] S503. Based on the initial damper configuration scheme and the vibration energy distribution of each bridge part, a search is performed within the initial damper configuration scheme through a simulated annealing algorithm. In each iteration, a new configuration scheme is generated according to the current configuration scheme, and its energy objective function value is calculated. After multiple iterations, until the damper configuration scheme with the minimum vibration energy is found, it is recorded as the optimal damper configuration scheme searched.
[0101] It should be noted that according to the vibration energy distribution of the main beam, bridge piers and bridge deck obtained in the previous steps, the energy values of each bridge part in different frequency ranges are calculated. For example, the vibration energy values of the main beam in the low frequency range (such as 0-10 Hz) and the high frequency range (such as 10-20 Hz) are Elow and Ehigh respectively. According to the vibration energy distribution, the damping requirements in different frequency ranges are calculated. Assuming that the damping demand D is proportional to the energy E, the damping requirements of the main beam, bridge piers and bridge deck in each frequency range can be determined. And according to the actual engineering requirements and damping requirements, the value range of each damper parameter is set. The initial damper configuration scheme is randomly generated in the parameter space. It is assumed that the configuration scheme consists of a set of parameters {Ki, Ci}, where i represents different bridge parts and frequency ranges. Next, the total vibration energy under the initial damper configuration scheme is calculated and recorded as the energy objective function value:
[0102]
[0103] In the formula, is the energy objective function value, E i is the vibration energy under the i-th configuration parameters, and (Ki, Ci) is a set of parameters of the configuration scheme.
[0104] Through multiple iterations of the simulated annealing algorithm, a new solution will be generated in each iteration, that is, a new configuration solution is generated based on the current configuration solution, which may be adjusted through random perturbations or other heuristic methods, and the energy objective function value Enew of the new solution is calculated, and whether to accept the new solution is determined based on the acceptance criteria, where the acceptance criteria usually include the probability acceptance criteria. After multiple iterations, the energy objective function value tends to be minimized, and the optimal damper configuration solution is finally determined.
[0105] Therefore, in the steps of this embodiment, by analyzing the vibration energy distribution, the damping requirements of each bridge part in different frequency ranges can be accurately calculated, providing a scientific basis for the damper configuration; by using the simulated annealing algorithm for global optimization search, the optimal damper configuration scheme can be efficiently found in a complex parameter space; by setting the value range of the damper parameters, it can be flexibly adjusted according to actual engineering needs to ensure the actual operability and effect of the scheme; this method not only provides a scientific bridge vibration control strategy, but also realizes the intelligent maintenance and management of the bridge structure by combining actual data and optimization algorithms.
[0106] Embodiment 2:
[0107] like Figure 2 As shown, this embodiment provides a coordinated control system of a multi-scale hybrid strategy for bridge damping, see Figure 2 The system comprises:
[0108] Acquisition module 701: used to acquire vibration signals of different parts of the bridge. According to the vibration distribution principle of the bridge, the vibration signals are collected by sensors, and fused through the graph neural network algorithm to obtain a multi-source vibration data matrix, where different parts include the main beam, piers and bridge deck;
[0109] Decomposition module 702: used for obtaining vibration signal components of various scales through a first decomposition process based on a multi-source vibration data matrix, and decomposing the signal into multiple intrinsic mode functions through a second decomposition process to obtain refined frequency components;
[0110] Analysis module 703: used to perform frequency domain analysis using Wigner quasi-probability distribution method according to the frequency components obtained by multi-scale decomposition, apply kernel principal component analysis to reduce the dimension of the frequency domain analysis results, analyze the instantaneous frequency and amplitude of the intrinsic mode function, obtain the frequency domain distribution of vibration energy at each scale, and apply Hilbert-Huang transform method to analyze the instantaneous frequency and amplitude of the intrinsic mode function to obtain the vibration energy distribution;
[0111] Determination module 704: used to divide the frequency range of vibration energy distribution into several clusters based on density-based spatial clustering algorithm, each cluster corresponds to a main vibration frequency range, obtain the distribution of main vibration energy in different frequency ranges, and then determine the damping requirements of each frequency range;
[0112] Search module 705: used to search for the best damper configuration scheme in the parameter space according to various damping requirements after processing by simulated annealing algorithm, with the goal of minimizing the vibration energy in each frequency range, and apply genetic algorithm to perform refined search to achieve the optimal solution for coordinated control of multi-scale hybrid strategies for vibrations in different frequency ranges.
[0113] Specifically, the acquisition module 701 includes:
[0114] Monitoring unit 7011: used to capture the deformation of the main beam structure and the first original vibration signal data of the dynamic response due to vibration using an accelerometer and a strain gauge, monitor the displacement of the bridge pier and the second original vibration signal data with vibration characteristics using a displacement sensor and an accelerometer, and monitor the third original vibration signal data of the dynamic response caused by the vehicle load on the bridge deck and the environmental vibration using an accelerometer and a strain gauge, wherein each data point in the first original vibration signal data, the second original vibration signal data and the third original vibration signal data corresponds to time and position;
[0115] The first processing unit 7012 is used to construct sensor data nodes according to the first original vibration signal data, the second original vibration signal data and the third original vibration signal data through topological modeling processing, regard each sensor as a node, and form a graph structure of the bridge to obtain a graph structure including the position, type and vibration signal data of each node;
[0116] The second processing unit 7013 is used to perform node embedding processing based on the constructed graph structure and graph neural network algorithm, and after node feature extraction, convert the vibration signal data of each node into a feature vector to obtain a node embedding vector containing spatiotemporal features;
[0117] Computational fusion unit 7014: used to perform multiple iterative calculations based on the node embedding vectors using the message passing mechanism in the graph neural network. After hierarchical transmission processing in the graph network, the embedding vector of each node is fused with the embedding vectors of adjacent nodes, and processed through the neighborhood aggregation function to obtain a multi-source vibration data matrix.
[0118] Specifically, the decomposition module 702 includes:
[0119] The obtaining unit 7021 is used to process the multi-source vibration data matrix through discrete wavelet transform, select a mother wavelet suitable for the characteristics of the bridge vibration signal, decompose each column vector in the multi-source vibration data matrix, decompose the signal into detail coefficients and approximation coefficients at different scales, and obtain vibration signal components at each scale, wherein the vibration signal components include different vibration characteristic data of the main beam, the pier and the bridge deck;
[0120] The third processing unit 7022 is used to reorganize the vibration signal components of each scale into a multi-scale signal component matrix through matrix construction processing;
[0121] Circulation unit 7023: used to traverse each signal component in the multi-scale signal component matrix, and select all the maximum and minimum values in the vibration signal components of each scale, respectively construct the upper envelope and the lower envelope through the maximum and minimum points, and calculate the mean of the upper envelope and the lower envelope; subtract the mean value from the signal to obtain a preliminary intrinsic mode function, iterate repeatedly until the preset conditions of the intrinsic mode function are met, separate the intrinsic mode function from the signal, and use the remaining part as a new signal, and calculate cyclically until there is no intrinsic mode function in the remaining signal, and obtain the refined frequency components, wherein the refined frequency components include the vibration frequency components of the main beam, the pier and the bridge deck, and the calculation formulas of the upper envelope and the lower envelope are as follows:
[0122]
[0123] Where u(t) is the upper envelope, l(t) is the lower envelope, and m(t) is the mean.
[0124] Specifically, the analysis module 703 includes:
[0125] The fourth processing unit 7031 is used to obtain the time-frequency energy distribution of the main beam vibration signal, the time-frequency energy distribution of the pier vibration signal and the time-frequency energy distribution of the bridge deck vibration signal based on the frequency components obtained by the multi-source vibration data matrix decomposition, and obtain the time-frequency energy distribution of the main beam vibration signal, the time-frequency energy distribution of the pier vibration signal and the time-frequency energy distribution of the bridge deck vibration signal through the Wigner quasi-probability distribution method, and record them as the main beam time-frequency distribution, the pier time-frequency distribution and the bridge deck time-frequency distribution respectively. The calculation formula of the Wigner quasi-probability distribution method is as follows;
[0126]
[0127] In the formula, W in it is the Wigner-Ville distribution, x is the time signal, t is the time variable, f is a frequency variable, x(t) represents the signal value at time t, Expressed as the complex conjugate of the signal at time t, is expressed as an integral variable, representing the time delay, represents the exponential term, indicating the frequency f and time delay The complex exponential function of represents the integration of time delay;
[0128] Dimensionality reduction unit 7032: used to perform dimensionality reduction based on the time-frequency distribution of the main beam, the time-frequency distribution of the bridge pier and the time-frequency distribution of the bridge deck through kernel principal component analysis to obtain frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the frequency domain distribution data of the bridge deck after dimensionality reduction;
[0129] Analysis unit 7033: used to perform instantaneous frequency and amplitude analysis on the intrinsic mode function based on the frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the frequency domain distribution data of the bridge deck by using the Hilbert-Huang transform method, so as to obtain the instantaneous frequency and amplitude data of the main beam, the instantaneous frequency and amplitude data of the bridge pier and the instantaneous frequency and amplitude data of the bridge deck;
[0130] Energy calculation unit 7034: used to calculate the instantaneous energy of the main beam, bridge piers and bridge deck according to the instantaneous frequency and amplitude data of the main beam, the instantaneous frequency and amplitude data of the bridge piers and the instantaneous frequency and amplitude data of the bridge deck, and divide the frequency range of the instantaneous frequency corresponding to the main beam, bridge piers and bridge deck into several frequency bands, integrate the instantaneous energy in each frequency band, and obtain the total energy in the frequency band, that is, the total energy of the main beam, the total energy of the bridge piers and the total energy of the bridge deck, and analyze the concentration of each total energy, determine the main concentrated frequency range, identify the frequency band with high energy, determine the vibration characteristics of the frequency range, and then obtain the vibration energy distribution.
[0131] Specifically, the search module 705 includes:
[0132] The calculation demand unit 7051 is used to calculate the damping demand of the main beam, the bridge pier and the bridge deck in different frequency ranges according to the vibration energy distribution of the main beam, the bridge pier and the bridge deck, wherein the damping demand is proportional to the energy;
[0133] Generating unit 7052: used to set the value range of damper parameters according to damping requirements and actual requirements, randomly generate damper parameters in each frequency range, and randomly generate an initial damper configuration scheme in the parameter space;
[0134] Iteration unit 7053: It is used to search in the initial damper configuration scheme based on the initial damper configuration scheme and the vibration energy distribution of each bridge part through the simulated annealing algorithm. In each iteration, a new configuration scheme is generated according to the current configuration scheme, and its energy objective function value is calculated. After multiple iterations, until the damper configuration scheme with the minimum vibration energy is found, it is recorded as the optimal damper configuration scheme is searched.
[0135] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.
[0136] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A coordinated control method of a multi-scale hybrid strategy for bridge damping, characterized in that: include: Obtain vibration signals from different parts of the bridge. According to the principle of bridge vibration distribution, use sensors to collect vibration signals, and perform fusion processing through the graph neural network algorithm to obtain a multi-source vibration data matrix. Different parts include the main beam, piers and bridge deck; Based on the multi-source vibration data matrix, after the first decomposition process, the vibration signal components of each scale are obtained. After the second decomposition process, the signal is decomposed into multiple intrinsic mode functions to obtain the refined frequency components. According to the frequency components obtained by multi-scale decomposition, the Wigner quasi-probability distribution method is used for frequency domain analysis. The kernel principal component analysis is applied to reduce the dimension of the frequency domain analysis results, and the instantaneous frequency and amplitude of the intrinsic mode function are analyzed to obtain the frequency domain distribution of vibration energy at each scale. The Hilbert-Huang transform method is used to analyze the instantaneous frequency and amplitude of the intrinsic mode function to obtain the vibration energy distribution. The density-based spatial clustering algorithm divides the frequency range of vibration energy distribution into several clusters. Each cluster corresponds to a main vibration frequency range. The distribution of the main vibration energy in different frequency ranges is obtained, and then the damping requirements of each frequency range are determined. According to each damping requirement, after processing by simulated annealing algorithm, the optimal damper configuration scheme is searched in the parameter space, with the goal of minimizing the vibration energy in each frequency range, and the genetic algorithm is applied for refinement search to achieve the optimal solution of coordinated control of multi-scale hybrid strategies for vibrations in different frequency ranges; Among them, the multi-source vibration data matrix is subjected to the first decomposition process to obtain vibration signal components of each scale, and the signal is decomposed into multiple intrinsic mode functions through the second decomposition process to obtain refined frequency components, including: The multi-source vibration data matrix is processed by discrete wavelet transform, and the mother wavelet suitable for the characteristics of bridge vibration signals is selected. Each column vector in the multi-source vibration data matrix is decomposed, and the signal is decomposed into detail coefficients and approximation coefficients at different scales to obtain vibration signal components at each scale, where the vibration signal components include different vibration characteristic data of the main beam, pier and bridge deck; Based on the vibration signal components of each scale, these components are reorganized into a multi-scale signal component matrix through matrix construction processing; Traverse each signal component in the multi-scale signal component matrix, and select all the maximum and minimum values in the vibration signal components of each scale, construct the upper envelope and the lower envelope through the maximum and minimum points respectively, and calculate the mean of the upper envelope and the lower envelope; subtract the mean value from the signal to obtain the preliminary intrinsic mode function, iterate repeatedly until the preset conditions of the intrinsic mode function are met, separate the intrinsic mode function from the signal, and use the remaining part as a new signal. Circular calculation until there is no intrinsic mode function in the remaining signal, and obtain the refined frequency components, where the refined frequency components include the vibration frequency components of the main beam, piers and bridge deck, and the calculation formulas of the upper envelope and the lower envelope are as follows: Where u(t) is the upper envelope, l(t) is the lower envelope, and m(t) is the mean.
2. The coordinated control method of the multi-scale hybrid strategy of bridge damping according to claim 1 is characterized in that: The vibration signals of different parts of the bridge are obtained. According to the vibration distribution principle of the bridge, the vibration signals are collected by sensors, and the multi-source vibration data matrix is obtained by fusion processing through the graph neural network algorithm, which includes: Using accelerometers and strain gauges, capture the first original vibration signal data of the deformation of the main beam structure and the dynamic response due to vibration, using displacement sensors and accelerometers, monitor the displacement of the pier and the second original vibration signal data with vibration characteristics, using accelerometers and strain gauges, monitor the third original vibration signal data of the dynamic response caused by the vehicle load and environmental vibration on the bridge deck, wherein each data point in the first original vibration signal data, the second original vibration signal data and the third original vibration signal data corresponds to time and position; According to the first original vibration signal data, the second original vibration signal data and the third original vibration signal data, after topological modeling processing, sensor data nodes are constructed, each sensor is regarded as a node, and a graph structure of a bridge is formed to obtain a graph structure including the position, type and vibration signal data of each node; Based on the constructed graph structure and graph neural network algorithm, the nodes are embedded. After node feature extraction, the vibration signal data of each node is converted into a feature vector to obtain a node embedding vector containing spatiotemporal features. According to the node embedding vector, the message passing mechanism in the graph neural network is used to perform multiple iterative calculations. After the hierarchical transmission processing of the graph network, the embedding vector of each node is fused with the embedding vectors of the adjacent nodes. Through the neighborhood aggregation function processing, the multi-source vibration data matrix is obtained.
3. The coordinated control method of the multi-scale hybrid strategy of bridge damping according to claim 1 is characterized in that: The frequency components obtained by multi-scale decomposition are subjected to frequency domain analysis using the Wigner quasi-probability distribution method, and the kernel principal component analysis is applied to reduce the dimension of the frequency domain analysis results, and the instantaneous frequency and amplitude of the intrinsic mode function are analyzed to obtain the frequency domain distribution of vibration energy at each scale, and the Hilbert-Huang transform method is applied to analyze the instantaneous frequency and amplitude of the intrinsic mode function to obtain the vibration energy distribution, which includes: Based on the frequency components obtained by matrix decomposition of multi-source vibration data, the time-frequency energy distribution of the main beam vibration signal, the time-frequency energy distribution of the pier vibration signal and the time-frequency energy distribution of the bridge deck vibration signal are obtained through the Wigner quasi-probability distribution method, and are recorded as the main beam time-frequency distribution, the pier time-frequency distribution and the bridge deck time-frequency distribution respectively. The calculation formula of the Wigner quasi-probability distribution method is as follows; In the formula, W in it is the Wigner-Ville distribution, x is the time signal, t is the time variable, f is a frequency variable, x(t) represents the signal value at time t, Expressed as the complex conjugate of the signal at time t, is expressed as an integral variable, representing the time delay, represents the exponential term, indicating the frequency f and time delay The complex exponential function of represents the integration of time delay; According to the time-frequency distribution of the main beam, the time-frequency distribution of the bridge pier and the bridge deck, the dimension is reduced by kernel principal component analysis to obtain the frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the bridge deck after dimension reduction; Based on the frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the frequency domain distribution data of the bridge deck, the Hilbert-Huang transform method is used to analyze the instantaneous frequency and amplitude of the intrinsic mode function, and the instantaneous frequency and amplitude data of the main beam, the instantaneous frequency and amplitude data of the bridge pier and the instantaneous frequency and amplitude data of the bridge deck are obtained; According to the instantaneous frequency and amplitude data of the main beam, the instantaneous frequency and amplitude data of the bridge piers and the bridge deck, the instantaneous energy of the main beam, the bridge piers and the bridge deck is calculated, and the frequency range of the instantaneous frequency corresponding to the main beam, the bridge piers and the bridge deck is divided into several frequency bands. The instantaneous energy in each frequency band is integrated to obtain the total energy in the frequency band, that is, the total energy of the main beam, the total energy of the bridge piers and the total energy of the bridge deck. The concentration of each total energy is analyzed to determine the main concentrated frequency range, identify the frequency band with high energy, determine the vibration characteristics of the frequency range, and then obtain the vibration energy distribution.
4. The coordinated control method of multi-scale hybrid strategy for bridge damping according to claim 1 is characterized in that: According to various damping requirements, the simulated annealing algorithm is used to search for the best damper configuration in the parameter space, with the goal of minimizing the vibration energy in each frequency range, including: According to the vibration energy distribution of the main beam, piers and bridge deck, the damping requirements of the main beam, piers and bridge deck in different frequency ranges are calculated, where the damping requirements are proportional to the energy; According to the damping requirements and actual needs, the value range of the damper parameters is set, the damper parameters are randomly generated in each frequency range, and the initial damper configuration scheme is randomly generated in the parameter space; Based on the initial damper configuration scheme and the vibration energy distribution of each bridge part, the simulated annealing algorithm is used to search within the initial damper configuration scheme. In each iteration, a new configuration scheme is generated according to the current configuration scheme, and its energy objective function value is calculated. After multiple iterations, until the damper configuration scheme with the minimum vibration energy is found, it is recorded as the optimal damper configuration scheme searched.
5. A coordinated control system of a multi-scale hybrid strategy for bridge damping, based on the coordinated control method of a multi-scale hybrid strategy for bridge damping according to claim 1, characterized in that: include: Acquisition module: used to obtain vibration signals from different parts of the bridge. According to the vibration distribution principle of the bridge, the vibration signals are collected by sensors and fused through the graph neural network algorithm to obtain a multi-source vibration data matrix. The different parts include the main beam, piers and bridge deck; Decomposition module: Based on the multi-source vibration data matrix, after the first decomposition process, the vibration signal components of each scale are obtained, and after the second decomposition process, the signal is decomposed into multiple intrinsic mode functions to obtain the refined frequency components; Analysis module: It is used to perform frequency domain analysis using Wigner quasi-probability distribution method based on the frequency components obtained by multi-scale decomposition, apply kernel principal component analysis to reduce the dimension of the frequency domain analysis results, analyze the instantaneous frequency and amplitude of the intrinsic mode function, obtain the frequency domain distribution of vibration energy at each scale, and apply Hilbert-Huang transform method to analyze the instantaneous frequency and amplitude of the intrinsic mode function to obtain the vibration energy distribution; Determination module: It is used to divide the frequency range of vibration energy distribution into several clusters based on density-based spatial clustering algorithm. Each cluster corresponds to a main vibration frequency range, and the distribution of main vibration energy in different frequency ranges is obtained, so as to determine the damping requirements of each frequency range. Search module: It is used to search for the best damper configuration scheme in the parameter space according to various damping requirements through simulated annealing algorithm processing, with the goal of minimizing the vibration energy in each frequency range, and to apply genetic algorithm for refined search to achieve the optimal solution for coordinated control of multi-scale hybrid strategies for vibrations in different frequency ranges; Wherein, the decomposition module includes: Acquisition unit: used for processing the multi-source vibration data matrix through discrete wavelet transform, selecting a mother wavelet suitable for the characteristics of bridge vibration signals, decomposing each column vector in the multi-source vibration data matrix, decomposing the signal into detail coefficients and approximation coefficients at different scales, and obtaining vibration signal components at each scale, wherein the vibration signal components include different vibration characteristic data of the main beam, piers and bridge deck; The third processing unit is used for reorganizing the vibration signal components of each scale into a multi-scale signal component matrix through matrix construction processing; Circular unit: used to traverse each signal component in the multi-scale signal component matrix, and select all the maximum and minimum values in the vibration signal components of each scale, respectively construct the upper envelope and the lower envelope through the maximum and minimum points, and calculate the mean of the upper envelope and the lower envelope; subtract the mean value from the signal to obtain the preliminary intrinsic mode function, iterate repeatedly until the preset conditions of the intrinsic mode function are met, separate the intrinsic mode function from the signal, and use the remaining part as a new signal. Circular calculation is performed until there is no intrinsic mode function in the remaining signal, and the refined frequency components are obtained, wherein the refined frequency components include the vibration frequency components of the main beam, piers and bridge deck, and the calculation formulas of the upper envelope and the lower envelope are as follows: Where u(t) is the upper envelope, l(t) is the lower envelope, and m(t) is the mean.
6. The coordinated control system of the multi-scale hybrid strategy for bridge damping according to claim 5 is characterized in that: The acquisition module includes: Monitoring unit: used to capture the deformation of the main beam structure and the first original vibration signal data of the dynamic response due to vibration using an accelerometer and a strain gauge, monitor the displacement of the bridge pier and the second original vibration signal data with vibration characteristics using a displacement sensor and an accelerometer, and monitor the third original vibration signal data of the dynamic response caused by the vehicle load on the bridge deck and the environmental vibration using an accelerometer and a strain gauge, wherein each data point in the first original vibration signal data, the second original vibration signal data and the third original vibration signal data corresponds to time and position; The first processing unit is used to construct sensor data nodes according to the first original vibration signal data, the second original vibration signal data and the third original vibration signal data through topological modeling processing, regard each sensor as a node, and form a graph structure of the bridge to obtain a graph structure including the position, type and vibration signal data of each node; The second processing unit is used to embed nodes based on the constructed graph structure and graph neural network algorithm, and convert the vibration signal data of each node into a feature vector after node feature extraction to obtain a node embedding vector containing spatiotemporal features; Computational fusion unit: It is used to perform multiple iterative calculations based on the node embedding vectors using the message passing mechanism in the graph neural network. After the hierarchical transmission processing of the graph network, the embedding vector of each node is fused with the embedding vectors of the adjacent nodes, and processed through the neighborhood aggregation function to obtain the multi-source vibration data matrix.
7. The coordinated control system of the multi-scale hybrid strategy for bridge damping according to claim 5 is characterized in that: The analysis module includes: The fourth processing unit is used to obtain the time-frequency energy distribution of the main beam vibration signal, the time-frequency energy distribution of the pier vibration signal and the time-frequency energy distribution of the bridge deck vibration signal based on the frequency components obtained by the matrix decomposition of the multi-source vibration data, and to obtain the time-frequency energy distribution of the main beam vibration signal, the time-frequency energy distribution of the pier vibration signal and the time-frequency energy distribution of the bridge deck vibration signal through the Wigner quasi-probability distribution method, and record them as the main beam time-frequency distribution, the pier time-frequency distribution and the bridge deck time-frequency distribution respectively. The calculation formula of the Wigner quasi-probability distribution method is as follows; In the formula, W in it is the Wigner-Ville distribution, x is the time signal, t is the time variable, f is a frequency variable, x(t) represents the signal value at time t, Expressed as the complex conjugate of the signal at time t, is expressed as an integral variable, representing the time delay, represents the exponential term, indicating the frequency f and time delay The complex exponential function of represents the integration of time delay; Dimensionality reduction unit: used to reduce the dimension according to the time-frequency distribution of the main beam, the time-frequency distribution of the bridge pier and the time-frequency distribution of the bridge deck through kernel principal component analysis, and obtain the frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the bridge deck after dimension reduction; Analysis unit: used to perform instantaneous frequency and amplitude analysis on the intrinsic mode function based on the frequency domain distribution data of the main beam, the frequency domain distribution data of the bridge pier and the frequency domain distribution data of the bridge deck by using the Hilbert-Huang transform method, so as to obtain the instantaneous frequency and amplitude data of the main beam, the instantaneous frequency and amplitude data of the bridge pier and the instantaneous frequency and amplitude data of the bridge deck; Energy calculation unit: used to calculate the instantaneous energy of the main beam, bridge piers and bridge deck according to the instantaneous frequency and amplitude data of the main beam, the instantaneous frequency and amplitude data of the bridge piers and the instantaneous frequency and amplitude data of the bridge deck, and divide the frequency range of the instantaneous frequency corresponding to the main beam, bridge piers and bridge deck into several frequency bands, integrate the instantaneous energy in each frequency band, and obtain the total energy in the frequency band, that is, the total energy of the main beam, the total energy of the bridge piers and the total energy of the bridge deck, and analyze the concentration of each total energy, determine the main concentrated frequency range, identify the frequency band with high energy, determine the vibration characteristics of the frequency range, and then obtain the vibration energy distribution.
8. The coordinated control system of the multi-scale hybrid strategy for bridge damping according to claim 5 is characterized in that: The search module includes: Calculation demand unit: used to calculate the damping demand of the main beam, bridge piers and bridge deck in different frequency ranges according to the vibration energy distribution of the main beam, bridge piers and bridge deck, where the damping demand is proportional to the energy; Generation unit: used to set the value range of damper parameters according to damping requirements and actual needs, randomly generate damper parameters in each frequency range, and randomly generate initial damper configuration schemes in the parameter space; Iteration unit: It is used to search within the initial damper configuration scheme based on the initial damper configuration scheme and the vibration energy distribution of each bridge part through the simulated annealing algorithm. In each iteration, a new configuration scheme is generated according to the current configuration scheme, and its energy objective function value is calculated. After multiple iterations, until the damper configuration scheme with the minimum vibration energy is found, it is recorded as the optimal damper configuration scheme is searched.
