Vortex shedding intelligent identification and tracking method based on bridge in-situ flow field data

Abstract

The invention discloses a vortex shedding intelligent identification and tracking method based on bridge in-situ flow field data, and the method specifically comprises the steps: collecting bridge global flow field data through cooperation of a laser radar and an anemograph, inverting flow field characteristic parameters, identifying a vortex structure, and carrying out the feature extraction, thereby obtaining vortex shedding characteristic parameters, and carrying out the tracking of the vortex shedding characteristic parameters. In combination with a deep ensemble learning algorithm, constructing a mapping relationship between the flow field characteristic parameters and the vortex shedding characteristic parameters, and establishing a bridge vortex shedding intelligent identification model; on the basis, a vortex shedding trajectory spatio-temporal evolution graph is constructed, the mapping relation of wind speed gradient-vortex shedding strength is analyzed, and a vortex shedding instability prediction model is established. According to the method, the in-situ vortex shedding phenomenon of the bridge can be effectively identified and tracked, and the wind-resistant intelligent level of the bridge is further improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge wind resistance safety monitoring, and in particular relates to an intelligent identification and tracking method for vortex shedding based on in-situ flow field data of a bridge. Background Art

[0002] With the continuous breakthroughs in the development of kilometer-long cable-supported bridges and the trend toward lightweight structural design, the sensitivity of long-span bridges to wind-induced vibrations has increased dramatically. Aeroelasticity has become a limiting factor in the wind-resistant design of the strength, stiffness, and stability of long-span and ultra-long-span bridges. Vortex-induced vibration, a core issue in bridge wind-resistant design, is caused by vortices periodically shed when fluid flows through a blunt-body section. The resulting alternating vortex excitation, if close to or identical to the bridge's natural frequency, can cause significant structural vibration. The occurrence of vortex-induced vibration is influenced not only by inherent factors such as the bridge's cross-sectional shape and structural stiffness, but also by the characteristics of the ambient wind. It is particularly prone to occur in complex environments such as wide rivers, deep-water channels, and mountainous canyons with significant wind speed variations, where long-span bridges are located. This directly leads to accelerated structural fatigue damage, a dramatic increase in traffic safety risks, and escalating operation and maintenance costs. Therefore, accurate sensing and prevention technologies for bridge vortex-induced vibration have become a critical requirement for effectively reducing the operational and maintenance risks of major infrastructure.

[0003] Traditional sensor signals are typically a superposition of multi-field coupled responses. Their frequency-domain characteristics are strongly correlated with the nonlinear dynamics of vortex shedding, making it difficult to effectively decouple vortex shedding energy through conventional frequency-domain filtering. Furthermore, key fluid dynamics parameters such as velocity gradients and pressure pulsations associated with vortex shedding cannot be obtained, leading to an unclear constitutive relationship between vortex shedding and the dynamic evolution of the flow field. Bridge vortex shedding identification and tracking technology, fundamental to vortex shedding mechanism research, constructs a high-precision prediction model of fluid-structure interaction and dynamically analyzes the vortex shedding energy transfer path. This technology can accurately capture the generation, evolution, and shedding of vortex cores across the entire bridge flow field. Furthermore, inverse inference techniques based on vortex shedding trajectories can guide the optimization of bridge aerodynamic shapes, reduce wind tunnel testing costs during the design phase, and significantly improve the safety and economic efficiency of bridges throughout their lifecycle. However, research on intelligent vortex shedding identification and tracking using in-situ bridge flow field data remains significant. In view of this, it is urgent to use the in-situ flow field data of bridges to establish an intelligent identification and tracking method for vortex shedding, fill the research deficiencies in the field of bridge wind engineering, further improve the intelligent level of bridge safety monitoring, and provide solid protection for the wind resistance safety of my country's large-span bridges. Summary of the Invention

[0004] The purpose of the present invention is to solve the problems of low accuracy and low intelligence level of existing bridge vortex shedding identification, and to provide a method for intelligent identification and tracking of vortex shedding based on in-situ flow field data of bridges;

[0005] To achieve the above objectives, the present invention adopts the following technical solutions: a method for intelligently identifying and tracking vortex shedding based on in-situ flow field data of bridges, comprising:

[0006] Step 1: Collect the flow field data of the entire bridge area, invert the flow field characteristic parameters and analyze its temporal and spatial evolution process;

[0007] Step 2: Identify the vortex structure and perform feature extraction to obtain vortex shedding characteristic parameters;

[0008] Step 3: Combined with the deep ensemble learning algorithm, the mapping relationship between flow field characteristic parameters and vortex shedding characteristic parameters is constructed to establish an intelligent identification model for bridge vortex shedding;

[0009] Step 4: Extract the vortex shedding characteristic parameters output by the bridge vortex shedding intelligent identification model and dynamically adjust the parameters and algorithms of the lidar;

[0010] Step 5: Based on the vortex shedding characteristic parameters output by the bridge vortex shedding intelligent identification model, track the vortex shedding evolution process, construct a spatiotemporal evolution map of the vortex shedding trajectory, and then obtain the dominant spatiotemporal modes of the vortex shedding evolution process and their energy contribution ratio;

[0011] Step 6: Based on the dominant spatiotemporal modes of the vortex shedding evolution process and their energy contribution ratio, a wind speed gradient-vortex shedding intensity mapping relationship is constructed, and a vortex shedding instability prediction model is established.

[0012] Furthermore, the specific steps of step 1 include:

[0013] Step 11: Collect global flow field data of the bridge by coordinating the lidar and anemometer, and then calculate the flow field characteristic parameters based on the collected global flow field data of the bridge, where the flow field characteristic parameters include extinction coefficient, spectral width, skewness, wind vector, rotation, and shear intensity;

[0014] Step 12: Extract multi-scale turbulence fluctuation information based on the wavelet transform method, and combine the wind vector and flow field characteristic parameters to finely analyze the flow field structure;

[0015] Step 13: Analyze the evolution of flow field characteristic parameters in time and space, and extract the spatiotemporal evolution characteristics of the flow field, wherein the spatiotemporal evolution characteristics of the flow field include the generation, evolution process and change cycle of the flow field structure.

[0016] Furthermore, in step 11:

[0017] The extinction coefficient is directly measured by the lidar and is part of the bridge's full-area flow field data;

[0018] The expression of spectral width is:

[0019] Where f(v) is the spectrum of the reflected signal in the global flow field data of the bridge, v represents the Doppler shift frequency of the laser signal, is the average frequency of the spectrum;

[0020] The expression for skewness is:

[0021] Where σ is the spectrum width;

[0022] The wind vector expression in the bridge global flow field data is:

[0023]

[0024] Where x is the position coordinate along the downwind direction, y is the position coordinate along the crosswind direction, z is the position coordinate along the vertical direction, t is the time, u(x,y,z,t) is the downwind wind speed, v(x,y,z,t) is the crosswind wind speed, and w(x,y,z,t) is the vertical wind speed.

[0025] The curl expression is:

[0026] Among them, the specific components of the curl along the x, y, and z axes are: In the formula is the symbol of partial derivative;

[0027] Shear strength:

[0028] In step 12:

[0029] The wavelet transform form is:

[0030] Where u(t) is the time series, a is the scale parameter, τ is the time shift, is the mother wavelet function;

[0031] The turbulent energy distribution at each scale is: E u (a,τ)=|W u (a,τ)| 2 ; After combining the wind vector and flow field characteristic parameters, the multi-scale flow intensity index expression is:

[0032]

[0033] Where, α i is the weight coefficient, i takes 1, 2, 3, 4, E u (a,τ)=|W u (a,τ)| 2is the turbulence energy distribution at each scale, ω(x,y,z,t), S(x,y,z,t), and σ(x,y,z,t) are the curl, shear intensity, and spectral width at a certain position (x,y,z) in the flow field at time t.

[0034] Furthermore, the specific steps of step 2 include:

[0035] Step 21: Determine the vortex shedding dominant frequency based on the spectrum analysis method;

[0036] Step 22: Analyze the spatial distribution of velocity field and scattering intensity in the bridge's full-area flow field data, and estimate the spatial scale parameters of vortex shedding based on the LiDAR measurement geometry.

[0037] Step 23: Calculate the vorticity distribution based on the velocity field data, and analyze the rotation intensity and structure inside the vortex.

[0038] Furthermore, the step 3 specifically includes:

[0039] Step 31: Convert the flow field characteristic parameter and vortex shedding characteristic parameter data into a unified format, delete duplicate data entries, and encode the categorical variables; after the data is standardized, construct a matrix containing time series and spatial distribution characteristics to obtain the flow field characteristic parameter data set and the vortex shedding characteristic parameter data set;

[0040] Step 32: Input the flow field characteristic parameter dataset and the vortex shedding characteristic parameter dataset into the deep learning model to train the deep learning model and obtain a bridge vortex shedding intelligent recognition model.

[0041] Furthermore, the specific steps of step 4 include:

[0042] Step 41: extracting vortex shedding characteristic parameters output by the bridge vortex shedding intelligent identification model, wherein the vortex shedding characteristic parameters include vortex shedding intensity, vortex shedding frequency, vortex shedding size, and vortex shedding position;

[0043] Step 42: Dynamically adjust the pulse repetition frequency and integration time of the lidar based on the accuracy and effectiveness of the vortex shedding characteristic parameters, and increase the scanning sampling in the high vortex shedding intensity area;

[0044] Step 43: Using Kalman filtering to perform real-time denoising on the raw data, and using the vortex shedding characteristic parameter feedback optimization algorithm based on the output of the bridge vortex shedding intelligent identification model;

[0045] Step 44: For asymmetric vortex shedding events, adaptively switch the scanning mode to ensure complete capture of the vortex shedding three-dimensional trajectory;

[0046] Step 45: Establish a parameter adjustment log library to record the influence of lidar configuration and algorithm version on the accuracy of vortex shedding identification.

[0047] Furthermore, the specific steps of step 5 include:

[0048] Step 51: Based on the vortex shedding intensity, frequency, size, and location output by the bridge vortex shedding intelligent identification model, the vortex cores at consecutive moments are matched for spatial proximity and intensity similarity to track the complete motion trajectory of vortex generation, shedding, merging, and dissipation. The spatial path, time history, and intensity evolution information of each vortex are integrated to construct a visual spatiotemporal evolution map of the vortex shedding trajectory.

[0049] Step 52: Different evolution modes are distinguished based on the cluster analysis method. The dominant space-time modes of the vortex shedding evolution process are obtained by using the intrinsic orthogonal decomposition, and the energy contribution ratio of the dominant space-time modes is analyzed. The different evolution modes include symmetric shedding, asymmetric merging, and secondary vortex excitation.

[0050] Furthermore, the specific steps of step 6 include:

[0051] Step 61: Based on the dominant spatiotemporal modes of the vortex shedding evolution process and their energy contribution ratios, analyze the interaction between the primary vortex and the secondary vortex, and quantify its impact on the vortex shedding frequency and energy dissipation. The interaction between the primary vortex and the secondary vortex includes vortex pair merging, vortex filament breakage, and vortex shedding locking.

[0052] Step 62: Calculate the modulation effect of non-uniform incoming flow on the spatial distribution and frequency of vortex shedding, construct a mapping relationship between wind speed gradient and vortex shedding intensity, and predict the critical threshold for the transition from periodic shedding to turbulence. Combined with the interaction between the primary vortex and the secondary vortex in step 61, a vortex shedding instability prediction model is obtained.

[0053] Beneficial effects:

[0054] 1. By establishing an intelligent identification model for bridge vortex shedding and a prediction model for vortex shedding instability, in-situ vortex shedding identification and tracking can be achieved on bridges, solving the bottleneck problem of the lack of in-situ vortex shedding tracking capabilities.

[0055] 2. Through the bridge vortex shedding intelligent identification model, dynamic adjustment of lidar measurement parameters and data processing algorithms can be achieved, effectively reducing measurement errors and improving the accuracy and reliability of vortex shedding identification and tracking.

[0056] 3. Based on the analysis of the influence of the interaction between the primary vortex and the secondary vortex on the vortex shedding frequency and energy dissipation, the action mechanism and influence mechanism of the dynamic evolution of vortex shedding on the aerodynamic stability of the bridge can be revealed. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a technical flow chart of the intelligent identification and tracking method of vortex shedding based on the in-situ flow field data of the bridge.

[0058] Figure 2 Schematic diagram of the lidar scanning mode.

[0059] Figure 3 Schematic diagram of the intelligent identification model framework for bridge vortex shedding.

[0060] In the figure: 1. Wind lidar, 2. Elevation angle, 3. Measurement area. DETAILED DESCRIPTION

[0061] The present invention will be further explained below with reference to the accompanying drawings.

[0062] like Figure 1 As shown, the present invention provides an intelligent identification and tracking method for vortex shedding based on in-situ flow field data of a bridge, comprising:

[0063] Step 1: Collect the flow field data of the entire bridge area, invert the flow field characteristic parameters and analyze its temporal and spatial evolution process.

[0064] Step 2: Identify the vortex structure and perform feature extraction to obtain vortex shedding characteristic parameters.

[0065] Step 3: Combined with the deep ensemble learning algorithm, the mapping relationship between flow field characteristic parameters and vortex shedding characteristic parameters is constructed to establish an intelligent identification model for bridge vortex shedding.

[0066] Step 4: Extract the vortex shedding characteristic parameters output by the bridge vortex shedding intelligent identification model and dynamically adjust the parameters and algorithms of the lidar.

[0067] Step 5: Based on the vortex shedding characteristic parameters output by the bridge vortex shedding intelligent identification model, track the vortex shedding evolution process, construct a spatiotemporal evolution map of the vortex shedding trajectory, and then obtain the dominant spatiotemporal modes of the vortex shedding evolution process and their energy contribution ratio;

[0068] Step 6: Based on the dominant spatiotemporal modes of the vortex shedding evolution process and their energy contribution ratio, a wind speed gradient-vortex shedding intensity mapping relationship is constructed, and a vortex shedding instability prediction model is established.

[0069] like Figure 2 As shown in the figure, the specific steps for collecting the flow field data of the entire bridge are as follows: two lidars and multiple anemometers are deployed at specific locations such as the 1 / 4 span of the main beam of the bridge. The lidar performs conical scanning at an elevation angle of 10° and adopts a synchronous trigger mode to cover the flow field areas such as the windward and leeward sides of the bridge, and synchronously collects the three-dimensional flow field data of the entire bridge, including horizontal wind speed, radial wind speed and wind direction.

[0070] In step 1, the process of inverting the flow field characteristic parameters and analyzing their spatiotemporal evolution includes:

[0071] First, based on the collected flow field data of the entire bridge area, the flow field characteristic parameters such as extinction coefficient, spectral width, skewness, wind vector, rotation, and shear intensity are calculated, and a flow field parameter dataset with matching temporal and spatial resolution is constructed.

[0072] Continuous wavelet transform is then used to extract multi-scale turbulence pulsation information, and combined with wind vector field data and flow field data such as extinction coefficient, spectral width, and skewness, the flow field structure characteristics are finely analyzed.

[0073] Finally, the evolution laws of flow field characteristic parameters in time and space are analyzed to extract the spatiotemporal evolution characteristics of the flow field structure, such as generation, evolution process, and change cycle.

[0074] In step 11 above:

[0075] The extinction coefficient is directly measured by the lidar and is part of the bridge's full-area flow field data;

[0076] The expression of spectral width is:

[0077] Where f(v) is the spectrum of the reflected signal in the global flow field data of the bridge, v represents the Doppler shift frequency of the laser signal, is the average frequency of the spectrum;

[0078] The expression for skewness is:

[0079] Where σ is the spectrum width;

[0080] The wind vector expression in the bridge global flow field data is:

[0081]

[0082] Where x is the position coordinate along the downwind direction, y is the position coordinate along the crosswind direction, z is the position coordinate along the vertical direction, t is the time, u(x,y,z,t) is the downwind wind speed, v(x,y,z,t) is the crosswind wind speed, and w(x,y,z,t) is the vertical wind speed.

[0083] The curl expression is:

[0084] Among them, the specific components of the curl along the x, y, and z axes are: In the formula is the symbol of partial derivative;

[0085] Shear strength:

[0086] In step 12:

[0087] The wavelet transform form is:

[0088] Where u(t) is the time series, a is the scale parameter, τ is the time shift, is the mother wavelet function;

[0089] The turbulent energy distribution at each scale is: E u (a,τ)=|W u (a,τ)| 2 ; After combining the wind vector and flow field characteristic parameters, the multi-scale flow intensity index expression is:

[0090]

[0091] Where, α i is the weight coefficient, i takes 1, 2, 3, 4, E u (a,τ)=|W u (a,τ)| 2 is the turbulence energy distribution at each scale, ω(x,y,z,t), S(x,y,z,t), and σ(x,y,z,t) are the curl, shear intensity, and spectral width at a certain position (x,y,z) in the flow field at time t.

[0092] In step 2, the specific steps for identifying the vortex structure and extracting features are as follows:

[0093] First, fast Fourier transform and power spectral density analysis are performed on the flow field data to identify the dominant frequency of vortex shedding.

[0094] Next, the spatial distribution of the velocity field and scattering intensity in the full-area flow field data of the bridge was analyzed. Combined with the spatial scanning angle and beam geometry of the lidar, the lateral scale and longitudinal spacing of the vortex core were estimated through triangulation to obtain the spatial scale parameters of the vortex shedding.

[0095] Then, the spatial distribution of vorticity is calculated based on the velocity field data, the vortex core boundary is identified using the Q criterion or the λ2 criterion, the vortex shedding intensity is quantified as the integral value of the closed velocity circulation, and the structural characteristics inside the vortex are revealed through the vortex gradient and velocity circulation distribution.

[0096] like Figure 3 As shown in Figure 1, the framework of the intelligent identification model for bridge vortex shedding is implemented. The input is a flow field parameter dataset (including wind speed, curl, shear intensity, etc.), and the output is labels such as vortex shedding intensity, frequency, and size.

[0097] Step 3: The specific steps include:

[0098] First, the flow field characteristic parameters and vortex shedding characteristic parameter data are converted into a unified format, duplicate data entries are deleted, and categorical variables are encoded; after the data are standardized, a matrix form containing time series and spatial distribution characteristics is constructed to obtain the flow field characteristic parameter data set and vortex shedding characteristic parameter data set.

[0099] Next, the flow field characteristic parameter dataset and the vortex shedding characteristic parameter dataset are input into the deep learning model to train the deep learning model and obtain the bridge vortex shedding intelligent recognition model.

[0100] Step 4: The specific steps include:

[0101] Firstly, the vortex shedding characteristic parameters output by the bridge vortex shedding intelligent identification model are extracted. The vortex shedding characteristic parameters include vortex shedding intensity, vortex shedding frequency, vortex shedding size and vortex shedding position.

[0102] Then, based on the accuracy and effectiveness of the vortex shedding characteristic parameters, the pulse repetition frequency and integration time of the lidar are dynamically adjusted, and scanning sampling is intensified in areas with high vortex shedding intensity.

[0103] Then, the Kalman filter is used to perform real-time denoising on the raw data, and the vortex shedding characteristic parameters are fed back to the optimization algorithm based on the output of the bridge vortex shedding intelligent identification model.

[0104] Then, for asymmetric vortex shedding events, the scanning mode is adaptively switched to ensure the complete capture of the vortex shedding three-dimensional trajectory.

[0105] Finally, a parameter adjustment log library is established to record the influence of lidar configuration and algorithm version on the accuracy of vortex shedding identification.

[0106] Step 5: The specific steps include:

[0107] Firstly, based on the vortex shedding intensity, vortex shedding frequency, vortex shedding size and vortex shedding position output by the bridge vortex shedding intelligent identification model, the spatial proximity and intensity similarity of the vortex cores at consecutive moments are matched to track the complete motion trajectory of vortex generation, shedding, merging and dissipation; the spatial path, time history and intensity evolution information of each vortex are integrated to construct a visual spatiotemporal evolution map of the vortex shedding trajectory.

[0108] Then, based on the cluster analysis method, different evolution modes are distinguished, and the dominant space-time modes of the vortex shedding evolution process are obtained by using the intrinsic orthogonal decomposition. The energy contribution ratio of the dominant space-time modes is analyzed, among which the different evolution modes include symmetric shedding, asymmetric merging and secondary vortex excitation.

[0109] Step 6: The specific steps include:

[0110] First, based on the dominant spatiotemporal modes of the vortex shedding evolution process and their energy contribution ratio, the interaction between the primary vortex and the secondary vortex is analyzed, and its impact on the vortex shedding frequency and energy dissipation is quantified. The interaction between the primary vortex and the secondary vortex includes vortex pair merging, vortex filament breakage, and vortex shedding locking.

[0111] Then, the modulation effect of non-uniform incoming flow on the spatial distribution and frequency of vortex shedding is calculated, the mapping relationship between wind speed gradient and vortex shedding intensity is constructed, and the critical threshold for the transition from periodic shedding to turbulence is predicted. Combined with the interaction between the primary vortex and the secondary vortex, a vortex shedding instability prediction model is obtained.

[0112] The present invention collects the entire flow field data of the bridge through the coordinated use of laser radar and anemometer, inverts the flow field characteristic parameters based on this, identifies the vortex structure and performs feature extraction, and then obtains the vortex shedding characteristic parameters. Then, combined with the deep integrated learning algorithm, a mapping relationship between the flow field characteristic parameters and the vortex shedding characteristic parameters is constructed, and an intelligent bridge vortex shedding recognition model is established. On this basis, a spatiotemporal evolution map of the vortex shedding trajectory is constructed, the mapping relationship of "wind speed gradient-vortex shedding intensity" is analyzed, and a vortex shedding instability prediction model is established. By establishing the bridge vortex shedding intelligent recognition model and the vortex shedding instability prediction model, in-situ vortex shedding recognition and tracking of the bridge are realized, and the bottleneck problem of the lack of in-situ vortex shedding tracking capability is solved. Through the bridge vortex shedding intelligent recognition model, dynamic adjustment of the laser radar measurement parameters and data processing algorithms is realized, which effectively reduces measurement errors and improves the accuracy and reliability of vortex shedding recognition and tracking.

[0113] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for intelligent identification and tracking of vortex shedding based on in-situ flow field data of bridges, characterized in that: include: Step 1: Collect the flow field data of the entire bridge area, invert the flow field characteristic parameters and analyze its temporal and spatial evolution process; Step 2: Identify the vortex structure and perform feature extraction to obtain vortex shedding characteristic parameters; Step 3: Combined with the deep ensemble learning algorithm, the mapping relationship between flow field characteristic parameters and vortex shedding characteristic parameters is constructed to establish an intelligent identification model for bridge vortex shedding; Step 4: Extract the vortex shedding characteristic parameters output by the bridge vortex shedding intelligent identification model and dynamically adjust the parameters and algorithms of the lidar; Step 5: Based on the vortex shedding characteristic parameters output by the bridge vortex shedding intelligent identification model, track the vortex shedding evolution process, construct a spatiotemporal evolution map of the vortex shedding trajectory, and then obtain the dominant spatiotemporal modes of the vortex shedding evolution process and their energy contribution ratio; Step 6: Based on the dominant spatiotemporal modes of the vortex shedding evolution process and their energy contribution ratio, a wind speed gradient-vortex shedding intensity mapping relationship is constructed, and a vortex shedding instability prediction model is established.

2. The method for intelligent identification and tracking of vortex shedding based on in-situ flow field data of a bridge according to claim 1 is characterized in that: The specific steps of step 1 include: Step 11: Collect global flow field data of the bridge by coordinating the lidar and anemometer, and then calculate the flow field characteristic parameters based on the collected global flow field data of the bridge, where the flow field characteristic parameters include extinction coefficient, spectral width, skewness, wind vector, rotation, and shear intensity; Step 12: Extract multi-scale turbulence fluctuation information based on the wavelet transform method, and combine the wind vector and flow field characteristic parameters to finely analyze the flow field structure; Step 13: Analyze the evolution of flow field characteristic parameters in time and space, and extract the spatiotemporal evolution characteristics of the flow field, wherein the spatiotemporal evolution characteristics of the flow field include the generation, evolution process and change cycle of the flow field structure.

3. The intelligent identification and tracking method of vortex shedding based on in-situ flow field data of a bridge according to claim 2 is characterized in that: In step 11: The extinction coefficient is directly measured by the lidar and is part of the bridge's full-area flow field data; The expression of spectral width is: Where f(v) is the spectrum of the reflected signal in the global flow field data of the bridge, v represents the Doppler shift frequency of the laser signal, is the average frequency of the spectrum; The expression for skewness is: Where σ is the spectrum width; The wind vector expression in the bridge global flow field data is: Where x is the position coordinate along the downwind direction, y is the position coordinate along the crosswind direction, z is the position coordinate along the vertical direction, t is the time, u(x,y,z,t) is the downwind wind speed, v(x,y,z,t) is the crosswind wind speed, and w(x,y,z,t) is the vertical wind speed. The curl expression is: Among them, the specific components of the curl along the x, y, and z axes are: In the formula is the symbol of partial derivative; Shear strength: In step 12: The wavelet transform form is: Where u(t) is the time series, a is the scale parameter, τ is the time shift, is the mother wavelet function; The turbulent energy distribution at each scale is: E u (a,τ)=|W u (a,τ)| 2 ; After combining the wind vector and flow field characteristic parameters, the multi-scale flow intensity index expression is: Where, α i is the weight coefficient, i takes 1, 2, 3, 4, E u (a,τ)=|W u (a,τ)| 2 is the turbulent energy distribution at each scale, S(x,y,z,t) and σ(x,y,z,t) are the curl, shear intensity, and spectral width at a certain position (x,y,z) in the flow field at time t.

4. The method for intelligent identification and tracking of vortex shedding based on in-situ flow field data of a bridge according to claim 1 is characterized in that: The specific steps of step 2 include: Step 21: Determine the vortex shedding dominant frequency based on the spectrum analysis method; Step 22: Analyze the spatial distribution of velocity field and scattering intensity in the bridge's full-area flow field data, and estimate the spatial scale parameters of vortex shedding based on the LiDAR measurement geometry. Step 23: Calculate the vorticity distribution based on the velocity field data, and analyze the rotation intensity and structure inside the vortex.

5. The intelligent identification and tracking method of vortex shedding based on in-situ flow field data of a bridge according to claim 1 is characterized in that: The step 3 specifically includes: Step 31: Convert the flow field characteristic parameter and vortex shedding characteristic parameter data into a unified format, delete duplicate data entries, and encode the categorical variables; after the data is standardized, construct a matrix containing time series and spatial distribution characteristics to obtain the flow field characteristic parameter data set and the vortex shedding characteristic parameter data set; Step 32: Input the flow field characteristic parameter dataset and the vortex shedding characteristic parameter dataset into the deep learning model to train the deep learning model and obtain a bridge vortex shedding intelligent recognition model.

6. The method for intelligent identification and tracking of vortex shedding based on in-situ flow field data of a bridge according to claim 1 is characterized in that: The specific steps of step 4 include: Step 41: extracting vortex shedding characteristic parameters output by the bridge vortex shedding intelligent identification model, wherein the vortex shedding characteristic parameters include vortex shedding intensity, vortex shedding frequency, vortex shedding size, and vortex shedding position; Step 42: Dynamically adjust the pulse repetition frequency and integration time of the lidar based on the accuracy and effectiveness of the vortex shedding characteristic parameters, and increase the scanning sampling in the high vortex shedding intensity area; Step 43: Using Kalman filtering to perform real-time denoising on the raw data, and using the vortex shedding characteristic parameter feedback optimization algorithm based on the output of the bridge vortex shedding intelligent identification model; Step 44: For asymmetric vortex shedding events, adaptively switch the scanning mode to ensure complete capture of the vortex shedding three-dimensional trajectory; Step 45: Establish a parameter adjustment log library to record the influence of lidar configuration and algorithm version on the accuracy of vortex shedding identification.

7. The method for intelligent identification and tracking of vortex shedding based on in-situ flow field data of a bridge according to claim 1 is characterized in that: The specific steps of step 5 include: Step 51: Based on the vortex shedding intensity, frequency, size, and location output by the bridge vortex shedding intelligent identification model, the vortex cores at consecutive moments are matched for spatial proximity and intensity similarity to track the complete motion trajectory of vortex generation, shedding, merging, and dissipation. The spatial path, time history, and intensity evolution information of each vortex are integrated to construct a visual spatiotemporal evolution map of the vortex shedding trajectory. Step 52: Different evolution modes are distinguished based on the cluster analysis method. The dominant space-time modes of the vortex shedding evolution process are obtained by using the intrinsic orthogonal decomposition, and the energy contribution ratio of the dominant space-time modes is analyzed. The different evolution modes include symmetric shedding, asymmetric merging, and secondary vortex excitation.

8. The method for intelligent identification and tracking of vortex shedding based on in-situ flow field data of a bridge according to claim 1 is characterized in that: The specific steps of step 6 include: Step 61: Based on the dominant spatiotemporal modes of the vortex shedding evolution process and their energy contribution ratios, analyze the interaction between the primary vortex and the secondary vortex, and quantify its impact on the vortex shedding frequency and energy dissipation. The interaction between the primary vortex and the secondary vortex includes vortex pair merging, vortex filament breakage, and vortex shedding locking. Step 62: Calculate the modulation effect of non-uniform incoming flow on the spatial distribution and frequency of vortex shedding, construct a mapping relationship between wind speed gradient and vortex shedding intensity, and predict the critical threshold for the transition from periodic shedding to turbulence. Combined with the interaction between the primary vortex and the secondary vortex in step 61, a vortex shedding instability prediction model is obtained.

Images