Fine reconstruction method for radar detection blind area wind field of anemograph collaborative networking

Abstract

The invention discloses a radar detection blind area wind field fine reconstruction method based on anemograph collaborative networking, and the method comprises the specific steps: firstly, separating wind speed measurement deviation caused by vibration displacement in real time, eliminating vibration interference, carrying out the automatic diagnosis and restoration of data based on machine learning, and optimizing the integrity of wind speed data and the measurement continuity; secondly, an ultrasonic anemograph array optimization arrangement method oriented to global wind field sensing is established, time-space consistency calibration of multi-sensor data is achieved, and the precision and reliability of wind speed measurement are improved; and finally, carrying out data-physical fusion bridge flow field high-resolution intelligent reconstruction, and improving the data integrity of a complex flow field area. The reconstruction method effectively cooperates with the wind measurement laser radar and the anemograph, the advantages of high-frequency single-point measurement and wide-area space scanning of the wind measurement laser radar and the anemograph can be exerted, high-precision reconstruction of the global flow field of the bridge is expected to be achieved, and reliable support is provided for guaranteeing wind resistance safety of a large-span bridge.

Description

Technical Field

[0001] The present invention belongs to the field of bridge wind engineering measurement, and in particular relates to a method for finely reconstructing a wind field in a radar detection blind area using a collaborative network of anemometers. Background Art

[0002] Accurately measuring wind field distribution is crucial for the safety monitoring of long-span bridges, including wind-resistant design, health monitoring, and traffic safety management. Existing wind speed measurement methods primarily include fixed anemometers and laser radar (LiDAR) wind measurement systems. However, single measurement devices have numerous limitations in practical applications, making it difficult to comprehensively and accurately capture complex flow fields around bridges. Fixed anemometers are typically deployed at key locations on bridges to provide localized wind speed data. While these devices offer high accuracy and minimal environmental impact, their limited measurement range prevents them from fully capturing the dynamics of the three-dimensional flow field around bridges. Furthermore, fixed anemometers cannot effectively capture localized turbulence, wind shear, and sudden wind speed changes in large-scale wind fields. LiDAR wind measurement systems can provide wind speed information over a wide range and are suitable for global flow field monitoring. However, because LiDAR relies on the reflective properties of light beams, it has blind spots around bridge structures, particularly in obstructed areas around pylons, main beams, and anemometers. Furthermore, LiDAR's relatively low measurement frequency makes it difficult to capture high-frequency wind speed fluctuations in real time.

[0003] Current research is gradually developing the concept of multi-sensor collaborative measurement, which uses anemometer array networking technology combined with LiDAR data to optimize wind field reconstruction accuracy. However, in practical applications, the rational deployment of anemometers, the spatiotemporal consistency of networking, and data fusion remain challenges. Traditional deployment methods often rely on empirical site selection, lack scientific basis, and are difficult to adapt to the wind field characteristics under different bridge environments. In addition, due to the differences in the measurement scales of anemometers and LiDAR, how to establish a spatial matching relationship between the two, compensate for radar detection blind spots, and improve the integrity of flow field reconstruction remains a key issue that needs to be addressed. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for finely reconstructing the wind field in the radar detection blind area of ​​a collaborative network of anemometers, so as to achieve high-precision wind field reconstruction and improve the wind resistance safety of long-span bridges.

[0005] In order to achieve the above object, the solution of the present invention is:

[0006] A method for finely reconstructing wind fields in radar blind areas detected by anemometer cooperative networking includes the following steps:

[0007] Step 1: pre-process the anemometer data and repair the abnormal data;

[0008] Step 2: Design the optimal ultrasonic anemometer array solution based on the flow characteristics of the bridge surface and the surrounding wind field;

[0009] Step 3: Establish a high-resolution intelligent reconstruction model of the bridge flow field based on data-physics fusion.

[0010] In the above step 1, the wavelet packet-Kalman filter cascade algorithm is used to separate the wind speed measurement deviation caused by vibration displacement;

[0011] First, perform J-level wavelet packet decomposition on the wind speed signal S(t) to obtain the subband signal

[0012]

[0013] in is the wavelet packet basis function;

[0014] Wavelet packet with signal Further decomposed into several modal components u i (t), represents the signal components in different frequency bands:

[0015]

[0016] Where H(.) represents Hilbert transform, ω i is the center frequency of each mode;

[0017] Non-Gaussian noise modeling is introduced, Kalman filtering is used for correction, and the state space model is constructed:

[0018] X t+1 =AX t +W t ,Z t =HX t +V t

[0019] Among them, X t+1 is the actual wind speed state, Z t is the observed wind speed data, W t 、V t is Gaussian white noise.

[0020] In step 1 above, identify abnormal data according to the following method:

[0021] The wind speed signal S(t) is converted into a Grammi angle field image, and a generative adversarial network is used to learn the normal wind speed data distribution, including:

[0022] Generator network G: input noise vector z, output generated samples

[0023]

[0024] Discriminator network D: Input the Grammi angle field image converted from the real wind speed distribution or generate samples Output true or false probability:

[0025] D(S)∈[0,1]

[0026] Through the optimal adversarial loss function:

[0027]

[0028] Train a generative adversarial network to generate samples that match the distribution of wind speed measurements;

[0029] Construct an encoding-decoding network, input the wind speed signal S(t), and output the reconstructed signal:

[0030]

[0031] Calculate the reconstruction error If E(t) exceeds the control chart threshold, the wind speed is determined to be abnormal.

[0032] In step 1 above, abnormal data is repaired based on the improved Bayesian dynamic linear model (BDLM), which includes four steps: model order selection, nonlinear state transition modeling, variational Bayesian inference optimization, and abnormal data self-repair.

[0033] Model order selection: Construct a set of K candidate BDLM models with different orders {M1,M2,…,M k The model score is calculated based on Akaike Information Criterion AIC and Root Mean Square Error RMSE, and then fused into the normalized score index S k , S k The larger the value, the better the model order:

[0034]

[0035]

[0036]

[0037] Where: p k is the number of parameters of the model; is the maximum likelihood estimate of the model on the data; y i is the true observation value; is the predicted output of the model; N is the total number of samples; exp(·) is the exponential function; min j AIC j Represents the minimum AIC value among all candidate models; ε is a very small positive number that prevents the denominator from being zero;

[0038] Multiple candidate models are weighted by Bayesian model average (BMA) to calculate the weighted prediction value:

[0039]

[0040] Where, is the predicted output of the model; is the final prediction value after weighted fusion;

[0041] Nonlinear state transition modeling: In view of the nonlinear characteristics of wind speed measurement error, a nonlinear term f(X t ), construct a generalized state transfer model:

[0042] X t =F t X t-1 +G t +f(X t-1 )+W t

[0043] in,

[0044] f(X t )=αtanh(βX t )

[0045] In the formula, α and β are parameters to be optimized;

[0046] Variational Bayesian Inference Optimization: Using variational inference to optimize the objective function:

[0047]

[0048] Where D KL (.) represents the KL divergence, which improves the model convergence speed by optimizing the variational distribution q(X);

[0049] Abnormal data self-repair: The improved BDLM is used to model and predict the normal time series distribution of wind speed. After detecting an abnormal point, the optimal autoregressive filter is used to repair the abnormal data:

[0050]

[0051] Where λ∈(0,1) is the dynamic adjustment factor.

[0052] The specific process of step 2 above is:

[0053] Assume that the Euler coordinate system of the wind field in the bridge area is expressed as:

[0054] V(x,y,z,t)=(u(x,y,z,t),v(x,y,z,t),w(x,y,z,t))

[0055] Among them, x, y, z represent the three-dimensional spatial position of wind speed measurement; t represents the time dimension of wind speed; (u, v, w) are the three components of wind speed; let the anemometer layout position set be P = {p1, p2, ..., p n}, where each sensor p i With performance parameter set:

[0056] Θ i ={R i ,S i ,T i ,E i}

[0057] Where: Ri is the sensor measurement accuracy; S i is the spatial resolution of the sensor; T i is the sensor signal transmission delay; E i is the energy consumption of the sensor;

[0058] Define the optimization objective function:

[0059]

[0060]

[0061] in: represents the intensity of the wind speed field gradient; is information entropy; represents the signal transmission delay loss; w1, w2, w3 are weight coefficients; τ1 is the wind speed gradient threshold; τ2 is the lower limit of information entropy; τ3 is the maximum allowable signal transmission delay; E max is the upper limit of total energy consumption;

[0062] Aiming at the above multi-objective optimization problem, a genetic algorithm is used to solve it and determine the optimal ultrasonic anemometer array solution.

[0063] The specific process of step 3 above is:

[0064] Step 31: scale-match the high-frequency local data of the anemometer with the low-frequency large-scale data of the lidar to establish a global alignment framework for the measured data of multiple devices;

[0065] Step 32: To address the radar detection blind spot problem, perform flow field spatial interpolation and dynamically correct the spatial interpolation results in combination with experimental calibration parameters;

[0066] Step 33: Reconstruct a high-precision three-dimensional flow field by combining the physical characteristics of the flow field with the measured data, and verify it with the measured data.

[0067] The specific process of the above step 31 is:

[0068] Scale matching and global alignment: Use scale normalization transformation S(f s ,f l ) matches the anemometer and lidar data in time and space, and builds a global alignment framework for multi-device measured data. The flow field distribution after scale matching and global alignment is:

[0069] V′(x,y,z,t)=S(f s ,f l )V(x,y,z,t)

[0070] Among them, V'(x, y, z, t) is the flow field coordinate system after scale matching and global alignment, V(x, y, z, t) is the Euler coordinate system of the wind field in the bridge area, x, y, and z represent the three-dimensional spatial positions of wind speed measurement, and t represents the time dimension of wind speed.

[0071] The specific process of the above step 32 is:

[0072] For the blind spots of lidar detection, radial basis function and Kriging interpolation are combined to complete the flow field. The flow field data after interpolation optimization is:

[0073]

[0074] Among them, λ i is the interpolation weight, φ(.) is the basis function, and h(x) is the polynomial trend term.

[0075] The specific process of the above step 33 is:

[0076] Based on the physical constraints of the flow field and experimental calibration parameters C calib Perform dynamic correction, and the high-precision flow field distribution after physical constraint correction is:

[0077]

[0078] Using measured flow field data V exp (x,y,z,t) Evaluate the reconstruction error:

[0079] ε=||V * (x,y,z,t)-V exp (x,y,z,t)||

[0080] To ensure the accuracy and physical consistency of the reconstructed flow field.

[0081] After adopting the above scheme, the beneficial effects of the present invention are as follows:

[0082] (1) Through collaborative networking of anemometers, the temporal and spatial consistency calibration of multi-sensor data is achieved, and compensation is performed in combination with radar measurement data, effectively reducing measurement errors and improving the accuracy and reliability of wind speed measurements;

[0083] (2) An adaptive anemometer deployment strategy based on flow field gradient characteristics and information entropy weights, combined with spatial interpolation methods, is used to refine and fill in radar detection blind spots, thereby improving data integrity in complex flow field areas.

[0084] (3) Using scale matching technology to fuse high-frequency local anemometer measurement data with low-frequency large-scale lidar data, combined with physical constraint optimization interpolation algorithm, high-precision three-dimensional flow field reconstruction is achieved;

[0085] (3) Through the spatial matching search compensation mechanism of geometric array element coupling, the collaborative networking of the anemometer array is realized, and the stability and adaptability of the measurement system under different environmental conditions are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 is a flow chart of the present invention;

[0087] Figure 2 This is the flow chart of GAN learning normal wind speed distribution;

[0088] Figure 3 This is a flow chart of automatic repair of wind speed data. DETAILED DESCRIPTION

[0089] The technical solutions and beneficial effects of the present invention will be described in detail below with reference to the accompanying drawings.

[0090] like Figure 1 As shown, the present invention provides a method for finely reconstructing the wind field in a radar detection blind area using anemometer cooperative networking, comprising the following steps:

[0091] Step 1: Isolate the measurement deviation caused by vibration displacement and perform automatic diagnosis and repair of anemometer data.

[0092] Step 2: Optimize the layout of ultrasonic anemometer arrays for global wind field sensing.

[0093] Step 3: Establish a high-resolution intelligent reconstruction model of the bridge flow field based on data-physics fusion.

[0094] In step 1, a wavelet packet-Kalman filter cascade algorithm is designed to separate the wind speed measurement deviation caused by vibration displacement in real time and eliminate vibration interference; the initial signal of the anemometer is converted into a visual Gramian Angular Field (GAF) image, and the Generative Adversarial Network (GAN) is used to learn the popular distribution of normal data. In combination with the autoencoder network (AE) and control chart theory, an anemometer test data anomaly diagnosis method is established; on this basis, a self-repair mechanism for anemometer abnormal data is established based on the improved Bayesian Dynamic Linear Model (BDLM), thereby optimizing data integrity and measurement continuity.

[0095] In step 2, the flow characteristics of the bridge surface and surrounding wind field are analyzed based on the CFD (Computational Fluid Dynamics) simulation system to identify key areas with significant wind speed changes and detection blind spots. Taking into account multiple factors such as sensor performance parameters, installation location, signal transmission path, and environmental conditions in the bridge area, anemometers are adaptively deployed based on flow field gradient characteristics and information entropy weights. Through the automated timing technology of synchronous sampling of multiple anemometers, a spatial matching search compensation mechanism for geometric array element coupling is established to calibrate the spatiotemporal consistency of the flow field data of the anemometer array network.

[0096] In step 3, the high-frequency local data of the anemometer is scale-matched with the low-frequency large-scale data of the lidar to establish a global alignment framework for the measured data of multiple devices; to address the problem of radar detection blind spots, spatial flow field interpolation is performed based on radial basis functions, Kriging interpolation and other technologies to improve the data integrity in complex flow field areas, and the spatial interpolation results are dynamically corrected in combination with experimental calibration parameters; the physical characteristics of the flow field are combined with the measured data to reconstruct a high-precision three-dimensional flow field and verify it with the measured data.

[0097] Furthermore, the wavelet packet-Kalman filter cascade algorithm described in step 1 includes the following steps:

[0098] First, perform J-level wavelet packet decomposition on the wind speed signal S(t) to obtain the subband signal

[0099]

[0100] in is the wavelet packet basis function.

[0101] Wavelet packet with signal Further decomposed into several modal components u i (t), represents the signal components in different frequency bands:

[0102]

[0103] Where H(.) represents Hilbert transform, ω i is the center frequency of each mode.

[0104] Non-Gaussian noise modeling is introduced, Kalman filtering is used for correction, and the state space model is constructed:

[0105] X t+1 =AX t +W t ,Z t =HX t +V t

[0106] Among them, X t+1 is the actual wind speed state, Z t is the observed wind speed data, W t 、V t is Gaussian white noise.

[0107] The anemometer test data abnormality diagnosis method described in step 1 includes the following process:

[0108] like Figure 2 As shown in the figure, the wind speed signal S(t) is converted into a GAF image, and GAN is used to learn the normal wind speed data distribution, including:

[0109] Generator network G: input noise vector z, output generated samples

[0110]

[0111] Discriminator network D: Input GAF image or generated sample converted from real wind speed distribution Output true or false probability:

[0112] D(S)∈[0,1]

[0113] Through the optimal adversarial loss function:

[0114]

[0115] GAN is trained to generate samples that conform to the distribution of wind speed measurements and improve anomaly detection performance.

[0116] Construct an encoding-decoding AE network, input the wind speed signal S(t), and output the reconstructed signal:

[0117]

[0118] Calculate the reconstruction error If E(t) exceeds the control chart threshold, the wind speed is determined to be abnormal.

[0119] Furthermore, if Figure 3 As shown in Figure 1, the wind speed anomaly detection and data self-repair method based on the improved BDLM described in step 1 includes four steps: model order selection, nonlinear state transition modeling, variational Bayesian inference optimization, and abnormal data self-repair. Specifically, adaptive model order selection is used to optimize parameter updates, nonlinear state transition modeling is combined to improve adaptability to non-Gaussian noise, and variational Bayesian inference is used to accelerate convergence. Furthermore, through the abnormal data self-repair mechanism, the optimal autoregressive filtering strategy is used to correct abnormal wind speed data, thereby improving the integrity and continuity of the measurement data. The detailed steps are as follows:

[0120] Model order selection: Construct a set of K candidate BDLM models with different orders {M1,M2,…,M k The model score is calculated based on the Akaike information criterion (AIC) and the root mean square error (RMSE), and then fused into the normalized score index S k , S k The larger the value, the better the model order:

[0121]

[0122]

[0123]

[0124] Where: p k is the number of parameters of the model; is the maximum likelihood estimate of the model on the data; y i is the true observation value; is the predicted output of the model; N is the total number of samples; exp(·) is the exponential function; min j AIC j Represents the minimum AIC value among all candidate models; ε is a very small positive number that prevents the denominator from being zero.

[0125] On the basis of selecting the optimal model order, the Bayesian Model Averaging (BMA) strategy is introduced to integrate the prediction results of multiple better candidate models to avoid model selection bias and improve the robustness and generalization ability of the system. Multiple candidate models are weighted by BMA to calculate the weighted prediction value:

[0126]

[0127] Where, is the predicted output of the model; is the final prediction value after weighted fusion.

[0128] Nonlinear state transition modeling: In view of the nonlinear characteristics of wind speed measurement error, a nonlinear term f(X t ), construct a generalized state transfer model:

[0129] X t =F t X t-1 +G t +f(X t-1 )+W t

[0130] in

[0131] f(X t )=αtanh(βX t )

[0132] In the formula, α and β are parameters to be optimized to enhance the adaptability of the model to nonlinear signals;

[0133] Variational Bayesian Inference Optimization: To address the problem of parameter update hysteresis in traditional Bayesian methods, variational inference is used to optimize the objective function:

[0134]

[0135] Where D KL (.) represents the Kullback-Leibler (KL) divergence, which improves the model convergence speed and the ability to identify outliers and repair data by optimizing the variational distribution q(X);

[0136] Abnormal data self-repair mechanism: The improved BDLM is used to model and predict the normal time series distribution of wind speed. After detecting an abnormal point, the optimal autoregressive filter is used to repair the abnormal data:

[0137]

[0138] Where λ∈(0,1) is the dynamic adjustment factor.

[0139] If abnormal points appear continuously, the model prediction part with a higher weight is used to ensure that the repaired wind speed data achieves a balance between continuity and physical consistency.

[0140] Furthermore, the anemometer adaptive deployment method described in step 2 is characterized in that the optimization process is as follows:

[0141] Assume that the Euler coordinate system of the wind field in the bridge area is expressed as:

[0142] V(x,y,z,t)=(u(x,y,z,t),v(x,y,z,t),w(x,y,z,t))

[0143] Among them, x, y, z represent the three-dimensional spatial position of wind speed measurement; t represents the time dimension of wind speed, reflecting the change of wind speed over time; (u, v, w) are the three components of wind speed. Suppose the set of anemometer locations is P = {p1, p2, ..., p n}, where each sensor p i With performance parameter set:

[0144] Θ i ={R i ,S i ,T i ,E i}

[0145] Where: Ri is the sensor measurement accuracy; S i is the spatial resolution of the sensor; T i is the sensor signal transmission delay; E i is the energy consumption of the sensor.

[0146] Define the optimization objective function:

[0147]

[0148]

[0149] in: Indicates the intensity of the wind speed field gradient, which is used to determine the key areas; Information entropy ensures that the information gain of the distribution of measurement points is maximized; represents the signal transmission delay loss, ensuring network synchronization after sensor deployment; w1, w2, and w3 are weight coefficients used to balance different optimization objectives; τ1 is the wind speed gradient threshold, ensuring that the anemometer deployment covers the area with significant wind field changes; τ2 is the lower limit of information entropy, avoiding sensor redundancy; τ3 is the maximum allowable signal transmission delay; E max The total energy consumption limit.

[0150] A genetic algorithm was used to solve the multi-objective optimization problem and determine the optimal ultrasonic anemometer array solution. The solution process is as follows: anemometer placement plans are randomly generated within the bridge area; the current plan is evaluated based on the objective function J(P); a new sensor placement plan is generated using roulette wheel selection and two-point crossover; small-probability perturbations are introduced to adjust individual anemometer positions to avoid local optimality; and if the objective function converges, the optimal placement plan is output.

[0151] Furthermore, the high-precision three-dimensional flow field reconstruction achieved through multi-device data fusion and flow field interpolation optimization described in step 3 includes the following steps:

[0152] Scale matching and global alignment: Use scale normalization transformation S(f s ,f l ) matches the anemometer (high frequency, local) and lidar (low frequency, large range) data in temporal and spatial scales, and builds a multi-device global data alignment framework. The flow field distribution after scale matching and global alignment is:

[0153] V′(x,y,z,t)=S(f s ,f l )V(x,y,z,t)

[0154] Flow field interpolation optimization: For the blind spots of lidar detection, radial basis function and Kriging interpolation are combined to complete the flow field. The flow field data after interpolation optimization (filling the radar blind spots) is:

[0155]

[0156] Among them, λ i is the interpolation weight, φ(.) is the basis function, and h(x) is the polynomial trend term.

[0157] High-precision flow field reconstruction: Based on flow field physical constraints (mass conservation ) and experimental calibration parameter C calib Perform dynamic correction. The high-precision flow field distribution after physical constraint correction is:

[0158]

[0159] Using measured flow field data V exp (x,y,z,t) Evaluate the reconstruction error:

[0160] ε=||V * (x,y,z,t)-V exp (x,y,z,t)||

[0161] To ensure the accuracy and physical consistency of the reconstructed flow field.

[0162] It will be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented in various computer languages, for example, the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0163] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0164] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0165] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0166] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0167] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A wind field fine reconstruction method for radar detection blind area using anemometer cooperative networking, characterized by The steps include: Step 1: pre-process the anemometer data and repair the abnormal data; Step 2: Design the optimal ultrasonic anemometer array solution based on the flow characteristics of the bridge surface and the surrounding wind field; Step 3: Establish a high-resolution intelligent reconstruction model of the bridge flow field based on data-physics fusion.

2. The method according to claim 1, wherein: In step 1, a wavelet packet-Kalman filter cascade algorithm is used to separate the wind speed measurement deviation caused by vibration displacement; First, perform J-level wavelet packet decomposition on the wind speed signal S(t) to obtain the subband signal in is the wavelet packet basis function; Wavelet packet with signal Further decomposed into several modal components u i (t), represents the signal components in different frequency bands: Where H(.) represents Hilbert transform, ω i is the center frequency of each mode; Non-Gaussian noise modeling is introduced, Kalman filtering is used for correction, and the state space model is constructed: X t+1 =AX t +W t ,Z t =HX t +V t Among them, X t+1 is the actual wind speed state, Z t is the observed wind speed data, W t 、V t is Gaussian white noise.

3. The method according to claim 1, wherein: In step 1, abnormal data is identified according to the following method: The wind speed signal S(t) is converted into a Grammi angle field image, and a generative adversarial network is used to learn the normal wind speed data distribution, including: Generator network G: input noise vector z, output generated samples Discriminator network D: Input the Grammi angle field image converted from the real wind speed distribution or generate samples Output true or false probability: D(S)∈[0,1] Through the optimal adversarial loss function: Train a generative adversarial network to generate samples that match the distribution of wind speed measurements; Construct an encoding-decoding network, input the wind speed signal S(t), and output the reconstructed signal: Calculate the reconstruction error If E(t) exceeds the control chart threshold, the wind speed is determined to be abnormal.

4. The method according to claim 1, wherein: In step 1, abnormal data is repaired based on the improved Bayesian dynamic linear model (BDLM), including four steps: model order selection, nonlinear state transition modeling, variational Bayesian inference optimization, and abnormal data self-repair. Model order selection: Construct a set of K candidate BDLM models with different orders {M1,M2,…,M k }, the model score is calculated based on Akaike information criterion AIC and root mean square error RMSE, and then integrated into the normalized score index S k , S k The larger the value, the better the model order: Where: p k is the number of parameters of the model; is the maximum likelihood estimate of the model on the data; y i is the true observation value; is the predicted output of the model; N is the total number of samples; exp(·) is the exponential function; min j AIC j Represents the minimum AIC value among all candidate models; ε is a very small positive number that prevents the denominator from being zero; Multiple candidate models are weighted by Bayesian model average (BMA) to calculate the weighted prediction value: Where, is the predicted output of the model; is the final prediction value after weighted fusion; Nonlinear state transition modeling: In view of the nonlinear characteristics of wind speed measurement error, a nonlinear term f(X t ), construct a generalized state transfer model: X t =F t X t-1 +G t +f(X t-1 )+W t in, f(X t )=αtanh(βX t ) In the formula, α and β are the parameters to be optimized; Variational Bayesian Inference Optimization: Using variational inference to optimize the objective function: Where D KL (.) represents the KL divergence, which improves the model convergence speed by optimizing the variational distribution q(X); Abnormal data self-repair: The improved BDLM is used to model and predict the normal time series distribution of wind speed. After detecting an abnormal point, the optimal autoregressive filter is used to repair the abnormal data: Where λ∈(0,1) is the dynamic adjustment factor.

5. The method according to claim 1, wherein: The specific process of step 2 is: Assume that the Euler coordinate system of the wind field in the bridge area is expressed as: V(x,y,z,t)=(u(x,y,z,t),v(x,y,z,t),w(x,y,z,t)) Among them, x, y, z represent the three-dimensional spatial position of wind speed measurement; t represents the time dimension of wind speed; (u, v, w) are the three components of wind speed; let the anemometer layout position set be P = {p1, p2, ..., p n }, where each sensor p i With performance parameter set: Θ i ={R i ,S i ,T i ,E i } Where: Ri is the sensor measurement accuracy; S i is the spatial resolution of the sensor; T i is the sensor signal transmission delay; E i is the energy consumption of the sensor; Define the optimization objective function: in: represents the intensity of the wind speed field gradient; is information entropy; represents the signal transmission delay loss; w1, w2, w3 are weight coefficients; τ1 is the wind speed gradient threshold; τ2 is the lower limit of information entropy; τ3 is the maximum allowable signal transmission delay; E max is the upper limit of total energy consumption; Aiming at the above multi-objective optimization problem, a genetic algorithm is used to solve it and determine the optimal ultrasonic anemometer array solution.

6. The method according to claim 1, wherein: The specific process of step 3 is: Step 31: scale-match the high-frequency local data of the anemometer with the low-frequency large-scale data of the lidar to establish a global alignment framework for the measured data of multiple devices; Step 32: To address the radar detection blind spot problem, perform flow field spatial interpolation and dynamically correct the spatial interpolation results in combination with experimental calibration parameters; Step 33: Reconstruct a high-precision three-dimensional flow field by combining the physical characteristics of the flow field with the measured data, and verify it with the measured data.

7. The method according to claim 6, wherein: The specific process of step 31 is: Scale matching and global alignment: Use scale normalization transformation S(f s ,f l ) matches the anemometer and lidar data in time and space, and builds a global alignment framework for multi-device measured data. The flow field distribution after scale matching and global alignment is: V′(x,y,z,t)=S(f s ,f l )V(x,y,z,t) Among them, V'(x, y, z, t) is the flow field coordinate system after scale matching and global alignment, V(x, y, z, t) is the Euler coordinate system of the wind field in the bridge area, x, y, and z represent the three-dimensional spatial positions of wind speed measurement, and t represents the time dimension of wind speed.

8. The method according to claim 7, wherein: The specific process of step 32 is: For the blind spots of lidar detection, radial basis function and Kriging interpolation are combined to complete the flow field. The flow field data after interpolation optimization is: Among them, λ i is the interpolation weight, φ(.) is the basis function, and h(x) is the polynomial trend term.

9. The method according to claim 8, wherein: The specific process of step 33 is: Based on the physical constraints of the flow field and experimental calibration parameters C calib Perform dynamic correction, and the high-precision flow field distribution after physical constraint correction is: Using measured flow field data V exp (x,y,z,t) Evaluate the reconstruction error: ε=||V * (x,y,z,t)-V exp (x,y,z,t)|| To ensure the accuracy and physical consistency of the reconstructed flow field.

Images