A method for field wind field data acquisition and processing for structural wind-induced fatigue assessment

Abstract

A kind of field wind field data acquisition and processing method for structural wind-induced fatigue assessment, comprising the following steps: step one, wind speed and direction monitoring system is laid out in target engineering field, and original wind field signal is obtained;Original wind field signal includes original wind speed voltage and original wind direction voltage;Step two, the analysis and outlier processing of original signal are carried out, and the pre-processed field measured wind time history data are obtained;Step three, measured wind time history data are converted into comparable daily scale wind force level and dominant wind direction index;Step four, measured wind time history data are corrected using correction model;Step five, local wind field data set for fatigue assessment is generated.The present application solves the technical problem that the prior art is difficult to obtain wind field data set with long-term statistical representativeness and local site accuracy simultaneously.

Description

Technical Field

[0001] This invention relates to the fields of structural health monitoring and wind engineering, and particularly to a method for collecting, processing, and predicting the long-term characteristics of field wind field data for large-span spatial structures. It focuses on using the correlation correction between short-term field measurement data and long-term historical meteorological data to generate a long-term, highly representative local wind field dataset, providing reliable data input for structural wind-induced fatigue durability assessment. Background Technology

[0002] Wind-induced fatigue durability assessments of large-span spatial structures (such as the National Speed ​​Skating Oval and large stadiums) require long-term, real-time wind field data from the structure's service location as the load basis. Currently, there are two main methods for obtaining this type of data: Short-term field measurements: Sensors are deployed at the engineering site for monitoring. The data accurately reflects the local wind field, but the monitoring period is limited (usually several months to one or two years), the sample size is small, and it is difficult to cover all wind and climate conditions (such as extreme winds with different return periods). Therefore, it cannot be directly used to evaluate the full life cycle performance of the structure over decades or even hundreds of years.

[0003] Long-term meteorological station data: Meteorological stations have observation data for decades, with complete time series. However, the observation locations are usually far from the specific engineering site. Affected by the urban environment, topography, and local structures, the wind field characteristics of the stations differ significantly from the local wind field at the engineering site. Direct use of such data will introduce bias.

[0004] In existing technologies, statistical analysis either relies solely on short-term measured data, resulting in highly random results, or simply references meteorological station data, failing to reflect local characteristics. Therefore, how to combine short-term, realistic but limited-sample field measured data with long-term, comprehensive but spatially varied meteorological station data to generate a wind field dataset that reflects both the local wind field characteristics of the engineering site and has long-term statistical representativeness has become a pressing technical problem to be solved in this field. Summary of the Invention

[0005] The purpose of this invention is to provide a method for on-site wind field data acquisition and processing for structural wind-induced fatigue assessment, in order to solve the technical problem that existing technologies struggle to obtain wind field datasets that simultaneously possess long-term statistical representativeness and local site accuracy.

[0006] To achieve the above objectives, the present invention adopts the following technical solution.

[0007] A method for acquiring and processing on-site wind field data for structural wind-induced fatigue assessment includes the following steps.

[0008] Step 1: Deploy a wind speed and direction monitoring system at the target project site to acquire raw wind field signals; the raw wind field signals include raw wind speed and voltage. and original wind direction voltage .

[0009] Step two involves analyzing the original signal and processing outliers to obtain preprocessed on-site measured wind time history data.

[0010] Step 3: Convert the measured wind time history data into comparable daily-scale wind force level and prevailing wind direction indicators.

[0011] Step 4: Correct the measured wind time history data using a correction model; Step 4.1: Background survey and analysis of wind field in the project area; Step 4.2: Statistically analyze the percentage of days with each wind force level on a daily scale, the percentage of days with each wind force level from the same period's meteorological station data, and the percentage of days with each wind force level from long-term meteorological station data; Step 4.3: Long-term wind climate correction and prediction based on statistical models: Based on the statistical results of Step 4.2, the short-term measured statistical characteristics are corrected and extrapolated to the long term using either the proportional correction-based Weiber distribution fitting prediction method or the weighted fusion-based Weiber distribution parameter correction method.

[0012] Step 5: Generate a local wind field dataset for fatigue assessment: Based on the long-term wind climate statistical characteristics predicted in Step 4 and the local wind field details obtained in Step 1, it is used to assess structural wind-induced fatigue.

[0013] Preferably, the wind speed acquisition device is an ultrasonic wind speed and direction sensor, which is installed around the building structure and located outside the wake region of the building.

[0014] Preferably, in step two, the analysis of the original signal specifically involves: Step 2.1: Collect the raw wind speed and voltage data from the wind speed and direction monitoring system. The analysis converts the wind speed to the corresponding height at the monitoring point. The calculation is as follows:

[0015] -Measured wind speed voltage, voltage range =4~20V, the wind speed corresponding to 4V is 0m / s, and the wind speed corresponding to 20V is 60m / s; - Height of the monitoring point; - Wind speed at the corresponding height of the monitoring point; Step 2.2, record the wind speed at the corresponding height of the monitoring point. Converted to wind speed at the corresponding altitude of the weather station : ; - Wind speed at a height of h; α - An index related to surface roughness; Then it is known The formula for converting wind speed at height h to wind speed at height h is as follows: ; Step 2.3: Transfer the raw wind direction and voltage signals collected by the system. Analysis converted to wind direction angle Voltage range =4~20V, where 4V corresponds to a wind direction angle of 0° and 20V corresponds to a wind direction angle of 360°. We define 0° as due north, 90° as due east, 180° as due south, and 270° as due west. The wind direction angle θ is related to the wind direction voltage signal voltage. The conversion relationship is as follows: .

[0016] Preferably, outlier processing in step two includes dead point identification and removal and / or data discontinuity processing and / or outlier identification and removal; The method for removing dead points is to directly delete consecutive and unchanging dead point data and move subsequent normal data forward to connect them; The method for handling data discontinuities is as follows: when individual sampling points are missing, the missing data is filled by linear interpolation of adjacent valid data. When data for a period longer than 10 minutes is missing, delete the null values ​​corresponding to the missing period, calculate the mean of the normal data adjacent to the deleted period, connect the data, and insert the mean into the data connection position.

[0017] Outlier identification and removal adopts A dual identification method combining criteria and empirical thresholds: First, calculate the mean μ and standard deviation of wind speed or wind direction every 10 minutes. It will exceed Data within a certain range are marked as suspected outliers; then, based on local extreme wind speed experience values, suspected outliers are re-verified to remove abnormal data exceeding the experience threshold; and finally, the data gaps after removing outliers are processed according to the data discontinuity handling method.

[0018] Preferably, the specific method for step three is as follows: Step 3.1: Take the wind force level corresponding to the maximum 10-minute average wind speed value of the day as the representative of the daily wind force level.

[0019] Step 3.2, calculate the average wind direction angle .

[0020] Step 3.3: Determine the measured daily prevailing wind direction for comparison with weather station data. The determination method is as follows: Step 3.3.1, Data preprocessing: Remove calm wind data with wind speeds less than 0.2 m / s; Step 3.3.2, outlier removal: If the wind direction at the measured point deviates from the average wind direction over the preceding and following 10 seconds by more than 90° and there is no sudden change in wind speed, it is judged as an outlier and will not be included in the prevailing wind direction statistics. Step 3.3.3, Wind direction zone division: based on the average wind direction angle every 2 minutes. Using these as basic units, the calculated basic units are categorized into eight standard wind direction intervals: North, Northeast, East, Southeast, South, Southwest, West, and Northwest. Step 3.3.4, Determining the prevailing wind direction: Locate the moment when the maximum 10-minute average wind speed of the day occurs, and statistically analyze the distribution of wind direction intervals within 5 minutes before and after that moment. The wind direction interval with the highest frequency of occurrence is determined as the prevailing wind direction of the day.

[0021] Preferably, the average wind direction angle The calculation process is as follows: Step 3.2.1, Vector decomposition: Determine the wind direction angle of sample i. Transform into a two-dimensional unit vector ( ), ; Step 3.2.2: After performing an arithmetic mean on the components of the m samples, the mean vector is obtained. ); ; ; Step 3.2.3, Mean wind direction angle Calculation: using the arctangent function in the four quadrants (arc) The mean wind direction angle can be obtained. for: ; Calculated The value range is Then The average wind direction angle can be obtained by converting the value to [0°, 360°). .

[0022] Preferably, the Weiber distribution fitting prediction method based on proportional correction in step four has the following specific steps: Step 4.3.1a: Correct the percentage of days with each wind level in the measured data by using the relationship between long-term meteorological station data and meteorological station data from the same period; ; in, This indicates the percentage of days with wind levels ≥ F after correction. This indicates the percentage of days with winds of force F or higher, based on long-term weather station data. This indicates the percentage of days with winds of level F or higher during the same period, as shown by meteorological station data. This indicates the percentage of days with winds ≥ Level F as measured. Step 4.3.1b: Based on the revised ,generate Number of days with F-level winds within a given time period : ; Furthermore, the wind force sample set is obtained. ; Indicates the maximum wind force level; Step 4.3.1c: Using the Weiber distribution formula and the maximum likelihood estimation method, fit the wind sample data; the likelihood function of the Weiber distribution is as follows: ; in, This refers to the number of days with winds classified as F-level. The shape parameters of the Weibull distribution determine the height of the Weibull distribution. The width of the Weber distribution is determined by the size parameter of the graph. Taking the derivative of this expression and setting it to zero, we obtain the maximum likelihood estimation condition: ; After simplification, we get: ; The values ​​of k and λ can be obtained by solving the system of equations; Step 4.3.1d: Substitute the values ​​of k and λ into the formula This allows us to obtain a Weber distribution prediction model that conforms to the characteristics of the local wind field, and thus obtain wind prediction data for the target time range.

[0023] Preferably, the steps of the weighted fusion-based Weiber distribution parameter correction method in step four are as follows: Step 4.3.1A: Correct the percentage of days with each wind level in the measured data by using the relationship between long-term meteorological station data and meteorological station data from the same period. ; in, This indicates the percentage of days with wind levels ≥ F after correction. This indicates the percentage of days with winds of force F or higher, based on long-term weather station data. This indicates the percentage of days with winds of level F or higher during the same period, as shown by meteorological station data. This indicates the percentage of days with winds ≥ Level F as measured. Step 4.3.1B: Based on the revised The number of days with an F-level wind within a given n-day period. : ; Furthermore, the wind force sample set is obtained. ; Indicates the maximum wind force level; Step 4.3.1C: Using the Weiber distribution formula and the maximum likelihood estimation method, fit the wind sample data; the likelihood function of the Weiber distribution is as follows: ; in, This refers to the number of days with winds classified as F-level. The shape parameters of the Weibull distribution determine the height of the Weibull distribution. The size parameter of the Weber distribution plot determines the width of the Weber distribution plot; Taking the derivative of this expression and setting it to zero, we obtain the maximum likelihood estimation condition: ; After simplification, we get: ; Step 4.3.1D: Perform Weiper distribution fitting on the field measured wind speed data and the long-term meteorological station wind speed data respectively, and finally obtain the parameters of the Weiper distribution model representing the measured wind force. , ) and the parameters of the Weiber distribution model characterizing the wind data from the meteorological station ( , ); Step 4.3.1E: Set the parameters ( , )and( , The parameters k and λ of the Weiber distribution prediction model are obtained by summing them according to their weights, and the calculation formula is as follows: ; Where 'a' represents the shape parameter of the measured wind Weber distribution model. The weighting coefficients, where b represents the size parameter of the measured wind Weber distribution model. The weighting coefficients.

[0024] Step 4.3.1F: Based on experience and historical wind patterns from weather stations, adjust the weighting coefficients a and b to obtain the adjusted Weber distribution prediction model parameters k and λ, and then extend the wind data to the target time range.

[0025] Step 4.3.1G: Substitute the values ​​of k and λ into the formula This allows us to obtain a Weber distribution prediction model that conforms to the characteristics of the local wind field, and thus obtain wind prediction data for the target time range.

[0026] Compared with the prior art, the present invention has the following features and beneficial effects.

[0027] 1. This invention establishes a statistical correlation model between short-term field measurement data and concurrent and long-term meteorological station data. It successfully integrates and extrapolates the precise local wind field characteristics (such as site-corrected wind speed and prevailing wind direction distribution) contained in short-term field measurement data into the complete wind climate statistical laws contained in long-term meteorological data. This generates an artificial wind field dataset that can accurately reflect the details of the wind field at the engineering site and has long-term statistical representativeness (covering various return periods). This fundamentally solves the contradiction between data "representativeness" and "locality" and provides an ideal data foundation that was previously unavailable for life assessment.

[0028] 2. The wind data input by this invention has both long-term statistical representativeness and local site authenticity, making the load sequence on which fatigue damage accumulation calculation is based more closely resemble the actual wind conditions that the structure may encounter during its service life, thus significantly improving the accuracy and reliability of wind-induced fatigue assessment.

[0029] Meanwhile, fatigue life prediction based on this dataset can more reasonably cover the impact of extreme wind events, significantly reducing the randomness or bias of the assessment results caused by insufficient data samples or spatial differences, making the durability assessment conclusions under long recurrence periods such as "once-in-a-century" events more scientific and credible.

[0030] 3. This invention comprehensively utilizes methods such as dead spot removal, outlier identification, and discontinuity interpolation to effectively purify the original monitoring signal, ensuring the quality and continuity of data used in subsequent analysis. By introducing the "daily maximum 10-minute average wind speed" as a representative of daily wind force level and employing the vector averaging method to calculate the average wind direction angle, it avoids directional errors caused by scalar averaging, making the daily-scale wind index more scientific and more in line with engineering comparison habits, facilitating effective comparison and correlation analysis with meteorological station data.

[0031] 4. This invention provides two core correction models: a "Weiper distribution fitting and prediction method based on proportional correction" and a "Weiper distribution parameter correction method based on weighted fusion." The former directly corrects statistical results through the proportion of days, with clear logic; the latter fuses measured data with meteorological station Weipai distribution parameters, offering greater flexibility. Both methods are based on the Weipai distribution, widely used in wind engineering, to statistically describe wind speed, with mature theory and parameters possessing clear physical meaning. Furthermore, the weighted fusion method allows for adjusting the weighting coefficients a and b to balance the reliance on local characteristics of measured data with the long-term patterns of meteorological stations, and can be optimized based on engineering experience or subsequent verification, making the method more adaptable to engineering and improving prediction accuracy.

[0032] 5. The entire process of this invention, from sensor deployment (requiring it to be located outside the wake region), data acquisition, analysis, preprocessing, to index extraction, correlation correction, and final dataset generation, has clear steps and rigorous logic, forming a complete technical closed loop, which is easy to promote and apply in engineering practice.

[0033] Furthermore, this invention eliminates the need for time-consuming and costly long-term continuous monitoring at the engineering site. Instead, it only requires representative short-term (such as a windy season or a year) field measurements, combined with existing free or low-cost historical data from weather stations, to meet the data requirements for long-term evaluation, thus greatly saving time and economic costs.

[0034] 6. This invention effectively overcomes the key challenge of data sources in structural wind-induced fatigue assessment. The generated dataset has both long-term representativeness and local authenticity, greatly improving the accuracy of the assessment. At the same time, the method system is scientific, complete, and economical, and has important theoretical value and broad engineering application prospects. Attached Figure Description

[0035] The present invention will now be described in further detail with reference to the accompanying drawings.

[0036] Figure 1 This is a diagram of the overall architecture of the health monitoring system.

[0037] Figure 2 This diagram illustrates data packet loss received by the storage server.

[0038] Figure 3 This is a schematic diagram showing the superposition of the measured wind force histogram and the fitted curve of the Weiber distribution function. Detailed Implementation

[0039] This method for acquiring and processing on-site wind field data for structural wind-induced fatigue assessment includes the following steps: Step 1: Deploy a wind speed and direction monitoring system at the target project site to acquire raw wind field signals; the raw wind field signals include raw wind speed and voltage. and original wind direction voltage ; The wind speed acquisition equipment is an ultrasonic wind speed and direction sensor. To reduce the interference of local turbulence caused by the target structure's own shape on the sensor, the sensor is installed in an area where the wake of the structure's shape has minimal impact. Specifically, the structure is the speed skating rink project. To reduce the interference of local turbulence caused by the venue's own shape on the sensor, the sensor is installed in an area far from the high-rise residential buildings to the west and relatively less affected by the venue's wake. After on-site survey, the sensor was finally selected to be deployed on the top of a temporary container house about 150m south of the venue. To the east and south of the sensor are parks, about 800m to the south is a residential area, and about 200m to the west is a high-rise residential building. The sensor was installed using steel pipes at a height of 1.2m to reduce the impact of roof turbulence on the collected data and ensure that the collected data accurately reflects the actual wind field conditions of the cable net structure's service environment. The total height of the sensor from the ground is approximately 6.0m.

[0040] The speed skating rink on-site wind field characteristics monitoring acquired approximately 21.38 million lines of data over 243 days, from November 23, 2021 to February 21, 2022; May 2, 2022 to August 9, 2022; August 15, 2022 to August 25, 2022; and April 23, 2024 to June 3, 2024. In addition, to study the wind field characteristics under strong wind conditions, two extra days of data were collected from April 12-13, 2025. The first step is to process the time series of the acquired data. The data received by the storage server is affected by the information transmission system, resulting in delays and packet loss. Figure 2 (The arrow indicates a data packet loss scenario). On-site data uploads also employ two modes: timed transmission and real-time transmission at uniform time intervals. For these reasons, data processing must be based on the time the wind field characteristic data is read from the field. When processing the raw data, the collected raw data must first be rearranged according to the sensor reporting time.

[0041] Step two: After the health monitoring system completes the acquisition of raw wind field data, it discovers that the raw data contains issues such as unusable signal types, dead points, and outliers, making it unsuitable for direct use in wind field characteristic studies. Therefore, systematic preliminary processing is required. Specifically, the raw data acquired by the system is a voltage signal, which needs to be analyzed and converted before dead points and outliers are removed to obtain preprocessed on-site measured wind time history data. Specifically, the analysis of the original signal involves: Step 2.1: Collect the raw wind speed and voltage data from the wind speed and direction monitoring system. The analysis converts the wind speed to the corresponding height at the monitoring point. The calculation is as follows:

[0042] - Measured wind speed voltage (unit: V), voltage range =4~20V, the original wind speed voltage collected by the sensor here. The wind speed corresponding to 4V is 0m / s, and the wind speed corresponding to 20V is 60m / s; - Height of the monitoring point; - Wind speed at the corresponding height of the monitoring point; Step 2.2, record the wind speed at the corresponding height of the monitoring point. Converted to wind speed at the corresponding altitude of the weather station : ; - Wind speed value at an altitude of h; α - an index related to ground roughness; in this project, the value of α is taken as 0.22 according to the urban environment. Then it is known The formula for converting wind speed at height h to wind speed at height h is as follows: ; Step 2.3: Transfer the raw wind direction and voltage signals collected by the system. Analysis converted to wind direction angle (°); Voltage range =4~20V, the wind direction voltage signal collected by the sensor here. At 4V, the corresponding wind direction angle is 0°, and at 20V, the corresponding wind direction angle is 360°. We define 0° as due north, 90° as due east, 180° as due south, and 270° as due west. The wind direction angle θ is related to the wind direction voltage signal voltage. The conversion relationship is as follows: .

[0043] The wind time-history data initially obtained through data analysis still contains outliers such as dead spots, data discontinuities, and outliers. These outliers can interfere with the accuracy of wind field characteristic analysis, leading to biased results. Therefore, targeted outlier identification and removal are necessary. Based on the characteristics of the field-measured data in this study, the following classification and processing strategy is formulated: The identification and removal of dead data points and / or handling of data discontinuities and / or identification and removal of outliers are crucial. Dead data points are characterized by wind speed and direction data remaining constant across multiple consecutive sampling times, contradicting the dynamic physical characteristics of actual wind fields. To balance the response characteristics of the monitoring system with the turbulent fluctuations of natural wind speed, this study uses a continuous 3-minute period of unchanged wind speed values ​​as the criterion for identifying dead data points. The removal method involves directly deleting continuously unchanged dead data points and shifting subsequent normal data forward to ensure the continuity and effectiveness of the data sequence.

[0044] Discontinuities manifest as gaps in the data sequence, specifically categorized into two types: missing individual data points and missing data over extended periods (usually longer than 10 minutes). The handling of data discontinuities involves using linear interpolation of adjacent valid data points to fill the missing data when individual sampling points are missing, ensuring the smoothness of the data sequence. When data for a period longer than 10 minutes is missing, delete the null values ​​corresponding to the missing period, calculate the mean of the normal data adjacent to the deleted period, connect the data, and insert the mean into the data connection position.

[0045] Outlier identification and removal: Outliers are characterized by abrupt changes in data values ​​and significant deviations from the normal distribution range of wind field data. In engineering, this is typically achieved through... A dual identification method combining criteria and empirical thresholds: First, calculate the mean μ and standard deviation of wind speed or wind direction every 10 minutes. It will exceed The data within the range is marked as suspected outliers; then, combined with the local extreme wind speed empirical value of 40m / s (instantaneous wind speed during summer storms and heavy rain), the suspected outliers are verified a second time to remove abnormal data that exceed the empirical threshold; then, the data gaps after removing outliers are processed according to the data discontinuity processing method.

[0046] Step 3: In order to establish the comparability between the field measured data and the meteorological station data, the measured wind time history data are converted into comparable daily-scale wind force level and prevailing wind direction indicators in accordance with meteorological observation specifications. The specific method is as follows: Step 3.1: Take the wind force level corresponding to the maximum 10-minute average wind speed value of the day as the representative of the daily wind force level; Step 3.2, calculate the average wind direction angle ; Mean wind angle The calculation process is as follows: Step 3.2.1, Vector decomposition: Determine the wind direction angle of sample i. Transform into a two-dimensional unit vector ( ), ; Step 3.2.2: After performing an arithmetic mean on the components of the m samples, the mean vector is obtained. ); ; ; Step 3.2.3, Mean wind direction angle Calculation: using the arctangent function in the four quadrants (arc) The mean wind direction angle can be obtained. for: ;

[0047] Calculated The value range is Then The average wind direction angle can be obtained by converting the value to [0°, 360°). .

[0048] Step 3.3: Determine the measured daily prevailing wind direction for comparison with weather station data. The determination method is as follows: Step 3.3.1, Data preprocessing: Remove calm wind data with wind speeds less than 0.2 m / s; Step 3.3.2, outlier removal: If the wind direction at the measured point deviates from the average wind direction over the preceding and following 10 seconds by more than 90° and there is no sudden change in wind speed, it is judged as an outlier and will not be included in the prevailing wind direction statistics. Step 3.3.3, Wind direction zone division: based on the average wind direction angle every 2 minutes. Using these as basic units, the calculated basic units are categorized into eight standard wind direction zones: North (N), Northeast (NE), East (E), Southeast (SE), South (S), Southwest (SW), West (W), and Northwest (NW). The specific divisions are as follows: .

[0049] Step 3.3.4, Determining the prevailing wind direction: Locate the moment when the maximum 10-minute average wind speed of the day occurs, and statistically analyze the distribution of wind direction intervals within 5 minutes before and after that moment. The wind direction interval with the highest frequency of occurrence is determined as the prevailing wind direction of the day.

[0050] Step 4: In order to generate a wind field dataset that has both local realism and long-term statistical representativeness, a correction model is used to correct the measured wind time history data. Step 4.1: Background survey and analysis of the wind field in the project area. The specific methods are as follows: Before data association, it is necessary to clarify the large-scale wind field background of the project area as a macro benchmark for data validity and association direction; Research content: Collect concurrent and long-term data and related research reports from meteorological stations in the project area, and analyze its climate type, seasonal wind speed and direction patterns, daily wind speed variation characteristics, frequency of strong winds, and historical extreme values.

[0051] Objective: To verify the measured data: By comparing the characteristics of the measured data initially processed in step three with the regional background, we can verify whether it conforms to the basic climate laws and ensure the reliability of the data acquisition system.

[0052] Establishing a correlation benchmark: This provides a background basis for subsequent quantification of differences between local measured data and regional concurrent and long-term meteorological station data.

[0053] In this embodiment, based on preprocessed measured wind time-history data and combined with the common characteristics of urban wind fields surveyed, the wind force characteristics of the local wind field at the engineering site are studied through statistical comparison with long-term meteorological station data, forming a correlation method between measured data and long-term meteorological station data. Specifically, the data basis is as follows: 243 days of valid measured wind time-history data at the site; meteorological station data selected from continuous observation data from December 25, 2020 to June 3, 2024, totaling 1257 days. During the study, the measured wind time-history data were first converted into daily-scale wind force level data, and then multi-dimensional statistical comparisons were performed with the data from the same period and long-term meteorological stations to explore the anemometer patterns in the engineering area and predict the number of days with each wind force level over 100 years. A comparison of daily wind speed data from the speed skating rink with concurrent wind speed data from meteorological stations revealed a significant difference. Within the 243-day effective measurement period, the number of days with wind speeds of force 1-4 were 4, 150, 68, and 21 days, respectively. In contrast, the meteorological station data for the same period (on the same dates) showed wind speeds of force 1-4 on 42, 155, 42, and 4 days, respectively. Further quantitative analysis showed that, in the 243-day comparison, 42% of the measured wind speeds were higher than the wind speeds observed at the meteorological stations; this proportion increased to 72% when the measured wind speeds reached force 3 and 4. Therefore, it can be determined that the local wind force at the location of the speed skating rink project is significantly higher than the average wind force level in the urban area.

[0054] Step 4.2: Calculate the percentage of days with each wind force level in the daily measured data, the percentage of days with each wind force level in the same period's meteorological station data, and the percentage of days with each wind force level in the long-term meteorological station data; the statistical method is as follows: Step 4.2.1, Data Preparation: Measured daily wind force sequence: Statistical analysis of the daily wind force levels in step three, and the number of days that each wind level occurred within the valid measured number of X days; Meteorological daily wind sequence: Obtain the number of days with winds of all levels from meteorological station data for the same period (X days with the same measured date); at the same time, obtain the number of days with winds of all levels from long-term (L consecutive days, L≫X) meteorological station data.

[0055] Step 4.2.2, Quantitative analysis of differences: The number of days and percentage of winds of each level in the measured sequence were statistically analyzed and compared with the meteorological sequence of the same period.

[0056] Based on the regional background in Step 1, determine whether the local wind field is enhanced, weakened, or basically the same as the regional average wind field.

[0057] In this embodiment, the comparison results of the number of days with wind at each level are shown in Table 1, which includes the measured data, the meteorological station data of the same period as the measured data, and the long-term meteorological station data of 1257 days.

[0058] Table 1

[0059]

[0060] As shown in Table 1, the percentage of days with winds of force 3 and 4 in the meteorological station data for the same period is significantly lower than the corresponding percentage in the long-term meteorological station big data of 1257 days. To ensure that the wind field data can reflect the characteristics of the long-term wind environment, it is necessary to correct the percentage of wind force of each level in the actual measurements based on the long-term meteorological station data of 1257 days, and then extend the number of days with wind occurrence to the target time limit. Two correction and prediction methods are provided here.

[0061] Step 4.3: Long-term wind climate correction and prediction based on statistical models: Based on the statistical results of Step 4.2, the short-term measured statistical characteristics are corrected and extrapolated to the long term using either the proportional correction-based Weiber distribution fitting prediction method or the weighted fusion-based Weiber distribution parameter correction method. Step 5: Generate a local wind field dataset for fatigue assessment: Based on the long-term wind climate statistical characteristics (frequency of wind occurrence at each level, prevailing wind direction, etc.) predicted in Step 4 and the local wind field details obtained in Step 1, it is used for the assessment of structural wind-induced fatigue.

[0062] In this embodiment, the system used for on-site wind field data acquisition and processing is a health monitoring system. The overall architecture diagram of the health monitoring system is shown below. Figure 1This system is a dedicated monitoring system for large-span spatial structures, independently developed based on information technology and the Internet of Things. It adopts a B / S architecture (browser / server mode) with a front-end and back-end separation design. The back-end is developed in Java and stores data in a MySQL database, while the front-end uses the Vue framework (a progressive JavaScript framework for building user interfaces) to construct the user interface. The system is divided into five levels: a sensing and acquisition layer, a network transmission layer, a data aggregation layer, an application analysis layer, and an information transmission and control layer. It integrates functions such as real-time acquisition, transmission, storage, processing, health assessment, anomaly alarms, and visualization of multi-source data. Functionally, the system includes a sensor subsystem, a data acquisition and transmission subsystem, a data processing and control subsystem, and a structural health assessment and early warning subsystem. Wind field monitoring functions fall under the environmental monitoring module of the sensor subsystem, providing stable and reliable technical support for the accurate acquisition of wind field data.

[0063] The core purpose of using this health monitoring system to obtain wind speed and direction data is to provide real and complete on-site wind field input data for the evaluation of wind-induced fatigue durability of cable net structures, and to ensure the reliability of subsequent scientific research. The specific data collection content and implementation details of the wind field characteristics study of the speed skating oval project are as follows: (1) The core sensing equipment is an ultrasonic wind speed and direction sensor. This sensor is included in the environmental monitoring category of the system sensor classification. It has the characteristics of high measurement accuracy, fast response speed and strong environmental adaptability, and can accurately capture the dynamic changes of the wind field.

[0064] (2) The system sampling frequency is set to 1Hz, which can meet the dynamic acquisition requirements of wind time history data and ensure that the original data fully reflects the instantaneous characteristics of the wind field on site.

[0065] In this embodiment, the specific steps of the Weiber distribution fitting prediction method based on proportional correction in step four are as follows: Step 4.3.1a: Correct the percentage of days with each wind level in the measured data by using the relationship between long-term meteorological station data and meteorological station data from the same period; ; in, This indicates the percentage of days with wind levels ≥ F after correction. This indicates the percentage of days with winds of force F or higher, based on long-term weather station data. This indicates the percentage of days with winds of level F or higher during the same period, as shown by meteorological station data. This indicates the percentage of days with winds ≥ Level F as measured. Step 4.3.1b: Based on the revised ,generate Number of days with F-level winds within a given time period : ; Furthermore, the wind force sample set is obtained. ; Indicates the maximum wind force level; Step 4.3.1c: Using the Weiber distribution formula and the maximum likelihood estimation method, fit the wind sample data; the likelihood function of the Weiber distribution is as follows: ; in, This refers to the number of days with winds classified as F-level. The shape parameters of the Weibull distribution determine the height of the Weibull distribution. The width of the Weber distribution is determined by the size parameter of the graph. Taking the derivative of this expression and setting it to zero, we obtain the maximum likelihood estimation condition: ; After simplification, we get: ; By solving the system of equations, we can obtain k. Numerical value; In this embodiment, the shape parameter k = 2.8 and the size parameter of the Weiber distribution can be calculated. = 2.8, Figure 3 The figure shows the superimposed display of the measured wind force histogram and the fitted curve of the Weiber distribution function. Based on the Weiber distribution model obtained from the above fitting, the number of days with winds of various levels over 100 years is predicted, and the statistical results are shown in Table 2 (the wind force levels in the table refer to the average wind force level). Table 2

[0066]

[0067] The data in Table 2 shows that the annual average number of predicted strong wind days of level 4 and above is about 53 days. This number is about twice the average number of strong wind days (28-32 days) of the past 10 years as recorded by meteorological stations, and is very close to the corrected number of measured strong wind days (49 days ≥ level 4).

[0068] Substitute the values ​​of k and λ into the formula By obtaining a Weber distribution prediction model that conforms to the characteristics of the local wind field, the wind force level (e.g., 5) is substituted into the fitted Weber distribution model and multiplied by the target time range (e.g., 100 years = 365 × 100) to obtain the predicted number of days for the wind of that level (level 5 wind), and then the wind force prediction data for the target time range (e.g., 100 years) can be obtained.

[0069] Preferably, the steps of the weighted fusion-based Weiber distribution parameter correction method in step four are as follows: Step 4.3.1A: Correct the percentage of days with each wind level in the measured data by using the relationship between long-term meteorological station data and meteorological station data from the same period. ; in, This indicates the percentage of days with wind levels ≥ F after correction. This indicates the percentage of days with winds of force F or higher, based on long-term weather station data. This indicates the percentage of days with winds of level F or higher during the same period, as shown by meteorological station data. This indicates the percentage of days with winds ≥ Level F as measured. Step 4.3.1B: Based on the revised The number of days with an F-level wind within a given n-day period. : ; Furthermore, the wind force sample set is obtained. ; Indicates the maximum wind force level; Step 4.3.1C: Using the Weiber distribution formula and the maximum likelihood estimation method, fit the wind sample data; the likelihood function of the Weiber distribution is as follows: ; in, This refers to the number of days with winds classified as F-level. The shape parameters of the Weibull distribution determine the height of the Weibull distribution. The width of the Weber distribution is determined by the size parameter of the graph. Taking the derivative of this expression and setting it to zero, we obtain the maximum likelihood estimation condition: ; After simplification, we get: ; Step 4.3.1D: Perform Weiper distribution fitting on the field measured wind speed data and the long-term meteorological station wind speed data respectively, and finally obtain the parameters of the Weiper distribution model representing the measured wind force. , ) and the parameters of the Weiber distribution model characterizing the wind data from the meteorological station ( , ); Step 4.3.1E: Set the parameters ( , )and( , The parameters k and λ of the Weiber distribution prediction model are obtained by summing them according to their weights, and the calculation formula is as follows: ; Where 'a' represents the shape parameter of the measured wind Weber distribution model. The weighting coefficients, where b represents the size parameter of the measured wind Weber distribution model. The weighting coefficients.

[0070] Step 4.3.1F: Based on experience and historical wind patterns from weather stations, adjust the weighting coefficients a and b to obtain the adjusted Weber distribution prediction model parameters k and λ.

[0071] Step 4.3.1G: Substitute the values ​​of k and λ into the formula By obtaining a Weber distribution prediction model that conforms to the characteristics of the local wind field, wind force prediction data for the target time range can be obtained. By substituting the wind force level (e.g., 5) into the fitted Weber distribution model and multiplying it by the target time range (e.g., 100 years = 365 × 100), the predicted number of days for that wind level (5 wind) can be obtained, and wind force prediction data for the target time range (e.g., 100 years) can be obtained.

[0072] In this embodiment, the shape parameters of the Weber distribution fitted to the measured data can be calculated. 2.68. Dimensional parameters 3.65, shape parameter corresponding to long-term weather station data 2.23 Dimensional Parameters 2.7, with weighting coefficients a and b each taking a value of 0.5, we obtain k. 2.46, λ 3.17. Based on the obtained Weiber distribution prediction model, the number of days with winds of various levels over 100 years is predicted. Comparison with urban wind field characteristics reveals that when the measured and meteorological station data are considered with equal weight, the parameters ( , )and( , The percentage of high-wind days obtained by weighted summation significantly increased, with the percentage of days experiencing level 4 winds reaching 18.64%, significantly higher than the corrected measured number of strong wind days. This indicates that by adjusting the parameters ( , )and( , The prediction result after weighted summation is too conservative, and the weighting coefficients need to be adjusted based on experience. Adjust the weighting coefficients so that a... 1. b 0 means that the shape parameter is fitted to the measured data and the size parameter is fitted to the meteorological station data. In this case, the prediction results of the two methods are close.

[0073] By comparing the wind field prediction results of the two methods, it can be seen that: Method 1 (directly calculating parameters k and λ) has a lower dependence on engineering experience, requires no complex parameter tuning, and can quickly obtain the number of days of wind prediction for each level that conforms to engineering experience judgment, making it suitable for rapid assessment scenarios; Method 2 (directly calculating parameters k and λ) , )and( , The weighted summation method offers the advantage of parameter adjustability, allowing for manual intervention in the values ​​of core parameters within a reasonable range. In practical engineering applications, it can flexibly adjust the proportion of days with strong winds (e.g., level 4 and above) according to assessment needs, thereby affecting the safety factor value of wind-induced fatigue damage assessment results. This is suitable for scenarios with personalized requirements for assessment accuracy and safety reserves. In practical applications, the larger the time scale of the measured data, that is, the more days of actual data collection, the higher the reliability after correction. Furthermore, a suitable prediction method can be selected based on specific engineering needs (such as assessment efficiency, accuracy requirements, and safety factor control targets).

[0074] The above embodiments are not exhaustive examples of specific implementation methods, and other embodiments are also possible. The purpose of the above embodiments is to illustrate the present invention, rather than to limit the scope of protection of the present invention. All applications derived from simple variations of the present invention fall within the scope of protection of the present invention.

Claims

1. A method for acquiring and processing on-site wind field data for structural wind-induced fatigue assessment, characterized in that, Includes the following steps: Step 1: Deploy a wind speed and direction monitoring system at the target project site to acquire raw wind field signals; the raw wind field signals include raw wind speed and voltage. and original wind direction voltage ; Step two involves analyzing the original signal and processing outliers to obtain preprocessed on-site measured wind time history data. Step 3: Convert the measured wind time history data into comparable daily-scale wind force level and prevailing wind direction indicators; Step 4: Correct the measured wind time history data using a correction model; Step 4.1: Background survey and analysis of wind field in the project area; Step 4.2: Statistically analyze the percentage of days with each wind force level on a daily scale, the percentage of days with each wind force level from the same period's meteorological station data, and the percentage of days with each wind force level from long-term meteorological station data; Step 4.3: Long-term wind climate correction and prediction based on statistical models: Based on the statistical results of Step 4.2, the short-term measured statistical characteristics are corrected and extrapolated to the long term using either the proportional correction-based Weiber distribution fitting prediction method or the weighted fusion-based Weiber distribution parameter correction method. Step 5: Generate a local wind field dataset for fatigue assessment: Based on the long-term wind climate statistical characteristics predicted in Step 4 and the local wind field details obtained in Step 1, it is used to assess structural wind-induced fatigue.

2. The method for on-site wind field data acquisition and processing for structural wind-induced fatigue assessment according to claim 1, characterized in that: The wind speed and direction monitoring system is an ultrasonic wind speed and direction sensor, which is installed around the building structure and located outside the wake region of the building.

3. The method for on-site wind field data acquisition and processing for structural wind-induced fatigue assessment according to claim 1, characterized in that: In step two, the analysis of the original signal specifically involves: Step 2.1: Collect the raw wind speed and voltage data from the wind speed and direction monitoring system. The analysis converts the wind speed to the corresponding height at the monitoring point. The calculation is as follows: ； -Measured wind speed voltage, voltage range =4~20V, the wind speed corresponding to 4V is 0m / s, and the wind speed corresponding to 20V is 60m / s; - Height of the monitoring point; - Wind speed at the corresponding height of the monitoring point; Step 2.2, record the wind speed at the corresponding height of the monitoring point. Converted to wind speed at the corresponding altitude of the weather station : ； - Wind speed at high altitudes; α - An index related to surface roughness; Then it is known The formula for converting wind speed at height h to wind speed at height h is as follows: ； Step 2.3: Transfer the raw wind direction and voltage signals collected by the system. Analysis converted to wind direction angle Voltage range =4~20V, the wind direction angle corresponding to 4V is 0°, and the wind direction angle corresponding to 20V is 360°. We define 0° as due north, 90° as due east, 180° as due south, and 270° as due west; the wind direction angle θ is related to the wind direction voltage signal voltage. The conversion relationship is as follows: 。 4. The method for on-site wind field data acquisition and processing for structural wind-induced fatigue assessment according to claim 1, characterized in that: Step two outlier handling includes dead spot identification and removal and / or data discontinuity handling and / or outlier identification and removal; The method for removing dead points is to directly delete consecutive and unchanging dead point data and move subsequent normal data forward to connect them; The method for handling data discontinuities is as follows: when individual sampling points are missing, the missing data is filled by linear interpolation of adjacent valid data. When data for a period longer than 10 minutes is missing, delete the null values ​​corresponding to the missing period, calculate the mean of the normal data adjacent to the deleted period, connect the data, and insert the mean into the data connection position. Outlier identification and removal employs 3 A dual identification method combining criteria and empirical thresholds: First, calculate the mean μ and standard deviation of wind speed or wind direction every 10 minutes. It will exceed Data within a certain range are marked as suspected outliers; then, based on local extreme wind speed experience values, suspected outliers are re-verified to remove abnormal data exceeding the experience threshold; and finally, the data gaps after removing outliers are processed according to the data discontinuity handling method.

5. The method for on-site wind field data acquisition and processing for structural wind-induced fatigue assessment according to claim 1, characterized in that: The specific method for step three is as follows: Step 3.1: Take the wind force level corresponding to the maximum 10-minute average wind speed value of the day as the representative of the daily wind force level; Step 3.2, calculate the average wind direction angle ; Step 3.3: Determine the measured daily prevailing wind direction for comparison with weather station data. The determination method is as follows: Step 3.3.1, Data preprocessing: Remove calm wind data with wind speeds less than 0.2 m / s; Step 3.3.2, outlier removal: If the wind direction at the measured point deviates from the average wind direction over the preceding and following 10 seconds by more than 90° and there is no sudden change in wind speed, it is judged as an outlier and will not be included in the prevailing wind direction statistics. Step 3.3.3, Wind direction zone division: based on the average wind direction angle every 2 minutes. Using these as basic units, the calculated basic units are categorized into eight standard wind direction intervals: North, Northeast, East, Southeast, South, Southwest, West, and Northwest. Step 3.3.4, Determining the prevailing wind direction: Locate the moment when the maximum 10-minute average wind speed of the day occurs, and statistically analyze the distribution of wind direction intervals within 5 minutes before and after that moment. The wind direction interval with the highest frequency of occurrence is determined as the prevailing wind direction of the day.

6. The method for on-site wind field data acquisition and processing for structural wind-induced fatigue assessment according to claim 4, characterized in that: Mean wind angle The calculation process is as follows: Step 3.2.1, Vector decomposition: Determine the wind direction angle of sample i. Transform into a two-dimensional unit vector ( ), ； Step 3.2.2: After performing an arithmetic mean on the components of the m samples, the mean vector is obtained. ); ； ； Step 3.2.3, Mean wind direction angle Calculation: using the arctangent function in the four quadrants (arc) The mean wind direction angle can be obtained. for: ； Calculated The value range is [-180, 180], then... Converting the value to [0, 360) will give the average wind direction angle. .

7. The method for on-site wind field data acquisition and processing for structural wind-induced fatigue assessment according to claim 1, characterized in that: Step four involves a Weiber distribution fitting prediction method based on proportional correction, and the specific steps are as follows: Step 4.3.1a: Correct the percentage of days with each wind level in the measured data by using the relationship between long-term meteorological station data and meteorological station data from the same period; ； in, This indicates the percentage of days with wind levels ≥ F after correction. This indicates the percentage of days with winds of force F or higher, based on long-term weather station data. This indicates the percentage of days with winds of level F or higher during the same period, as shown by meteorological station data. This indicates the percentage of days with winds ≥ Level F as measured. Step 4.3.1b: Based on the revised The number of days with an F-level wind within a given n-day period. : Furthermore, a wind force sample set is obtained. ; Indicates the maximum wind force level; Step 4.3.1c: Using the Weiber distribution formula and the maximum likelihood estimation method, fit the wind sample data; the likelihood function of the Weiber distribution is as follows: ； in, This refers to the number of days with winds classified as F-level. The shape parameters of the Weibull distribution determine the height of the Weibull distribution. The size parameter of the Weber distribution plot determines the width of the Weber distribution plot; Taking the derivative of this expression and setting it to zero, we obtain the maximum likelihood estimation condition: 。 After simplification, we get: ； The values ​​of k and λ can be obtained by solving the system of equations; Step 4.3.1d: Substitute the values ​​of k and λ into the formula This allows us to obtain a Weber distribution prediction model that conforms to the characteristics of the local wind field, and thus obtain wind prediction data for the target time range.

8. The method for on-site wind field data acquisition and processing for structural wind-induced fatigue assessment according to claim 1, characterized in that: The steps of the weighted fusion-based Weiber distribution parameter correction method described in step four are as follows: Step 4.3.1A: Correct the percentage of days with each wind level in the measured data by using the relationship between long-term meteorological station data and meteorological station data from the same period. ； in, This indicates the percentage of days with wind levels ≥ F after correction. This indicates the percentage of days with winds of force F or higher, based on long-term weather station data. This indicates the percentage of days with winds of level F or higher during the same period, as shown by meteorological station data. This indicates the percentage of days with winds ≥ Level F as measured. Step 4.3.1B: Based on the revised ,generate Number of days with F-level winds within a given time period : ； Furthermore, a wind force sample set is obtained. ; Indicates the maximum wind force level; Step 4.3.1C: Using the Weiber distribution formula and the maximum likelihood estimation method, fit the wind sample data; the likelihood function of the Weiber distribution is as follows: ； in, This refers to the number of days with winds classified as F-level. The shape parameters of the Weibull distribution determine the height of the Weibull distribution. The size parameter of the Weber distribution plot determines the width of the Weber distribution plot; Taking the derivative of this expression and setting it to zero, we obtain the maximum likelihood estimation condition: ； After simplification, we get: ； Step 4.3.1D: Perform Weiper distribution fitting on the field measured wind speed data and the long-term meteorological station wind speed data respectively, and finally obtain the parameters of the Weiper distribution model representing the measured wind force. , ) and the parameters of the Weiber distribution model characterizing the wind data from the meteorological station ( , ); Step 4.3.1E: Set the parameters ( , )and( , The parameters k of the Weiber distribution prediction model are obtained by summing the weights. The calculation formula is as follows: ； Where 'a' represents the shape parameter of the measured wind Weber distribution model. The weighting coefficients, where b represents the size parameter of the measured wind Weber distribution model. Weighting coefficients; Step 4.3.1F: Based on experience and historical wind patterns from weather stations, adjust the weighting coefficients a and b to obtain the adjusted Weber distribution prediction model parameters k and λ, and then extend the wind data to the target time range. Step 4.3.1G: Substitute the values ​​of k and λ into the formula This allows us to obtain a Weber distribution prediction model that conforms to the characteristics of the local wind field, and thus obtain wind prediction data for the target time range.

Images