CN106330197A

CN106330197A - A data compression method for building wind tunnel pressure measurement test

Info

Publication number: CN106330197A
Application number: CN201610624414.0A
Authority: CN
Inventors: 苏宁; 孙瑛; 武岳; 沈世钊
Original assignee: Harbin Institute of Technology Shenzhen
Current assignee: Heilongjiang Industrial Technology Research Institute Asset Management Co ltd
Priority date: 2016-08-02
Filing date: 2016-08-02
Publication date: 2017-01-11
Anticipated expiration: 2036-08-02
Also published as: CN106330197B

Abstract

The invention relates to a building wind tunnel pressure measurement test data compression method. The invention aims at solving the problem that the little attention is paid to frequency domain characteristics of wind load and correlation in the prior art, resulting in a data compression error and an actual application error. The statistical information of the data is reconstructed by the Hermite polynomial, and the spectrum information is reconstructed by the Beta function theory, and a compressed wind pressure filed can be reconstructed at last in combination with random simulation technology. By adoption of the building wind tunnel pressure measurement test data compression method provided by the invention, the building wind tunnel pressure measurement data of GB and TB levels can be compressed to the KB and MB levels, and a novel information extracting and modeling method is provided for the high-dimension time-course large data of the wind load changing with time and space so as to achieve the data compression purpose at last. The building wind tunnel pressure measurement test data compression method is applied to the field of building wind tunnel pressure measurement tests.

Description

A data compression method for building wind tunnel pressure measurement test

技术领域technical field

本发明涉及建筑风洞测压试验数据压缩方法。The invention relates to a compression method for building wind tunnel pressure measurement test data.

背景技术Background technique

风洞测压试验是目前大型复杂工程结构抗风设计的关键步骤之一。风洞测压试验获得的脉动风荷载时程数据量大，可达GB甚至TB级别，且蕴含丰富的信息。大数据量导致了数据存储和分析的困难，因此，有必要对数据进行特性提取和压缩存储，以便于数据的积累、分析和预测。建筑风洞试验获得的风荷载时程数据特点可归结为非高斯部分相关的标量场。目前的风洞试验数据压缩，多采用本征向量法，关注风荷载场在时域的主坐标信息，或对风荷载时程的时域统计信息进行建模，对风荷载频域特性及相关性关注较少，导致数据压缩的偏差和实际应用中的误差。Wind tunnel pressure testing is one of the key steps in the wind resistance design of large and complex engineering structures. The time-history data of fluctuating wind load obtained from wind tunnel pressure test is large, reaching GB or even TB level, and contains rich information. The large amount of data leads to difficulties in data storage and analysis. Therefore, it is necessary to extract the characteristics of the data and compress the storage to facilitate data accumulation, analysis and prediction. The characteristics of wind load time history data obtained from building wind tunnel tests can be attributed to non-Gaussian partially correlated scalar fields. The current wind tunnel test data compression mostly uses the eigenvector method, focusing on the principal coordinate information of the wind load field in the time domain, or modeling the time domain statistical information of the wind load time history, and analyzing the frequency domain characteristics of wind load and related Sexuality is less concerned, leading to bias in data compression and errors in practical applications.

发明内容Contents of the invention

本发明是为了解决现有技术中对风荷载频域特性及相关性关注较少，导致数据压缩的偏差和实际应用中的误差的问题，而提出的一种建筑风洞测压试验数据压缩方法。The present invention is to solve the problem that less attention is paid to the frequency domain characteristics and correlation of wind load in the prior art, which leads to the deviation of data compression and the error in practical application, and proposes a data compression method for building wind tunnel pressure measurement test .

一种建筑风洞测压试验数据压缩方法按以下步骤实现：A method for compressing building wind tunnel pressure test data is realized in the following steps:

步骤一：将建筑风洞试验测压得到的风压时间序列数据无量纲化为风压系数时程数据，并计算风压系数的无偏估计平均值、均方根值、偏度值和和峰度值；Step 1: Dimensionally convert the wind pressure time series data obtained from building wind tunnel test pressure measurement into wind pressure coefficient time history data, and calculate the unbiased estimated average value, root mean square value, skewness value and sum of wind pressure coefficient kurtosis value;

步骤二：对风压系数时程进行自功率谱估计得到自功率谱，计算无量纲功率谱曲线的峰值以及曲线高频段在双对数坐标下的斜率；Step 2: Estimate the self-power spectrum for the time history of the wind pressure coefficient to obtain the self-power spectrum, and calculate the peak value of the dimensionless power spectrum curve and the slope of the high-frequency band of the curve in log-logarithmic coordinates;

步骤三：采用Welch法估计风压系数场的相干函数，用指数函数拟合相干函数；Step 3: Estimate the coherence function of the wind pressure coefficient field by Welch method, and fit the coherence function with exponential function;

步骤四：根据步骤二求解带Bata函数的方程得到无量纲自功率谱的表达式，并根据Beta函数得到任意ν阶无量纲谱矩；Step 4: Solve the equation with the Bata function according to the step 2 to obtain the expression of the dimensionless self-power spectrum, and obtain any ν-order dimensionless spectral moment according to the Beta function;

步骤五：根据步骤一、步骤三和步骤四形成压缩后的风荷载数据；Step 5: Form compressed wind load data according to Step 1, Step 3 and Step 4;

步骤六：根据Hermite多项式转换函数，形成互谱矩阵并对风压场进行重构；Step 6: According to the Hermite polynomial conversion function, a cross-spectrum matrix is formed and the wind pressure field is reconstructed;

步骤七：估计极值风荷载；Step 7: Estimate extreme wind loads;

步骤八：计算结构风振响应。Step 8: Calculate the wind-induced response of the structure.

发明效果：Invention effect:

本发明为一种基于Hermite多项式及Beta函数的建筑风洞测压试验数据压缩方法，可将GB、TB级别的建筑风洞测压数据压缩到KB、MB级别，是对风荷载随时间、空间变化的高维时程大数据的一种新型信息提取和建模方法，最终达到数据压缩的目的。本发明的优势在于，建模过程明确简单，能够得到高效存储和应用的数据格式，便于数据的深入挖掘。可通过Hermite多项式对数据的统计信息进行重构，以及通过Beta函数理论对频谱信息进行重构，最终结合随机模拟技术可重建压缩的风压场。从结构抗风设计的应用角度，压缩数据可进行工程要求的抗风分析，简化了繁琐的时程和频谱分析过程，较为实用。The invention is a method for compressing building wind tunnel pressure test data based on Hermite polynomials and Beta functions, which can compress building wind tunnel pressure measurement data of GB and TB levels to KB and MB levels, and is a method for analyzing wind loads over time and space. A new information extraction and modeling method for changing high-dimensional time-history big data, and finally achieves the purpose of data compression. The advantages of the present invention are that the modeling process is clear and simple, a data format for efficient storage and application can be obtained, and it is convenient for deep mining of data. The statistical information of the data can be reconstructed through the Hermite polynomial, and the spectrum information can be reconstructed through the Beta function theory. Finally, the compressed wind pressure field can be reconstructed by combining the stochastic simulation technology. From the application point of view of structural wind resistance design, compressed data can be used for wind resistance analysis required by engineering, which simplifies the tedious time course and spectrum analysis process, which is more practical.

附图说明Description of drawings

图1为本发明流程图；Fig. 1 is a flowchart of the present invention;

图2为基于Bata函数的风压谱建模结果图；Fig. 2 is the wind pressure spectrum modeling result figure based on Bata function;

图3(a)为归一化的风压系数时程图；图中横坐标为时间t_k(s)，纵坐标为归一化的风压系数时程 Figure 3(a) is the time history diagram of the normalized wind pressure coefficient; the abscissa in the figure is the time t _k (s), and the ordinate is the time history of the normalized wind pressure coefficient

图3(b)为风压系数概率密度函数图，图中横坐标为归一化的风压系数纵坐标为概率密度函数；Figure 3(b) is the probability density function diagram of the wind pressure coefficient, and the abscissa in the figure is the normalized wind pressure coefficient The ordinate is the probability density function;

图3(c)风压系数无量纲功率谱密度函数图；图中横坐标为频率f(Hz)；纵坐标为无量纲功率谱密度函数S；Fig. 3(c) The dimensionless power spectral density function diagram of the wind pressure coefficient; the abscissa in the figure is the frequency f (Hz); the ordinate is the dimensionless power spectral density function S;

图4为基于Hermite多项式转换函数法的极值风荷载估计结果与真实值的对比图；Figure 4 is a comparison chart between the estimated results of extreme wind loads and the real values based on the Hermite polynomial transfer function method;

图5为基于压缩数据进行的平板网架风振响应与原始数据计算结果的对比图。Figure 5 is a comparison chart of the wind-induced response of the flat grid based on the compressed data and the calculation results of the original data.

具体实施方式detailed description

具体实施方式一：如图1所示，一种建筑风洞测压试验数据压缩方法包括以下步骤：Specific embodiment one: as shown in Figure 1, a kind of building wind tunnel pressure measurement test data compression method comprises the following steps:

步骤七：估计极值风荷载；Step 7: Estimate extreme wind loads;

具体实施方式二：本实施方式与具体实施方式一不同的是：所述步骤一中将建筑风洞试验测压得到的风压时间序列数据无量纲化为风压系数时程数据，计算风压系数的无偏估计平均值、均方根值、偏度值和和峰度值具体为：Specific embodiment 2: The difference between this embodiment and specific embodiment 1 is that in the step 1, the wind pressure time series data obtained by building wind tunnel test pressure measurement is dimensionless into wind pressure coefficient time history data, and the wind pressure is calculated. The unbiased estimated mean, root mean square, skewness, and kurtosis values of the coefficients are:

将建筑风洞试验测压得到的风压时间序列数据p_i(t_k)，无量纲化为风压系数时程数据其中所述i表示测点号，t为时间，k表示时间序列号，取值为1,2,…,N，N为采样长度，ρ为空气密度，U表示参考高度处来流风速；并计算风压系数的无偏估计平均值均方根值偏度值和峰度值The wind pressure time series data p _i (t _k ) obtained from building wind tunnel test pressure measurement is dimensionless into wind pressure coefficient time history data Wherein said i represents the measuring point number, t is the time, k represents the time series number, and the value is 1, 2, ..., N, N is the sampling length, ρ is the air density, and U represents the incoming wind speed at the reference height; and Compute unbiased estimated mean of wind pressure coefficient RMS Skewness value and kurtosis value

${C C}_{p p,, k k u u} = = [[\frac{N N + + 11}{N N} {((\frac{N N - - 11}{N N}))}^{22} {Σ Σ}_{k k = = 11}^{N N} {[[\frac{{C C}_{p p} (({t t}_{k k})) - - {\overset{&OverBar; &OverBar;}{C C}}_{p p}}{{\overset{~ ~}{C C}}_{p p}}]]}^{44} - - 33 ((N N - - 11))]] \frac{N N - - 11}{((N N - - 22)) ((N N - - 33))} + + 3. 3.$

其它步骤及参数与具体实施方式一相同。Other steps and parameters are the same as those in Embodiment 1.

具体实施方式三：本实施方式与具体实施方式一或二不同的是：所述步骤二中对风压系数时程进行自功率谱估计得到自功率谱，计算无量纲功率谱曲线的峰值以及曲线高频段在双对数坐标下的斜率具体为：Specific embodiment three: the difference between this embodiment and specific embodiment one or two is that in the step two, the wind pressure coefficient time history is estimated from the power spectrum to obtain the self-power spectrum, and the peak value and curve of the dimensionless power spectrum curve are calculated. The slope of the high frequency band under the logarithmic coordinates is specifically:

采用自回归AR模型对进行风压系数时程进行自功率谱估计，得到自功率谱S_Cp(f)，对其无量纲化，表示为频率f无量纲化为其中L表示参考尺度；计算无量纲功率谱曲线S-F曲线的峰值，即S_m＝max{S(F)}，F_m＝argmax{S(F)}；以及曲线高频段在双对数坐标下的斜率The autoregressive AR model is used to estimate the autopower spectrum for the time history of the wind pressure coefficient, and the autopower spectrum S _Cp (f) is obtained, which is dimensionless and expressed as The frequency f is dimensionless as Where L represents the reference scale; calculate the peak value of the dimensionless power spectrum curve SF curve, that is, S _m =max{S(F)}, F _m =argmax{S(F)}; and the high frequency section of the curve is in the double logarithmic coordinates The slope of

${k k}_{22} = = \frac{((K K - - j j + + 11)) \cdot &Center Dot; {Σ Σ}_{k k = = j j}^{K K} l l n no [[S S (({F f}_{k k}))]] \cdot &Center Dot; l l n no (({F f}_{k k})) - - {Σ Σ}_{k k = = j j}^{K K} l l n no [[S S (({F f}_{k k}))]] \cdot \cdot {Σ Σ}_{k k = = j j}^{K K} l l n no (({F f}_{k k}))}{((K K - - j j + + 11)) \cdot &Center Dot; {Σ Σ}_{k k = = j j}^{K K} {[[l l n no (({F f}_{k k}))]]}^{22} - - {[[{Σ Σ}_{k k = = j j}^{K K} ln ln (({F f}_{k k}))]]}^{22}}$

式中，K为半傅里叶变换长度，F_k(k＝1,2,…,K)为离散的无量纲频率，j为高频段频率的指标，取F_j＝1.5F_m。In the formula, K is the half-Fourier transform length, F _k (k=1,2,...,K) is a discrete dimensionless frequency, j is an index of high-frequency frequency, and F _j =1.5F _m .

其它步骤及参数与具体实施方式一或二相同。Other steps and parameters are the same as those in Embodiment 1 or Embodiment 2.

具体实施方式四：本实施方式与具体实施方式一至三之一不同的是：所述步骤三中采用Welch法估计风压系数场的相干函数，用指数函数拟合相干函数的具体过程为：Specific embodiment four: what this embodiment is different from one of specific embodiments one to three is: adopt Welch method to estimate the coherence function of wind pressure coefficient field in described step 3, the concrete process of fitting coherence function with exponential function is:

采用Welch法估计风压系数场的相干函数Coh_ij(f)，用指数函数Coh_ij(f)＝exp(-k_c||f·D_ij/U)拟合相干函数，其中D_ij表示两点间的距离，即，The coherence function Coh _ij (f) of the wind pressure coefficient field is estimated by the Welch method, and the coherence function is fitted by the exponential function Coh _ij (f)=exp(-k _c ||f·D _ij /U), where D _ij represents two The distance between the points, that is,

${k k}_{c c} = = {{{Σ Σ}_{k k = = 11}^{K K} k k \cdot &Center Dot; ln ln [[{Coh Coh}_{i i j j} ((k k \cdot &Center Dot; Δ Δ f f))]] - - \frac{K K + + 11}{22} \cdot &Center Dot; {Σ Σ}_{k k = = 11}^{K K} ln ln [[{Coh Coh}_{i i j j} ((k k \cdot &Center Dot; Δ Δ f f))]]}} \cdot &Center Dot; \frac{66}{K K ((K K + + 11)) ((K K + + 22))} \cdot &Center Dot; \frac{U u}{{D D.}_{i i j j} \cdot &Center Dot; Δ Δ f f}$

式中，Δf＝f_s/2K为频率间隔，f_s为采样频率，k_c为相干指数，U为来流参考风速。In the formula, Δf=f _s /2K is the frequency interval, f _s is the sampling frequency, k _c is the coherence index, and U is the incoming reference wind speed.

其它步骤及参数与具体实施方式一至三之一相同。Other steps and parameters are the same as those in Embodiments 1 to 3.

具体实施方式五：本实施方式与具体实施方式一至四之一不同的是：所述步骤四中求解带Bata函数的方程得到无量纲自功率谱的表达式，并根据Beta函数得到任意ν阶无量纲谱矩的具体过程为：Specific embodiment five: this embodiment is different from one of specific embodiments one to four: in the step 4, solving the equation with the Bata function obtains the expression of the dimensionless self-power spectrum, and obtains any ν order infinite according to the Beta function The specific process of the outline spectrum moment is:

求解带Bata函数的方程Solving equations with Bata functions

${S S}_{m m} \cdot \cdot \frac{11}{α α} \cdot &Center Dot; {((11 - - \frac{11}{{k k}_{22}}))}^{\frac{11 - - {k k}_{22}}{α α}} \cdot &Center Dot; {((- - {k k}_{22}))}^{\frac{11}{α α}} \cdot \cdot B B ((\frac{11}{α α},, \frac{- - {k k}_{22}}{α α})) = = 11$

得到风压谱的频率指数α，进一步得到无量纲自功率谱的表达式其中F′＝F/F_m；根据Beta函数，得到任意ν阶无量纲谱矩The frequency index α of the wind pressure spectrum is obtained, and the expression of the dimensionless self-power spectrum is further obtained Among them, F′=F/F _m ; according to the Beta function, any ν-order dimensionless spectral moment can be obtained

${S S}_{v v} = = {&Integral; &Integral;}_{00}^{∞ ∞} {F f}^{' ' v v - - 11} S S {dF f}^{' '} = = {((- - {k k}_{22}))}^{v v / / α α} \cdot &Center Dot; B B ((\frac{11 + + v v}{α α},, \frac{- - {k k}_{22} - - v v}{α α})) / / B B ((\frac{11}{α α},, \frac{- - {k k}_{22}}{α α}))$

进一步得到归一化的二阶谱矩：Further get the normalized second-order spectral moment:

${λ λ}_{00} = = \sqrt{\frac{{S S}_{22}}{{S S}_{00}}} = = {((- - {k k}_{22}))}^{11 / / α α} \cdot \cdot \sqrt{B B ((\frac{33}{α α},, \frac{- - {k k}_{22} - - 22}{α α})) / / B B ((\frac{11}{α α},, \frac{- - {k k}_{22}}{α α}))} . .$

其它步骤及参数与具体实施方式一至四之一相同。Other steps and parameters are the same as in one of the specific embodiments 1 to 4.

具体实施方式六：本实施方式与具体实施方式一至五之一不同的是：所述步骤五中形成压缩后的风荷载数据具体为：Specific embodiment six: the difference between this embodiment and one of the specific embodiments one to five is that the compressed wind load data formed in the step five is specifically:

形成压缩后的风荷载数据，表达为一个13列数据：The compressed wind load data is formed, expressed as a 13-column data:

$[\begin{matrix} x x & y the y & z z & {\overset{&OverBar; &OverBar;}{C C}}_{p p} & {\overset{~ ~}{C C}}_{p p} & {C C}_{p p,, s the s k k} & {C C}_{p p,, k k u u} & {F f}_{m m} & {S S}_{m m} & {k k}_{22} & {k k}_{c c} & α α & {λ λ}_{00} \end{matrix}]$

前3列是测点三维几何坐标；4～7列为风荷载前四阶统计矩，分别表示平均风压系数、均方根风压系数、风压系数偏度、风压系数峰度，8～10列为风荷载的自功率谱模型，分别表示无量纲谱峰值频率、无量纲谱峰值和风压谱衰减斜率，11列为风荷载相干函数模型，表示相干指数，12～13列为导出参数，分别表示风压谱的频率指数、归一化的二阶谱矩。The first 3 columns are the three-dimensional geometric coordinates of the measuring points; the 4th to 7th columns are the first four order statistical moments of the wind load, respectively representing the average wind pressure coefficient, the root mean square wind pressure coefficient, the wind pressure coefficient skewness, and the wind pressure coefficient kurtosis, 8 Columns ~10 are self-power spectrum models of wind loads, respectively representing the peak frequency of dimensionless spectrum, peak value of dimensionless spectrum, and attenuation slope of wind pressure spectrum; columns 11 are coherence function models of wind loads, indicating coherence index; columns 12-13 are derived parameters , represent the frequency index and the normalized second-order spectral moment of the wind pressure spectrum, respectively.

其它步骤及参数与具体实施方式一至五之一相同。Other steps and parameters are the same as one of the specific embodiments 1 to 5.

具体实施方式七：本实施方式与具体实施方式一至六之一不同的是：所述步骤六中根据Hermite多项式转换函数，形成互谱矩阵并对风压场进行重构具体为：Specific embodiment seven: this embodiment is different from one of specific embodiments one to six in that: according to the Hermite polynomial conversion function in the step six, the cross-spectrum matrix is formed and the wind pressure field is reconstructed as follows:

基于Hermite多项式转换函数法重建风压场，根据Hermite多项式转换函数，结合表征非高斯特性的统计参数γ₃＝C_p,sk、γ₄＝C_p,ku，建立非高斯过程x(t)与高斯过程u(t)的联系，即：The wind pressure field is reconstructed based _on the _Hermite polynomial transfer function method _, and the non _- Gaussian process x(t) and The connection of Gaussian process u(t), namely:

当C_p,ku≥3时，x＝h(u)＝κ[u+h₃(u²-1)+h₄(u³-u)]，或表示为ξ(x)＝1.5b(a+x/κ)-a³，a＝h₃/3h₄，b＝1/3h₄，c＝(b-1-a²)³；When C _p,ku ≥ 3, x=h(u)=κ[u+h ₃ (u ² -1)+h ₄ (u ³ -u)], or expressed as ξ(x)=1.5b(a+x/κ)-a ³ , a=h ₃ /3h ₄ , b=1/3h ₄ , c=(b-1-a ² ) ³ ;

当C_p,ku<3时，u＝h^-1(x)＝b₂x+b₃(x²-γ₃x-1)+b₄(x³-γ₄x-γ₃)， When C _p,ku <3, u=h ^-1 (x)=b ₂ x+b ₃ (x ² -γ ₃ x-1)+b ₄ (x ³ -γ ₄ x-γ ₃ ),

结合互功率谱的表达式形成的互谱矩阵[S_Cp(ω)]，按Combining Expressions for Cross Power Spectrum The formed cross spectrum matrix [S _Cp (ω)], according to

${C C}_{p p} (({t t}_{k k})) = = {\overset{&OverBar; &OverBar;}{C C}}_{p p} + + {\overset{~ ~}{C C}}_{p p} 22 \sqrt{Δ Δ ω ω} {Σ Σ}_{m m = = 11}^{k k} {Σ Σ}_{l l = = 11}^{N N} | | {H h}_{k k m m} (({ω ω}_{m m l l})) | | c c o o s the s [[{ω ω}_{m m l l} t t - - {θ θ}_{k k m m} (({ω ω}_{m m l l})) + + {φ φ}_{m m l l}]],, k k = = 11,, 22,, ... ...,, n no$

对风压场进行重构，其中H_km(ω_ml)为互功率谱矩阵[S_Cp(ω)]的Cholesky分解，θ_km(ω_ml)为H_km(ω_ml)的辅角，为离散的频率，Δω为圆频率间隔，φ_ml为附加相位角。Reconstruct the wind pressure field, where H _km (ω _ml ) is the Cholesky decomposition of the cross power spectrum matrix [S _Cp (ω)], θ _km (ω _ml ) is the auxiliary angle of H _km (ω _ml ), is the discrete frequency, Δω is the circular frequency interval, and φ _ml is the additional phase angle.

其它步骤及参数与具体实施方式一至六之一相同。Other steps and parameters are the same as one of the specific embodiments 1 to 6.

具体实施方式八：本实施方式与具体实施方式一至七之一不同的是：所述步骤七中根据步骤六估计极值风荷载具体为：Embodiment 8: The difference between this embodiment and one of Embodiments 1 to 7 is that in Step 7, the extreme value wind load estimated according to Step 6 is specifically:

基于步骤六的Hermite多项式转换函数估计极值风荷载，g_NG＝h(g)，n₀＝λ₀F_mU/L为平均穿越频率，T＝600s为参考时长。h()为根据步骤六所确定的Hermite函数x＝h(u)＝κ[u+h₃(u²-1)+h₄(u³-u)]或者u＝h^-1(x)＝b₂x+b₃(x²-γ₃x-1)+b₄(x³-γ₄x-γ₃)。Based on the Hermite polynomial transfer function in step 6, the extreme wind load is estimated, _gNG = h(g), n ₀ =λ ₀ F _m U/L is the average crossing frequency, and T=600s is the reference duration. h() is the Hermite function x=h(u)=κ[u+h ₃ (u ² -1)+h ₄ (u ³ -u)] or u=h ^-1 (x) determined according to step six =b ₂ x+b ₃ (x ² -γ ₃ x-1)+b ₄ (x ³ -γ ₄ x-γ ₃ ).

其它步骤及参数与具体实施方式一至七之一相同。Other steps and parameters are the same as one of the specific embodiments 1 to 7.

具体实施方式九：本实施方式与具体实施方式一至八之一不同的是：所述步骤八中根据步骤六计算结构风振响应的具体过程为：Specific embodiment nine: the difference between this embodiment and one of the specific embodiments one to eight is that: in the step eight, the specific process of calculating the structural wind-induced response according to the step six is:

根据步骤六中还原的风荷载互谱矩阵按其中所述[H(ω)]＝{ω²[M]+iω[C]+[K]}-¹为频响函数矩阵，[M]为质量矩阵，[C]为阻尼矩阵，[K]为刚度矩阵，为虚数单位，[S_Cp(ω)]为根据步骤六计算的互谱矩阵，[R]为附属面积转换矩阵，上标*表示共轭转置，T表示转置，-1表示逆矩阵，计算结构风振响应协方差[∑_x]。According to the wind load cross-spectrum matrix restored in step 6, press Wherein [H(ω)]={ω ² [M]+iω[C]+[K]}- ¹ is the frequency response function matrix, [M] is the mass matrix, [C] is the damping matrix, [K ] is the stiffness matrix, is the imaginary unit, [S _Cp (ω)] is the cross-spectrum matrix calculated according to step 6, [R] is the auxiliary area transformation matrix, superscript * means conjugate transpose, T means transpose, -1 means inverse matrix, Calculate the structural wind-induced response covariance [∑ _x ].

其它步骤及参数与具体实施方式一至八之一相同。Other steps and parameters are the same as those in Embodiments 1 to 8.

实施例一：Embodiment one:

平屋盖系列风洞试验，考察了屋盖的长宽比，高宽比，地貌，风速，缩尺比，风向的影响，共进行了286工况的风洞试验，实验数据大小为84.9GB，按照本发明数据压缩后，数据大小为12.9MB，具体实施步骤如下：The series of wind tunnel tests on flat roofs investigated the effects of roof aspect ratio, height-to-width ratio, topography, wind speed, scale ratio, and wind direction. A total of 286 working conditions of wind tunnel tests were carried out, and the experimental data size was 84.9GB. After data compression according to the present invention, data size is 12.9MB, and specific implementation steps are as follows:

步骤一：将风洞试验测得的风压时程无量纲化后进行统计分析，求得风压系数的前四阶统计矩。Step 1: Dimensionless the wind pressure time history measured by the wind tunnel test and then perform statistical analysis to obtain the first four order statistical moments of the wind pressure coefficient.

步骤二：对风压系数时程进行自功率谱分析，求解了自功率谱参数S_m、F_m、k₂，图2给出了建模结果。Step 2: Carry out autopower spectrum analysis on the time history of wind pressure coefficient, and solve autopower spectrum parameters S _m , F _m , k ₂ . Figure 2 shows the modeling results.

步骤三：对风压系数时程进行了相干函数的分析和拟合，求解了各工况下的相干指数k_c。Step 3: Analyze and fit the coherence function of the wind pressure coefficient time history, and solve the coherence index k _c under each working condition.

步骤四：对步骤三得到的自功率谱参数求解了含Beta函数的方程，得到了导出参数α和λ₀。Step 4: Solve the equation containing the Beta function for the autopower spectrum parameters obtained in Step 3, and obtain the derived parameters α and λ ₀ .

步骤五：将步骤一～四得到的结果整合形成压缩后的风荷载数据文件输出。Step 5: Integrate the results obtained in steps 1 to 4 to form a compressed wind load data file output.

为验证该发明的有效性，本文还将压缩数据进行了重构(步骤六)，发现基于该发明的方法，压缩数据与原始数据的统计、频谱、相关特性保持一致，说明了该方法的适用性。此外，为便于工程应用，本文还采用压缩数据估计了平屋盖的极值风荷载(步骤七)，与多样本(1000样本)的精确方法进行了对比，发现模拟值能包络95％以上的真实值，低估时相对误差在10％以内，可以用于工程围护结构设计中；对于主体结构的风振响应分析，按照步骤八给出了不同平板网架的计算结果，发现位移响应的误差在5％以内，结果较为精确，可用于工程结构抗风设计中，实验结果如图3(a)至图5所示。In order to verify the effectiveness of the invention, this paper also reconstructed the compressed data (step 6), and found that based on the method of the invention, the statistics, frequency spectrum, and correlation characteristics of the compressed data and the original data are consistent, which shows the applicability of the method sex. In addition, for the convenience of engineering application, this paper also uses the compressed data to estimate the extreme wind load of the flat roof (step 7). Compared with the accurate method of multi-sample (1000 samples), it is found that the simulated value can cover more than 95% of the The real value, when underestimated, the relative error is within 10%, can be used in the design of the engineering envelope; for the analysis of the wind vibration response of the main structure, the calculation results of different flat grids are given according to step 8, and the error of the displacement response is found Within 5%, the result is relatively accurate and can be used in the wind resistance design of engineering structures. The experimental results are shown in Figure 3(a) to Figure 5.

综上所述，本发明在处理建筑风洞测压试验数据中，能够大大缩减数据的存储空间，得到的数据结果具有较强的可还原性，和较为广泛的工程使用价值。To sum up, the present invention can greatly reduce the storage space of data in the processing of building wind tunnel pressure measurement test data, and the obtained data results have strong reducibility and relatively extensive engineering use value.

Claims

1. a building wind tunnel pressure measuring test data compression method, it is characterised in that described building wind tunnel pressure measuring test data pressure The detailed process of compression method is:

Step one: number of passes when blast time series data nondimensionalization building wind tunnel test pressure measurement obtained is coefficient of wind pres According to, and the unbiased esti-mator meansigma methods of rated wind pressure coefficient, root-mean-square value, degree of bias value and and kurtosis value；

Step 2: coefficient of wind pres time-histories is carried out auto-power spectrum and estimates to obtain auto-power spectrum, calculate dimensionless power spectrum curve Peak value and curve high band slope under log-log coordinate；

Step 3: use Welch method to estimate the coherent function of coefficient of wind pres field, with exponential function matching coherent function；

Step 4: solve the equation of band Bata function according to step 2 and obtain the expression formula of dimensionless auto-power spectrum, and according to Beta function call is to any ν rank dimensionless spectral moment；

Step 5: form the wind load data after compression according to step one, step 3 and step 4；

Step 6: according to Hermite multinomial transfer function, forms cross-spectrum matrix and is reconstructed wind-pressure field；

Step 7: estimate extreme value wind load；

Step 8: computation structure wind vibration response.

A kind of building wind tunnel pressure measuring test data compression method the most according to claim 1, it is characterised in that described step Blast time series data nondimensionalization building wind tunnel test pressure measurement obtained in one is coefficient of wind pres time course data, calculates wind Pressure the unbiased esti-mator meansigma methods of coefficient, root-mean-square value, degree of bias value and and kurtosis value particularly as follows:

Blast time series data p that building wind tunnel test pressure measurement is obtained_i(t_k), nondimensionalization is coefficient of wind pres time course dataWherein said i represents measuring point number, and t is the time, k express time serial number, and value is 1,2 ..., N, N are Sampling length, ρ is atmospheric density, and U represents and flows wind speed at reference altitude；And the unbiased esti-mator meansigma methods of rated wind pressure coefficientRoot-mean-square valueDegree of bias value And kurtosis value

C_{p, k u} = [\frac{N + 1}{N} {(\frac{N - 1}{N})}^{2} Σ_{k = 1}^{N} {[\frac{C_{p} (t_{k}) - {\overset{&OverBar;}{C}}_{p}}{{\tilde{C}}_{p}}]}^{4} - 3 (N - 1)] \frac{N - 1}{(N - 2) (N - 3)} + 3.

A kind of building wind tunnel pressure measuring test data compression method the most according to claim 2, it is characterised in that described step Coefficient of wind pres time-histories carries out in two auto-power spectrum estimate to obtain auto-power spectrum, calculate dimensionless power spectrum curve peak value and Curve high band slope under log-log coordinate particularly as follows:

Use autoregression AR model to carry out auto-power spectrum estimation to carrying out coefficient of wind pres time-histories, obtain auto-power spectrum S_CpF (), to it Nondimensionalization, is expressed asFrequency f nondimensionalization isWherein L represents with reference to yardstick；Calculate immeasurable The peak value of guiding principle power spectrum curve S-F curve, i.e. S_m=max{S (F) }, F_m=argmax{S (F) }；And curve high band is double Slope under logarithmic coordinates

k_{2} = \frac{(K - j + 1) \cdot Σ_{k = j}^{K} l n [S (F_{k})] \cdot l n (F_{k}) - Σ_{k = j}^{K} l n [S (F_{k})] \cdot Σ_{k = j}^{K} l n (F_{k})}{(K - j + 1) \cdot Σ_{k = j}^{K} {[l n (F_{k})]}^{2} - {[Σ_{k = j}^{K} \ln (F_{k})]}^{2}}

In formula, K is half Fourier transformation length, F_kFor discrete dimensionless frequency, k value is 1,2 ..., K；J is high band frequency The index of rate, takes F_j=1.5F_m。

A kind of building wind tunnel pressure measuring test data compression method the most according to claim 3, it is characterised in that described step Use Welch method to estimate the coherent function of coefficient of wind pres field in three, by the detailed process of exponential function matching coherent function be:

Welch method is used to estimate the coherent function Coh of coefficient of wind pres field_ijF (), uses exponential function Coh_ij(f)=exp (-k_cf· D_ij/ U) matching coherent function, wherein D_ijRepresent the distance of point-to-point transmission, i.e.

k_{c} = {Σ_{k = 1}^{K} k \cdot \ln [{Coh}_{i j} (k \cdot Δ f)] - \frac{K + 1}{2} \cdot Σ_{k = 1}^{K} \ln [{Coh}_{i j} (k \cdot Δ f)]} \cdot \frac{6}{K (K + 1) (K + 2)} \cdot \frac{U}{D_{i j} \cdot Δ f}

In formula, △ f=f_s/ 2K is frequency interval, f_sFor sample frequency.

A kind of building wind tunnel pressure measuring test data compression method the most according to claim 4, it is characterised in that described step The equation solving band Bata function in four obtains the expression formula of dimensionless auto-power spectrum, and according to Beta function call to any ν rank The detailed process of dimensionless spectral moment is:

Solve the equation of band Bata function

S_{m} \cdot \frac{1}{α} \cdot {(1 - \frac{1}{k_{2}})}^{\frac{1 - k_{2}}{α}} \cdot {(- k_{2})}^{\frac{1}{α}} \cdot B (\frac{1}{α}, \frac{- k_{2}}{α}) = 1

Obtain frequency index α of blast spectrum, obtain the expression formula of dimensionless auto-power spectrum further Wherein F '=F/F_m；According to Beta function, obtain any ν rank dimensionless spectral moment

S_{ν} = {&Integral;}_{0}^{\infty} F^{' ν - 1} S {dF}^{'} = {(- k_{2})}^{ν / α} \cdot B (\frac{1 + ν}{α}, \frac{- k_{2} - ν}{α}) / B (\frac{1}{α}, \frac{- k_{2}}{α})

Obtain normalized second order spectral moment further:

λ_{0} = \sqrt{\frac{S_{2}}{S_{0}}} = {(- k_{2})}^{1 / α} \cdot \sqrt{B (\frac{3}{α}, \frac{- k_{2} - 2}{α}) / B (\frac{1}{α}, \frac{- k_{2}}{α})} .

A kind of building wind tunnel pressure measuring test data compression method the most according to claim 5, it is characterised in that described step In five formed compression after wind load data particularly as follows:

Form the wind load data after compression, be expressed as 13 column data:

[\begin{matrix} x & y & z & {\overset{&OverBar;}{C}}_{p} & {\tilde{C}}_{p} & C_{p, sk} & C_{p, ku} & F_{m} & S_{m} & k_{2} & k_{c} & α & λ_{0} \end{matrix}]

Front 3 row are measuring point three-dimensional geometry coordinates；4～7 are classified as Fourth square before wind load, 8～10 be classified as wind load from merit Rate spectrum model, 11 are classified as wind load coherency function model, and 12～13 are classified as derived parameter.

A kind of building wind tunnel pressure measuring test data compression method the most according to claim 6, it is characterised in that described step According to Hermite multinomial transfer function in six, formed cross-spectrum matrix and wind-pressure field is reconstructed particularly as follows:

Wind-pressure field is rebuild, according to Hermite multinomial transfer function, in conjunction with characterizing based on Hermite multinomial transfer function method The statistical parameter γ of non-Gaussian feature₃=C_p,sk、γ₄=C_p,ku, set up the connection of nongausian process x (t) and Gaussian process u (t) System, it may be assumed that

Work as C_p,kuWhen >=3, x=h (u)=κ [u+h₃(u²-1)+h₄(u³-u)], Or be expressed asξ (x)=1.5b (a+x/ κ)-a³, a=h₃/ 3h₄, b=1/3h₄, c=(b-1-a²)³；

Work as C_p,ku< when 3, u=h^-1(x)=b₂x+b₃(x²-γ₃x-1)+b₄(x³-γ₄x-γ₃),

Expression formula in conjunction with crosspower spectrumCross-spectrum matrix [the S formed_Cp (ω)], press

K=1,2 ..., n

Wind-pressure field is reconstructed, wherein H_km(ω_ml) it is crosspower spectrum matrix [S_Cp(ω) Cholesky] decomposes, θ_km(ω_ml) For H_km(ω_ml) explement,L=1,2 ..., N is discrete frequency, and Δ ω is circular frequency interval, φ_mlFor additive phase angle.

A kind of building wind tunnel pressure measuring test data compression method the most according to claim 7, it is characterised in that described step In seven according to step 6 estimate extreme value wind load particularly as follows:

Hermite multinomial transfer function based on step 6 estimates extreme value wind load,g_NG=h (g),n₀=λ₀F_mU/L is average cross-over frequency, and T=600s is with reference to duration.

A kind of building wind tunnel pressure measuring test data compression method the most according to claim 8, it is characterised in that described step In eight, the detailed process according to step 6 computation structure wind vibration response is:

Press according to the wind load cross-spectrum matrix of reduction in step 6Its Described in [H (ω)]={ ω²[M]+iω[C]+[K]}^-1For frequency response function matrix, [R] is attached area transition matrix, calculates Wind induced structural vibration response covariance [∑_x]。