A method and system for simulating a steady non-Gaussian wind field

Abstract

The present invention discloses a method and system for simulating a stationary non-Gaussian wind field, and relates to the technical field of single-point non-Gaussian wind field process simulation. The method includes establishing an autoregressive model according to a target power spectrum density, determining the relationship between the autoregressive model parameters and the marginal moments between the model input and output, obtaining the moments of non-Gaussian white noise, reconstructing the non-Gaussian white noise signal using the maximum entropy method, obtaining a probability density function and a cumulative probability distribution function, obtaining a translation function using the cumulative probability distribution function, simulating the non-Gaussian white noise process using the translation function, obtaining a non-Gaussian white noise sample, inputting the non-Gaussian white noise sample into the autoregressive model, obtaining a target non-Gaussian process output sample, and establishing a non-Gaussian wind field simulation. The present invention is helpful to simulate a wind field with a stronger non-Gaussian property with higher precision.

Description

Technical Field

[0001] The present invention relates to the technical field of single-point non-Gaussian wind field process simulation, and in particular to a simulation method and system for a stable non-Gaussian wind field. Background Art

[0002] Due to the complexity of wind fields in mountainous areas, scholars have adopted a variety of methods to conduct research, including theoretical modeling, numerical simulation, wind tunnel tests and field measurements. In comparison, field measurements are the most intuitive and reliable method for wind field monitoring, which can not only obtain the average wind speed distribution, but also the pulsating wind speed time history that is more concerned with the structural dynamic response. Since the 1950s, scholars in the field of meteorology have been studying the characteristics of wind in mountainous areas. However, the research mainly focuses on the average wind characteristics of wind in mountainous areas, or the influence of micro-topography such as hills.

[0003] Subsequently, scholars in the field of wind engineering also conducted a series of field measurements of mountain wind. Holmes et al. studied the wind field characteristics of steep slopes, and the results showed that the field measurement results were consistent with the wind tunnel test results. Domestic scholars Zhu Ledong et al. conducted field measurements on the deep canyon wind field at the site of the Baling River Bridge. The results showed that the wind field characteristics were different from those under flat terrain, and its turbulence intensity had a high degree of randomness. Huang Guoqing et al. analyzed the wind field characteristics of a canyon in a remote mountain area, and the study pointed out that the wind field characteristics were significantly different from the results specified in the design specifications. In summary, the complexity of mountain terrain will significantly change the vertical distribution of wind speed and turbulent structure of near-surface layer flow, resulting in significant differences in wind characteristics in mountainous areas and those in flat and open areas. For example, the average wind profile is difficult to describe with a simple exponential law or logarithmic law, and the pulsating wind characteristics are also significantly different from the recommended results of the design specifications. At present, there are few studies on the non-Gaussian pulsating wind characteristics in mountainous areas, and the corresponding wind field mathematical model has not been reported. Therefore, it is necessary to carry out simulation research on non-Gaussian wind fields in mountainous areas.

[0004] At present, many scholars have solved the problem of non-Gaussian wind field or process simulation and proposed some methods, such as time series method, Karhunen-Loeve expansion method, correlation distortion method, and high-order spectrum representation method. The conceptual directness of the correlation function deviation method proposed by Grigoriu (1998) has attracted more attention among these methods and has been widely used in non-Gaussian process simulation.

[0005] Such moment-based translation functions can be explicit, thus facilitating non-Gaussian simulations. Both CDF-based and moment-based correlation distortion methods require the generation of an underlying Gaussian process, which is then converted to a non-Gaussian process based on a translation function. In order to directly generate non-Gaussian processes in correlation distortion methods without any conversion relationship between non-Gaussian and Gaussian, Li and Li (2012) have proposed a time series method to simulate univariate non-Gaussian processes based on time series models, i.e., autoregressive (AR) model and autoregressive moving average (ARMA) model. In this method, the first four moments of the input process (i.e., non-Gaussian white noise) are first fitted in a time series model as a function of the output process (i.e., target process) using a linear regression technique. Based on the first four moments of the input process, they use the Johnson Transformation Model (JTM) to generate non-Gaussian white noise, which is then filtered into a target non-Gaussian process through a time series model. However, the method proposed by Li and Li (2012) for developing the relationship between the first four moments of the input and output processes requires a lot of computation and can only be applied to univariate simulations. In addition, the results provided by this method are not analytical but approximate. Therefore, Ma and Xu (2019) derived the relationship between the first four moments of the input and output processes for multivariate simulations and also used JTM and AR models as the main simulation tools to generate the target non-Gaussian process. However, due to the limited accuracy of JTM, these time series simulation methods perform poorly for processes with strong non-Gaussian characteristics. Moreover, these methods cannot reproduce the bimodal characteristics of hardening processes with bimodal distribution.

[0006] Therefore, it is an urgent problem for those skilled in the art to propose a simulation method and system for a steady non-Gaussian wind field to solve the problem of poor performance of the prior art time series simulation method. Summary of the invention

[0007] In view of this, the present invention provides a method and system for simulating a steady non-Gaussian wind field, which can achieve the effects of accurate data acquisition and manpower saving.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] A method and system for simulating a steady non-Gaussian wind field, comprising the following steps:

[0010] Model building steps: Establish an autoregressive model based on the target power spectral density;

[0011] Moment acquisition step: determining the relationship between the autoregressive model parameters and the marginal moments between the model input and output according to the autoregressive model, and obtaining the moment of non-Gaussian white noise;

[0012] Function calculation steps: Use the maximum entropy method to reconstruct the non-Gaussian white noise signal and obtain the probability density function and cumulative probability distribution function;

[0013] Sample acquisition steps: using the cumulative probability distribution function to obtain the translation function, using the translation function to simulate the non-Gaussian white noise process to obtain the non-Gaussian white noise sample;

[0014] Wind field simulation steps: Input non-Gaussian white noise samples into the autoregressive model to obtain the target non-Gaussian process output samples and establish a non-Gaussian wind field simulation.

[0015] In the above method, optionally, in the model building step, the autoregressive model includes:

[0016]

[0017] Where x(n) is the output random process at time n, p is the order of the autoregressive model, and a k is the kth (k=1,…p) model parameter, and w(n) is the Gaussian white noise input with zero mean.

[0018] In the above method, optionally, the autoregressive parameters include: R(γ)a=R l ;

[0019] Where R(γ) is the p×p order autocorrelation function matrix;

[0020] The twisted relationship between the cross-edge moments in the moment acquisition step determines the input, i.e., the moment of non-Gaussian white noise.

[0021] In the above method, optionally, in the function calculation step, the marginal probability density function is calculated by:

[0022]

[0023] Among them, λ0,λ1,…,λ m is the Lagrange multiplier, and w is a random variable.

[0024] In the above method, optionally, in the function calculation step, a convergence mechanism is used to determine the moment order, including:

[0025] ||H m -H m-2 || / ||H m ||≤ε H ;

[0026] Among them, H m is information entropy, ε H is the accuracy value of relative error.

[0027] In the above method, optionally, in the sample acquisition step, a Gaussian white noise is converted into a non-Gaussian white noise process by a transfer function; the specific calculation process includes:

[0028] w(n)=g[y(n)]=F W -1 {Φ G [y(n)]},

[0029] where g(·) is the translation function; Φ G (·) and F W (·) are the cumulative probability distribution functions of Gaussian and non-Gaussian white noise processes, respectively; F W -1 Yes F W The inverse function of .

[0030] A simulation system for a steady non-Gaussian wind field, applied to any one of the above-mentioned simulation methods for a steady non-Gaussian wind field, comprising a model building module, a moment acquisition module, a function calculation module, a sample acquisition module and a wind field simulation module connected in sequence;

[0031] Model building module: build an autoregressive model based on the target power spectral density;

[0032] Moment acquisition module: Determine the relationship between the autoregressive model parameters and the marginal moments between the model input and output according to the autoregressive model, and obtain the moment of non-Gaussian white noise;

[0033] Function calculation module: Use the maximum entropy method to reconstruct the non-Gaussian white noise signal and obtain the probability density function and cumulative probability distribution function;

[0034] Sample acquisition module: using the cumulative probability distribution function to obtain the translation function, using the translation function to simulate the non-Gaussian white noise process to obtain the non-Gaussian white noise sample;

[0035] Wind field simulation module: non-Gaussian white noise samples are input into the autoregressive model to obtain the target non-Gaussian process output samples and establish non-Gaussian wind field simulation.

[0036] It can be seen from the above technical solution that compared with the prior art, the present invention provides a simulation method and system for a stationary non-Gaussian wind field, which has the following beneficial effects: the stationary non-Gaussian wind field simulation method based on autoregression and maximum entropy provided by the present invention is suitable for the problem of simulating a single variable stationary non-Gaussian wind field with only target wind speed statistical moments and target power spectrum. Compared with traditional moment-based simulation methods (such as JTM), this method can provide better simulation accuracy, especially for wind fields with strong non-Gaussianity. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0038] Figure 1 A flow chart of a method for simulating a steady non-Gaussian wind field disclosed in the present invention;

[0039] FIG2( a ) is a schematic diagram showing a comparison of wind speed time histories at measuring point 1 between the JTM-based method disclosed in this embodiment and the proposed method;

[0040] FIG2( b ) is a schematic diagram showing a comparison of wind speed time histories at measuring point 2 between the JTM-based method disclosed in this embodiment and the proposed method;

[0041] FIG2( c ) is a schematic diagram showing a comparison of wind speed time histories at measuring point 3 between the JTM-based method disclosed in this embodiment and the proposed method;

[0042] FIG3( a ) is a schematic diagram of the power spectrum density of the univariate non-Gaussian wind speed corresponding to the measuring point 1 disclosed in this embodiment;

[0043] FIG3( b ) is a schematic diagram of the power spectrum density of the univariate non-Gaussian wind speed corresponding to the measuring point 1 disclosed in this embodiment;

[0044] FIG3( c ) is a schematic diagram of the power spectrum density of the univariate non-Gaussian wind speed corresponding to the measuring point 1 disclosed in this embodiment;

[0045] FIG4( a ) is a schematic diagram of the normal ordinate of the probability density function of the simulated univariate non-Gaussian wind speed at measuring point 1 disclosed in this embodiment;

[0046] FIG4( b ) is a logarithmic coordinate diagram of the probability density function of the simulated univariate non-Gaussian wind speed at measuring point 1 disclosed in this embodiment;

[0047] FIG4( c ) is a schematic diagram of the probability density function of the simulated single variable non-Gaussian wind speed at measuring point 2 disclosed in this embodiment;

[0048] FIG4( d ) is a schematic diagram of the probability density function of the normal ordinate of the measuring point 3 disclosed in this embodiment simulating the univariate non-Gaussian wind speed;

[0049] Figure 5 The present invention discloses a block diagram of a simulation system for a steady non-Gaussian wind field. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] In this application, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of more restrictions, the elements defined by the sentence "comprise one..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0052] Reference Figure 1 As shown, the present invention discloses a method and system for simulating a steady non-Gaussian wind field, comprising the following steps:

[0053] Model building steps: Establish an autoregressive model based on the target power spectral density;

[0054] Moment acquisition step: determining the relationship between the autoregressive model parameters and the marginal moments between the model input and output according to the autoregressive model, and obtaining the moment of non-Gaussian white noise;

[0055] Function calculation steps: Use the maximum entropy method to reconstruct the non-Gaussian white noise signal and obtain the probability density function and cumulative probability distribution function;

[0056] Sample acquisition steps: using the cumulative probability distribution function to obtain the translation function, using the translation function to simulate the non-Gaussian white noise process to obtain the non-Gaussian white noise sample;

[0057] Wind field simulation steps: Input non-Gaussian white noise samples into the autoregressive model to obtain the target non-Gaussian process output samples and establish a non-Gaussian wind field simulation.

[0058] Furthermore, in the model building step, the autoregressive model includes:

[0059]

[0060] Where x(n) is the output random process at time n, p is the order of the autoregressive model, and a kis the kth (k=1,…p) model parameter, and w(n) is the Gaussian white noise input with zero mean.

[0061] Furthermore, the autoregressive parameters include:

[0062] R(γ)a=R l (2);

[0063] Where R(γ) is the p×p order autocorrelation function matrix;

[0064] The twisted relationship between the edge moments in the moment acquisition step determines the moment of the input, i.e., the non-Gaussian white noise.

[0065] Specifically, in (2),

[0066] a=[a1 a2 … a p ] T ; R l =[R x (1) R x (2) … R x (p)] T ; R(γ) is the p×p order autocorrelation function matrix, which can be written in the form of a block matrix as shown below:

[0067]

[0068] Among them, R x (γ)=E[x(n)x(n-γ)] is the autocorrelation function of the random process x(n), which can be calculated based on the actual wind speed time history data, where γ is the time step lag.

[0069] For low-frequency harmonics in the target power spectral density (PSD), the magnitude of its value has a great impact on the simulation accuracy, so the determination of its value is crucial. Therefore, a simple method can be used to determine the appropriate value of p. First, determine a suitable range around zero, within which the ACF can be considered to have little effect on the PSD value. Generally speaking, the range can be set to ±ασ x 2 , where σ x 2 is the variance of x(n). α can be set empirically. Second, p in the autoregressive model can be considered as R that makes γ fall within a specified range. x The minimum value of (γ).

[0070] Furthermore, in the moment acquisition step, the output process of the autoregressive model can also be expressed as:

[0071]

[0072] where h(i) is the unit impulse response of the system.

[0073] Based on equation (4), the moment-distortion relationship can be analyzed. According to these studies, the moment-distortion relationship of the first six moments can be given by:

[0074] r w,1 =0 (5)

[0075] r w,2 =r x,2 / A (6)

[0076]

[0077]

[0078]

[0079]

[0080] The corresponding parameters are as follows:

[0081]

[0082]

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] After determining the distorted relationship between the edge moments of the input and output, the original moment of the input non-Gaussian white noise process can be given by the following formula:

[0092] U w,i =E(w i )=r w,i σ w i (twenty one),

[0093] In the formula, σ wis the standard deviation of the input process and can be calculated as follows:

[0094]

[0095] Furthermore, in the function calculation step, the marginal probability density function is calculated using:

[0096]

[0097] Among them, λ0,λ1,…,λ m is the Lagrange multiplier, and w is a random variable.

[0098] Specifically, through the relationship between the input and output moments, the first six moments of the input, i.e., non-Gaussian white noise, can be obtained according to the first six moments of the target output process, and its corresponding f W It can be developed using the Maximum Entropy Method (MEM) through the following optimization problem:

[0099]

[0100] Where H is the information entropy of w(n); f w (w) represents the probability density function of w(n); w represents a random variable; i represents the order of the origin moment; m represents the mth order;

[0101] By introducing the Euler-Lagrange equation into equation (23), the probability density function can be approximated as:

[0102]

[0103] where λ0 and λ k (i=1,2,…,m) are Lagrange multipliers, which can be determined by the following convex unconstrained optimization problem:

[0104]

[0105] Furthermore, in the function calculation step, a convergence mechanism is used to determine the moment order, including:

[0106] ||H m -H m-2 || / ||H m ||≤ε H ;

[0107] Among them, H m is information entropy, ε H is the accuracy value of relative error.

[0108] Specifically, ε H Recommended value: 10 -5; Generally, it is necessary to know the integral range of the maximum entropy distribution to ensure that the method (26) converges automatically. In this embodiment, the lower and upper limits of the integral are the true solutions of the following equation about the point of interest C:

[0109]

[0110] m i is the moment of linear shift; The mathematical theory behind the process of determining the integral limits is relatively complex, so for the sake of brevity, other details are omitted here. Once the integral limits are determined, the Lagrange multipliers λ0 and λ in equation (24) are k (i=1,2,…,m) can be obtained through Newton's method and improved Gram-Schmidt algorithm.

[0111] Furthermore, in the sample acquisition step, the Gaussian white noise is converted into a non-Gaussian white noise process through a transfer function; the specific calculation process includes:

[0112] w(n)=g[y(n)]=F W -1 {Φ G [y(n)]},

[0113] where g(·) is the translation function; Φ G (·) and F W (·) are the cumulative probability distribution functions of Gaussian and non-Gaussian white noise processes, respectively; F W -1 Yes F W The inverse function of .

[0114] Specifically, F W -1 It can be calculated by the following formula

[0115]

[0116] Among them, f W is the probability density function (probability density function) of the non-Gaussian white noise w(n).

[0117] generally, An explicit expression for y(n) as a function of w(n) cannot be given, which makes non-Gaussian simulation relatively inconvenient. In order to simulate non-Gaussian processes more directly and conveniently, several moment-based models such as HPM and JTM can be used as translation functions. By using these models, whose parameters can only be determined by the first four moments (mean, standard deviation, skewness, kurtosis) of the target non-Gaussian process, the non-Gaussian process can be explicitly expressed as a function of the underlying standard Gaussian process. However, these moment-based models perform poorly in simulating processes with strong non-Gaussian characteristics, which can be attributed to the fact that the first four moments cannot well reflect the probability characteristics of the non-Gaussian process. Moreover, for some non-Gaussian processes (e.g., binormal variables), their probability distribution cannot be well represented by their first four moments. Therefore, according to the maximum entropy method, the target high-order moment (e.g., the 6th-order moment) can be used to determine f W .

[0118] A simulation system for a steady non-Gaussian wind field, applied to a simulation method for a steady non-Gaussian wind field as described in any one of the above, referring to Figure 5 As shown, it includes a model building module, a moment acquisition module, a function calculation module, a sample acquisition module and a wind field simulation module connected in sequence;

[0119] Model building module: build an autoregressive model based on the target power spectral density;

[0120] Moment acquisition module: Determine the relationship between the autoregressive model parameters and the marginal moments between the model input and output according to the autoregressive model, and obtain the moment of non-Gaussian white noise;

[0121] Function calculation module: Use the maximum entropy method to reconstruct the non-Gaussian white noise signal and obtain the probability density function and cumulative probability distribution function;

[0122] Sample acquisition module: using the cumulative probability distribution function to obtain the translation function, using the translation function to simulate the non-Gaussian white noise process to obtain the non-Gaussian white noise sample;

[0123] Wind field simulation module: non-Gaussian white noise samples are input into the autoregressive model to obtain the target non-Gaussian process output samples and establish non-Gaussian wind field simulation.

[0124] Furthermore, a simulation system for a steady non-Gaussian wind field also includes a memory, a processor, and a computer program stored in the memory and executable on the processor.

[0125] In a specific embodiment:

[0126] In order to verify the effectiveness of the present invention, three numerical examples are considered. Referring to Figures 2(a), 2(b) and 2(c), the wind speeds at measuring points 1, 1 and 3 are used as examples of univariate non-Gaussian simulation. Similarly, the first ten origin matrices of these standardized wind speeds are shown in Tables 1-3, respectively. The results show that the wind speed at measuring point 3 is located in the SU system of JTM, showing strong non-Gaussianity, while the wind speeds at measuring points 1 and 2 are located in the SB system in JTM, showing weaker non-Gaussianity.

[0127] Table 1. Statistical moments of target and simulated wind speed processes at measuring point 1

[0128]

[0129] Table 2. Statistical moments of target and simulated wind speed processes at measuring point 2

[0130]

[0131] Table 3. Statistical moments of target and simulated wind speed processes at measuring point 3

[0132]

[0133] From the data in Tables 1-3, the first six moments of non-Gaussian white noise can be obtained. Using JTM and the proposed method, 100 samples of standardized wind speed time series were simulated. The first six marginal moments and power spectrum density of these samples were easily estimated. It can be seen that the simulation accuracy of the method proposed in this paper is higher than that based on the JTM method. The time series diagrams of simulated wind speed at three measuring points and the ensemble average of the power spectrum density and the ensemble average of the PDF are provided, as shown in Figures 3(a), 3(b) and 3(c). This shows that the power spectrum density obtained by both methods is very close to the target value. Referring to Figures 4(a), 4(b), 4(c) and 4(d), the probability density functions (PDFs) of the samples simulated by JTM and the proposed method are compared. It can be seen that for wind speeds with strong non-Gaussianity, JTM usually underestimates the value of the positive tail of the probability density function, while the PDF estimated by the samples simulated by the method proposed in this paper is in good agreement with the target.

[0134] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without creative work.

[0135] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for simulating a steady non-Gaussian wind field, characterized in that: The following steps are involved: Model building steps: Establish an autoregressive model based on the target power spectral density; Moment acquisition step: determining the relationship between the autoregressive model parameters and the marginal moments between the model input and output according to the autoregressive model, and obtaining the moment of non-Gaussian white noise; Function calculation steps: Use the maximum entropy method to reconstruct the non-Gaussian white noise signal and obtain the probability density function and cumulative probability distribution function; Sample acquisition steps: using the cumulative probability distribution function to obtain the translation function, using the translation function to simulate the non-Gaussian white noise process to obtain the non-Gaussian white noise sample; Wind field simulation steps: non-Gaussian white noise samples are input into the autoregressive model to obtain the target non-Gaussian process output samples and establish non-Gaussian wind field simulation; The moment-distortion relationship of the first six moments of the edge moment is given by: r w,1 ＝0； r w,2 ＝r x,2 / A； The corresponding parameters are as follows: h(i) is the unit impulse response of the system.

2. A method for simulating a steady non-Gaussian wind field according to claim 1, characterized in that: In the model building step, the autoregressive model includes: Where x(n) is the output random process at time n, p is the order of the autoregressive model, and a k is the kth model parameter, k=1,…,p, and w(n) is a Gaussian white noise input with zero mean.

3. A method for simulating a steady non-Gaussian wind field according to claim 2, characterized in that: The autoregressive parameters include: R(γ)a=R l ,a=[a1 a2 … a p ] T ; R l =[R x (1) R x (2) ... R x (p)] T ; Where R(γ) is the p×p order autocorrelation function matrix; The twisted relationship between the cross-edge moments in the moment acquisition step determines the input, i.e., the moment of non-Gaussian white noise.

4. The method for simulating a steady non-Gaussian wind field according to claim 1, characterized in that: In the function calculation step, the marginal probability density function is calculated using: Among them, λ0,λ1,...,λ m is the Lagrange multiplier, and w is a random variable.

5. The method for simulating a steady non-Gaussian wind field according to claim 1, characterized in that: In the function calculation step, a convergence mechanism is used to determine the moment order, including: ||H m -H m-2 || / ||H m ||≤ε H ; Among them, H m is information entropy, ε H is the accuracy value of relative error.

6. A method for simulating a steady non-Gaussian wind field according to claim 1, characterized in that: In the sample acquisition step, the Gaussian white noise is converted into non-Gaussian white noise through a transfer function; The specific calculation process includes: w(n)=g[y(n)]=F W -1 {Φ G [y(n)]}, where g(·) is the translation function; Φ G (·) and F W (·) are the cumulative probability distribution functions of Gaussian and non-Gaussian white noise processes, respectively; F W -1 Yes F W The inverse function of .

7. A simulation system for a steady non-Gaussian wind field, characterized in that: A method for simulating a steady non-Gaussian wind field as described in any one of claims 1 to 6, comprising a model building module, a moment acquisition module, a function calculation module, a sample acquisition module and a wind field simulation module connected in sequence; Model building module: build an autoregressive model based on the target power spectral density; Moment acquisition module: Determine the relationship between the autoregressive model parameters and the marginal moments between the model input and output according to the autoregressive model, and obtain the moment of non-Gaussian white noise; Function calculation module: Use the maximum entropy method to reconstruct the non-Gaussian white noise signal and obtain the probability density function and cumulative probability distribution function; Sample acquisition module: using the cumulative probability distribution function to obtain the translation function, using the translation function to simulate the non-Gaussian white noise process to obtain the non-Gaussian white noise sample; Wind field simulation module: non-Gaussian white noise samples are input into the autoregressive model to obtain the target non-Gaussian process output samples and establish non-Gaussian wind field simulation.

Images