Wind turbine reliability assessment method, device and medium

Abstract

The present invention belongs to the field of electric power technology, and discloses a method, device and medium for evaluating the reliability of a wind turbine. The method comprises: establishing a degradation-impact dependent competitive failure model of each component of the wind turbine considering self-healing according to a nonlinear Wiener process and the Markov principle; establishing a multi-fault failure model of each component of the wind turbine based on a multidimensional Copula function; selecting a Copula function and establishing a correlation model of failures between different components of the wind turbine; and performing parameter estimation on the degradation amount and the Copula function based on the maximum likelihood method and the Monte Carlo method. The present invention can obtain the reliability curves of each component and system of the wind turbine, and then evaluate the reliability of the wind turbine.

Description

Technical Field

[0001] The present invention relates to the field of electric power technology, and more specifically, to a wind turbine reliability assessment method, device and medium. Background Art

[0002] As one of the most promising power generation methods in the field of new energy, wind power generation has a complex structure and a high degree of coupling. At the same time, the working environment of wind turbines during service is complex and changeable, and they are often subjected to multiple stress loads, which leads to the complexity and concealment of their failure modes. The failure form is often the result of competition between multiple failure modes. At present, the failure of multiple failure modes competition is often in extreme weather, and the recovery characteristics of each component of the wind turbine after extreme impact are not considered, resulting in inaccurate evaluation of the reliability of wind turbines. Summary of the invention

[0003] The present invention provides a reliability assessment method, device and medium for a wind turbine. Considering that the degradation process of each component of a wind turbine is affected by multiple factors, is random, and is not linearly distributed, a nonlinear Wiener process and Markov principle are used to establish the natural degradation and random impact process of the wind turbine, and considering the recovery characteristics of each component of the wind turbine after being impacted, a self-healing model of each component of the wind turbine is established; in addition, in order to solve the problem of correlation between failures of various components of the wind turbine, a multi-dimensional Archimedean Copula function model is established to obtain the correlation between different components; the maximum likelihood number and Monte Carlo method are used to estimate the parameters of the degradation amount and the Copula function model to obtain the reliability curves of the wind turbine system and each component.

[0004] The present invention adopts the following technical solution:

[0005] According to a first aspect of the present invention, a method for evaluating the reliability of a wind turbine is provided, the method comprising:

[0006] Based on the nonlinear Wiener process and Markov principle, a degradation-impact dependent competitive failure model of each component of the wind turbine considering self-healing is established;

[0007] Based on the multi-dimensional Copula function, a multi-fault failure model of each component of the wind turbine is established;

[0008] Selection of Copula function and establishment of correlation model of failure between different components of wind turbine;

[0009] The degradation amount and Copula function parameters are estimated based on the maximum likelihood method and Monte Carlo method.

[0010] As a preferred technical solution, the degradation-impact dependent competitive failure model of each component of the wind turbine generator set considering self-healing includes a natural degradation model of the wind turbine generator set, a random impact process model and a self-healing model of each component of the wind turbine generator set. The degradation-impact dependent competitive failure model of each component of the wind turbine generator set considering self-healing is established according to the nonlinear Wiener process and the Markov principle, including:

[0011] A natural degradation model of wind turbines is established based on the nonlinear Wiener process;

[0012] Establish a random shock process model based on the Markov principle;

[0013] The self-healing model of each component of the wind turbine is established based on the natural degradation model of the wind turbine and the random impact process model.

[0014] As a preferred technical solution, a natural degradation model of wind turbines is established based on the nonlinear Wiener process, including:

[0015] The degradation model of nonlinear Wiener process is established as the natural degradation model of wind turbines, which is expressed as:

[0016] ;

[0017] in: represent The degradation amount of the moment degradation process; is the drift coefficient, indicating the degradation rate; is a non-decreasing time scale function, indicating the nonlinear characteristics of the degradation process. It represents the change of scale per unit time. is the drift function, is the time scale parameter, is the diffusion coefficient; is the standard Brownian motion;

[0018] The closed expression of the probability density function of the natural degradation model of the wind turbine is transformed into the probability problem of the first penetration distribution of the standard Brownian motion exceeding the time-varying threshold, and the approximate analytical expression of the natural degradation failure is obtained as follows:

[0019] ;

[0020] in: represents the natural degradation failure function, represents the time-varying threshold function, represents a given critical degradation failure threshold, represents the exponential function with e as base;

[0021] The approximate calculation formula of the cumulative distribution function of the first penetration time distribution is:

[0022] ;

[0023] in: express The cumulative distribution function of the distribution when the threshold is first crossed, represents the probability function, represents the moment when the threshold is reached, is a composite function, representing the natural degradation failure function, represents the cumulative distribution function.

[0024] As a preferred technical solution, a random shock process model is established based on the Markov principle, including:

[0025] For performance parameters, assuming that the shock occurrence and natural degradation satisfy the 0-1 state Markov chain, Indicates the status, expressed as:

[0026] ;

[0027] The state transition matrix of the Markov chain is used to describe the state transition process of the performance parameters to show the probability of mutual conversion between different states. If the current state is a shock, the probability of the shock occurring at the next moment is , the probability of returning to natural degradation is ,and ; If the current state is natural degradation, the probability of maintaining natural degradation in the next moment is The probability of a shock occurring is ,and ;

[0028] Based on natural degradation and impact effects, The degradation of performance parameters It is expressed as:

[0029] ;

[0030] in: To represent p The performance parameters are t The degradation amount of the natural degradation process at each moment, for t The additional degradation caused by the impact at time t, the impact degradation, Subject to independent and identical expectations , the variance is The normal distribution of .

[0031] As a preferred technical solution, a self-healing model of each component of a wind turbine is established based on the natural degradation model of the wind turbine and the random impact process model, including:

[0032] When the system is subjected to After the first impact, self-healing occurs, and the time is determined by the following formula t The actual amount of damage after the system self-healed :

[0033] ;

[0034] in: For the Secondary shock duration; is the self-healing influence function; For the The amount of degradation of the secondary impact;

[0035] The time is determined by the following formula t The cumulative damage of the system :

[0036] ;

[0037] in: For the Secondary shock duration; For the The degradation amount of each impact, is the number of impact arrivals.

[0038] As a preferred technical solution, a multi-fault failure model of each component of a wind turbine is established based on a multi-dimensional Copula function, including:

[0039] Determine a Copula function based on the joint distribution function and marginal distribution function of multidimensional random variables Satisfies the following relationship:

[0040] ;

[0041] in: is the joint distribution function, is the marginal distribution function, For wind turbines n component degradation; if the marginal distribution function is continuous, then the Copula function is a definite function; if the marginal distribution function is a univariate distribution, F (⋅,⋯,⋅) is a marginal distribution F 1(⋅) , F 2(⋅) ,⋯, F nThe joint distribution function of ( ⋅ );

[0042] If the wind turbine Components exist h There are multiple coupling relationships between the failure modes, among which the failure modes with multiple coupling relationships are k Then the failure modes with independent relationships are h - k The corresponding joint distribution function is expressed as:

[0043] ;

[0044] in: is the corresponding component life, is a joint distribution function with multivariate coupling relationship, For failure modes with multi-coupling relationships, the component life is less than the rated life. Rating life of components with failure modes with multi-coupling relationships;

[0045] Establish k-dimensional Copula function , expressed as

[0046] ;

[0047] in: For all k The joint distribution function of the component rated life of the multivariate coupling failure mode is For the i Has k The component life distribution function of the multivariate coupling failure mode, is the cumulative distribution function;

[0048] At the initial moment, the component is not in a faulty state, so The reliability of each component is:

[0049] ;

[0050] in, For the The reliability of each component, is the probability occurrence function, For the Has k Component life of a multi-coupling failure mode, For the The life of each component;

[0051] The failure mode with multivariate coupling relationship is expressed as Copula form, and the The reliability of each component is:

[0052] ;

[0053] in: For the The corresponding component The cumulative distribution function value of the failure mode, For the Components corresponding to and The joint distribution function value of the two failure modes; is the reliability of the failure mode without coupling relationship, is the joint distribution function of the failure modes with k multivariate coupling relationships in the Copula form, Mathematical operation symbol, expressed as the product of the reliabilities of failure modes that do not have a coupling relationship.

[0054] As a preferred technical solution, the selection of Copula function and the establishment of a correlation model of failures between different components of a wind turbine generator set include:

[0055] Establish the first Copula function, the second Copula function and the third Copula function;

[0056] The first Copula function, the second Copula function and the third Copula function are used to analyze the multi-fault failure model of each component of the wind turbine to establish a correlation model of failures between different components of the wind turbine;

[0057] The distribution function expression and probability density expression of the first Copula function are:

[0058] ;

[0059] in: , For different faults; , is the first parameter; is the Archimedean Copula distribution function; is the probability density function, is the distribution function expression of the first Copula function;

[0060] The distribution function expression and probability density expression of the second Copula function are:

[0061] ;

[0062] in: is the second parameter, , ; is the distribution function expression of the second Copula function, is the probability density expression of the second Copula function;

[0063] The distribution function expression and probability density expression of the third Copula function are respectively:

[0064] ;

[0065] in: is the third parameter, ; is the distribution function expression of the third Copula function, is the probability density expression of the third Copula function;

[0066] The failure correlation model between different components of the wind turbine generator set includes one of a series model, a parallel model and a voting model, and a combination thereof;

[0067] In the described series model, failure of any one component will lead to failure of the entire system;

[0068] In the parallel model described, the system fails when all components that make up the system fail;

[0069] In the voting model, the n There are at least r The system works normally when there are no faults in the components; r ≤ n .

[0070] As a preferred technical solution, the degradation amount and the Copula function are estimated based on the maximum likelihood method and the Monte Carlo method, including:

[0071] A maximum likelihood degradation model is constructed, and the maximum likelihood degradation model is used to perform parameter estimation on the degradation amount; wherein the maximum likelihood degradation model is expressed as:

[0072] ;

[0073] in: is the maximum likelihood degradation function, For the The scale parameter of the component degradation, is the degradation process of the ith component, n is the number of components, is the number of measurements, For the In the degradation process The first fault The degradation amount measured is For the The total number of measurements for each fault, is the distribution function of the standard normal variable, For the h The first fault W The time taken for each measurement, For the i The first component h The first fault W The variance of the time taken for each measurement, For the i The first component h The first fault W The expected time for each measurement;

[0074] Parameter estimation of Copula function based on Monte Carlo method and semi-parametric estimation method:

[0075] The parameter value is solved by using the semi-parametric estimation method, and the kernel density distribution function is Replace the marginal distribution function:

[0076] ;

[0077] in: For the The kernel density distribution function of the components; is a semi-parametric estimate.

[0078] According to a second aspect of the present invention, a wind turbine reliability assessment device is provided, the device comprising:

[0079] The first model building unit is configured to establish a degradation-impact dependent competitive failure model of each component of the wind turbine considering self-healing according to a nonlinear Wiener process and a Markov principle;

[0080] The second model building unit is configured to build a multi-fault failure model of each component of the wind turbine generator based on a multi-dimensional Copula function;

[0081] The third model building unit is configured to select the Copula function and establish a correlation model of failures between different components of the wind turbine;

[0082] The parameter estimation unit is configured to perform parameter estimation on the degradation amount and the Copula function based on the maximum likelihood method and the Monte Carlo method.

[0083] According to a third aspect of the present invention, there is provided a readable storage medium storing one or more programs, wherein the one or more programs can be executed by one or more processors to implement the method as described above.

[0084] The present invention has at least the following beneficial effects:

[0085] The present invention can better simulate the degradation process caused by multiple factors such as the wind turbine environment by adopting the nonlinear Wiener method, and has more advantages in dealing with random problems through the Markov principle. In addition, each component of the wind turbine will adopt a self-protection mechanism when impacted, and self-recover from the damage caused by the impact. By establishing a self-healing model, the reliability of the wind turbine can be more accurately evaluated. Secondly, the Archimedean Copula function family has good applicability in solving nonlinear relationships, has a simple structure and can accurately describe the correlation between fault modes and between components. Finally, since the Monte Carlo method and the semi-parametric estimation method have data simulation and iterative processes, the speed of estimating unknown parameters can be further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 A flow chart of a wind turbine reliability assessment method according to an embodiment of the present invention is shown;

[0087] Figure 2 A structural diagram of a wind turbine reliability assessment device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0088] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and specific embodiments, but are not intended to limit the present invention. For the various steps described herein, if there is no necessity for a causal relationship between each other, the order in which they are described as examples herein should not be regarded as a limitation, and those skilled in the art should know that they can be adjusted in order, as long as the logic between them is not destroyed, resulting in the inability to implement the entire process.

[0089] The embodiment of the present invention provides a method for evaluating the reliability of a wind turbine generator system. Figure 1 As shown in FIG. 1 , it is a flow chart of the method, and the wind turbine reliability assessment method comprises the following steps:

[0090] Step 1: Based on the nonlinear Wiener process and Markov principle, a degradation-impact dependent competitive failure model considering self-healing of each component of the wind turbine is established.

[0091] In this embodiment, the degradation-impact dependent competitive failure model of each wind turbine component taking into account self-healing includes a natural degradation model of the wind turbine, a random impact process model and a self-healing model of each wind turbine component.

[0092] In some embodiments, step 1 is specifically implemented by the following steps:

[0093] Step 11: Establish a natural degradation model of wind turbines based on the nonlinear Wiener process.

[0094] Currently, most competitive failure studies use linear regression models X ( t )= a + bt To describe the degradation process of the system. However, the linear regression model cannot describe the volatility and nonlinear characteristics in the degradation process of the system. This embodiment uses a nonlinear Wiener process to model the degradation process to more accurately describe the degradation behavior of the system.

[0095] Degradation model based on nonlinear Wiener process N 0 is:

[0096] ;

[0097] in: represent The degradation amount of the moment degradation process; is the drift coefficient, indicating the degradation rate; is a non-decreasing time scale function, indicating the nonlinear characteristics of the degradation process. It represents the change of scale per unit time. is the drift function, is the time scale parameter, is the diffusion coefficient; This is a standard Brownian motion.

[0098] When the degradation The specified critical degradation failure threshold is exceeded for the first time k , then it is judged to be invalid, and the time at this time t It is called the first wear-out time and can be used as the life of the system when degradation failure occurs. N 0 , the closed expression of the probability density function of the nonlinear degradation failure model is transformed into the probability problem of the first penetration distribution of the standard Brownian motion exceeding the time-varying threshold, and the approximate analytical expression of natural degradation failure is obtained as follows:

[0099] ;

[0100] in: represents the natural degradation failure function, represents the time-varying threshold function, represents a given critical degradation failure threshold, Represents an exponential function with base e.

[0101] but The approximate calculation formula of the cumulative distribution function of the first penetration time distribution is:

[0102] ;

[0103] in: express The cumulative distribution function of the distribution when the threshold is first crossed, represents the probability function, represents the moment when the threshold is reached, is a composite function, representing the natural degradation failure function, represents the cumulative distribution function.

[0104] Step 12: Establish a random shock process model based on the Markov principle.

[0105] Due to the influence of factors such as material properties and working characteristics, the impact effect of performance parameters is recoverable, that is, for a certain performance parameter, its value returns to the natural degradation process after an impact change occurs.

[0106] For performance parameters, assuming that the shock occurrence and natural degradation satisfy the 0-1 state Markov chain, Indicates the status, expressed as:

[0107] ;

[0108] The state transition matrix of the Markov chain is used to describe the state transition process of the performance parameters, and then the probability of mutual conversion between different states is shown. If the current state is a shock, the probability of the shock still occurring at the next moment is , the probability of returning to natural degradation is ,and ; If the current state is natural degradation, the probability of maintaining natural degradation in the next moment is The probability of a shock occurring is ,and , expressed as:

[0109] ;

[0110] Taking into account the natural degradation and impact effects, the final The degradation of performance parameters It is expressed as:

[0111] ;

[0112] in: To represent p The performance parameters are t The degradation amount of the natural degradation process at each moment, for t The additional degradation caused by the impact at time t, the impact degradation, Subject to independent and identical expectations , the variance is The normal distribution of .

[0113] Step 13: Establish a self-healing model for each component of the wind turbine generator set based on the natural degradation model of the wind turbine generator set and the random impact process model.

[0114] When the system is subjected to After the first impact, self-healing occurs, and the time is determined by the following formula t The actual amount of damage after the system self-healed :

[0115] ;

[0116] in: For the Secondary shock duration; is the self-healing influence function; For the The amount of degradation of the secondary impact;

[0117] The time is determined by the following formula t The cumulative damage of the system :

[0118] ;

[0119] in: For the Secondary shock duration; For the The degradation amount of each impact, is the number of impact arrivals.

[0120] Due to the non-monotonicity of the degradation process of the system with self-healing mechanism, it is impossible to directly give For a dependent competitive failure system with a self-healing mechanism, in order to prevent the system from failing, it is necessary to ensure that the size of each impact is smaller than the failure threshold. , and the total degradation of the system is less than .

[0121] Step 2: Based on the multi-dimensional Copula function, a multi-fault failure model of each component of the wind turbine is established.

[0122] N The Copula function has a simple structure and can accurately describe the correlation between components and between components and failure modes. N The Copula function has the following properties: its domain is [0,1]; N The zero reference plane of increasing dimension; for any variable , satisfying the marginal distribution .

[0123] According to the multidimensional Sklar theorem, assuming that the joint distribution function of multidimensional random variables is , the marginal distribution function is , then there exists a Copula function Satisfies the relationship:

[0124] ;

[0125] in: is the joint distribution function, is the marginal distribution function, For wind turbines n The degradation of components, if the marginal distribution function is continuous, then the Copula function is a definite function; if the marginal distribution function is a univariate distribution, F (⋅,⋯,⋅) is a marginal distribution F 1(⋅) , F 2(⋅) ,⋯, F n The joint distribution function of ( ⋅ )

[0126] Assume that the first i Components exist h There are (h≥2) kinds of failure modes, and there are multi-coupling relationships between the failure modes. It means that the failure modes with multi-coupling relationships are k Then the fault types with independent relationships are hk , use ; indicates. Failure modes of components . Then the corresponding joint distribution function can be expressed as:

[0127] ;

[0128] in: is the actual life of the component with the i-th failure mode with multivariate coupling relationship, is the joint distribution function of component life with multivariate coupling relationship, For failure modes with multi-coupling relationships, the component life is less than the rated life. Rating life of components with failure modes with multi-coupling relationships;

[0129] Establish k-dimensional Copula function , expressed as

[0130] ;

[0131] in: For all k The joint distribution function of the component rated life of the multivariate coupling failure mode is For the i Has k The component life distribution function of the multivariate coupling failure mode, is the expression of a composite function, and its value is equal to ;

[0132] At the initial moment, the component is in a non-faulty state. The reliability of each component is:

[0133] ;

[0134] in, For the The reliability of each component, is the probability occurrence function, For the Has k Component life of a multi-coupling failure mode, For the The life of each component;

[0135] According to the disjoint property and multidimensional Sklar theorem, the failure mode with multi-dimensional coupling relationship is expressed as Copula form, and the first The reliability of each component is:

[0136] ;

[0137] in: For the The corresponding component The cumulative distribution function value of the failure mode, For the Components corresponding to and The joint distribution function value of the two failure modes; is the reliability of the failure mode without coupling relationship, is the joint distribution function of the failure modes with k multivariate coupling relationships in the Copula form, Mathematical operation symbol, expressed as the product of the reliabilities of failure modes that do not have a coupling relationship.

[0138] Step 3: Select the Copula function and establish the failure correlation model between different components of the wind turbine.

[0139] By using different Copula functions, different dependency structures can be described, thereby more accurately characterizing the joint distribution of multidimensional random variables. Compared with the elliptic Copula function, the advantage of the Archimedean Copula function is that it is not restricted by radial symmetry and can accurately capture the dependencies between different upper and lower tails.

[0140] The Archimedean Copula distribution function expression is:

[0141] ;

[0142] in: , and denote the Gumbel Copula function, the Clayton Copula function, and the Frank Copula function, respectively. a, β, and are the corresponding parameters, , ,and is the weighting coefficient, and .

[0143] Derivative the above formula to obtain the probability density function of the Copula function:

[0144] ;

[0145] in: They are The derivative function of .

[0146] In this embodiment, three Copula functions are set, namely the first Copula function, the second Copula function and the third Copula function, which are specifically described as follows:

[0147] (1) The first Copula function is a binary Gumbel Copula function, and the distribution function expression and probability density expression are:

[0148] ;

[0149] in: , For different faults; , is the first parameter; is the Archimedean Copula distribution function; is the probability density function, is the distribution function expression of the first Copula function.

[0150] (2) The second Copula function is a binary Clayton Copula function, and the distribution function expression and probability density expression are:

[0151] ;

[0152] in: is the second parameter, , ; is the distribution function expression of the second Copula function, is the probability density expression of the second Copula function;

[0153] (3) The third Copula function is a binary Frank Copula function, and the distribution function expression and probability density expression are:

[0154] ;

[0155] in: is the third parameter, ; is the distribution function expression of the third Copula function, is the probability density expression of the third Copula function;

[0156] The model constructed in step 2 is analyzed based on the Copula function to establish a correlation model. The typical reliability models between different components mainly include series model, parallel model and voting model:

[0157] In the series model, if any component fails, the entire system will fail. i The system is composed of components, each of which is independent of each other, that is, the failure of a component will not affect the reliability of other components. The system reliability expression is

[0158] ;

[0159] in: is the system reliability, n is the number of components.

[0160] A parallel system will fail only when all components of the system fail. i If independent components are connected in parallel, the system reliability expression is:

[0161] ;

[0162] in: is the system reliability, n is the number of components.

[0163] In the voting model, the n Of the components, at least r (1≤ r ≤ n ) components are fault-free for the system to work properly. The system reliability expression is

[0164] ;

[0165] in: is the system reliability, n is the number of components.

[0166] Step 4: Estimate the parameters of the degradation amount and Copula function based on the maximum likelihood method and Monte Carlo method.

[0167] In one embodiment, step 4 is specifically implemented by the following steps:

[0168] Step 41: Parameter (degradation amount) estimation is performed using the maximum likelihood degradation model.

[0169] The maximum likelihood degradation model is expressed as:

[0170] ;

[0171] in: is the maximum likelihood degradation function, For the The scale parameter of the component degradation, is the degradation process of the ith component, n is the number of components, is the number of measurements, For the In the degradation process The first fault The degradation amount measured is For the The total number of measurements for each fault, is the distribution function of the standard normal variable, For the h The first fault WThe time taken for each measurement, For the i The first component h The first fault W The variance of the time taken for each measurement, For the i The first component h The first fault W The expected time for each measurement;

[0172] Step 42: Estimate the parameters of the Copula function based on the Monte Carlo method and the semi-parametric estimation method.

[0173] Since the sample likelihood function involves multi-dimensional integration and differentiation, and the integral function itself is relatively complex, the Markov chain Monte Carlo (MCMC) method has data simulation and iteration processes. Therefore, the MCMC method is used to estimate the unknown parameters. The semi-parametric estimation method is used to solve the parameter value and the kernel density distribution function is Replace the marginal distribution function:

[0174] ;

[0175] in: For the The kernel density distribution function of the components; is a semi-parametric estimate.

[0176] In summary, based on the above content, the reliability curves of each component and system of the wind turbine can be obtained, and then the reliability of the wind turbine can be evaluated.

[0177] The embodiment of the present invention also provides a wind turbine reliability assessment device, such as Figure 2 As shown, the device comprises:

[0178] The first model building unit 401 is configured to establish a degradation-impact dependent competitive failure model of each component of the wind turbine generator considering self-healing according to a nonlinear Wiener process and a Markov principle;

[0179] The second model building unit 402 is configured to build a multi-fault failure model of each component of the wind turbine generator based on a multi-dimensional Copula function;

[0180] The third model building unit 403 is configured to select a Copula function and establish a correlation model of failures between different components of the wind turbine generator set;

[0181] The parameter estimation unit 404 is configured to perform parameter estimation on the degradation amount and the Copula function based on the maximum likelihood method and the Monte Carlo method.

[0182] It should be noted that the various device structures described in this embodiment belong to the same technical concept as the previously described method, and achieve the same technical effect through the same principle, which will not be repeated here.

[0183] An embodiment of the present invention further provides a readable storage medium, which stores one or more programs. The one or more programs can be executed by one or more processors to implement the method described in any of the above embodiments.

[0184] In addition, although exemplary embodiments have been described herein, the scope includes any and all embodiments based on the present invention with equivalent elements, modifications, omissions, combinations (e.g., various embodiments intersecting schemes), adaptations or changes. The elements in the claims are to be interpreted broadly based on the language adopted in the claims, and are not limited to the examples described in this specification or during the prosecution of this application, which examples are to be interpreted as non-exclusive. Therefore, this specification and examples are intended to be considered as examples only, and the true scope and spirit are indicated by the following claims and the full scope of their equivalents.

[0185] The above description is intended to be illustrative rather than restrictive. For example, the above examples (or one or more of them) may be used in combination with each other. For example, a person of ordinary skill in the art may use other embodiments when reading the above description. In addition, in the above-mentioned specific embodiments, various features may be grouped together to simplify the present invention. This should not be interpreted as an intention that a feature of an invention that is not claimed for protection is necessary for any claim. On the contrary, the subject matter of the present invention may be less than all the features of an embodiment of a particular invention. Thus, the following claims are incorporated herein as examples or embodiments in the specific embodiments, wherein each claim is independently a separate embodiment, and it is considered that these embodiments may be combined with each other in various combinations or arrangements. The scope of the present invention should be determined with reference to the appended claims and the full scope of equivalent forms granted by these claims.

Claims

1. A wind turbine reliability assessment method, characterized in that: The method comprises: Based on the nonlinear Wiener process and Markov principle, a degradation-impact dependent competitive failure model of each component of the wind turbine considering self-healing is established; Based on the multi-dimensional Copula function, a multi-fault failure model of each component of the wind turbine is established; Selection of Copula function and establishment of correlation model of failure between different components of wind turbine; The degradation amount and Copula function parameters are estimated based on the maximum likelihood method and Monte Carlo method; The degradation-impact dependent competitive failure model of each wind turbine component considering self-healing includes a natural degradation model of the wind turbine, a random impact process model and a self-healing model of each wind turbine component. The degradation-impact dependent competitive failure model of each wind turbine component considering self-healing is established according to the nonlinear Wiener process and the Markov principle, including: A natural degradation model of wind turbines is established based on the nonlinear Wiener process; Establish a random shock process model based on the Markov principle; The self-healing model of each component of the wind turbine is established based on the natural degradation model of the wind turbine and the random impact process model; The natural degradation model of wind turbines is established based on the nonlinear Wiener process, including: The degradation model of nonlinear Wiener process is established as the natural degradation model of wind turbines, which is expressed as: ； in: represent The degradation amount of the moment degradation process; is the drift coefficient, indicating the degradation rate; is a non-decreasing time scale function, indicating the nonlinear characteristics of the degradation process. It represents the change of scale per unit time. is the drift function, is the time scale parameter, is the diffusion coefficient; is the standard Brownian motion; The closed expression of the probability density function of the natural degradation model of the wind turbine is transformed into the probability problem of the first penetration distribution of the standard Brownian motion exceeding the time-varying threshold, and the approximate analytical expression of the natural degradation failure is obtained as follows: ； in: ; represents the natural degradation failure function, represents the time-varying threshold function, represents a given critical degradation failure threshold, represents the exponential function with e as base; The approximate calculation formula of the cumulative distribution function of the first penetration time distribution is: ； in: express The cumulative distribution function of the distribution when the threshold is first crossed, represents the probability function, represents the moment when the threshold is reached, is a composite function, representing the natural degradation failure function, represents the cumulative distribution function; A random shock process model is established based on the Markov principle, including: For performance parameters, assuming that the shock occurrence and natural degradation satisfy the 0-1 state Markov chain, Indicates the status, expressed as: ； The state transition matrix of the Markov chain is used to describe the state transition process of the performance parameters to show the probability of mutual conversion between different states. If the current state is a shock, the probability of the shock occurring at the next moment is , the probability of returning to natural degradation is ,and ; If the current state is natural degradation, the probability of maintaining natural degradation in the next moment is The probability of a shock occurring is ,and ; Based on natural degradation and impact effects, The degradation of performance parameters It is expressed as: ； in: To represent p The performance parameters are t The degradation amount of the natural degradation process at each moment, for t The amount of impact degradation caused by the impact at time, Subject to independent and identical expectations , the variance is Normal distribution of Based on the natural degradation model of wind turbines and the random impact process model, the self-healing model of each component of the wind turbine is established, including: When the system is subjected to After the first impact, self-healing occurs, and the time is determined by the following formula t The actual amount of damage after the system self-healed : ； in: For the Secondary shock duration; is the self-healing influence function; For the The amount of degradation of the secondary impact; The time is determined by the following formula t The cumulative damage of the system : ； in: For the Secondary shock duration; For the The degradation amount of each impact, is the number of arrivals of the shock; The degradation amount and Copula function parameters are estimated based on the maximum likelihood method and Monte Carlo method, including: A maximum likelihood degradation model is constructed, and the maximum likelihood degradation model is used to perform parameter estimation on the degradation amount; wherein the maximum likelihood degradation model is expressed as: ； in: is the maximum likelihood degradation function, For the The scale parameter of the component degradation, is the degradation process of the ith component, n is the number of components, is the number of measurements, For the In the degradation process The first fault The degradation amount measured is For the The total number of measurements for each fault, is the distribution function of the standard normal variable, For the h The first fault W The time taken for each measurement, For the i The first component h The first fault W The variance of the time taken for each measurement, For the i The first component h The first fault W The expected time for each measurement; Parameter estimation of Copula function based on Monte Carlo method and semi-parametric estimation method: The parameter value is solved by using the semi-parametric estimation method, and the kernel density distribution function is Replace the marginal distribution function: ； in: For the The kernel density distribution function of the components; is a semi-parametric estimate.

2. The method according to claim 1, characterized in that Based on the multi-dimensional Copula function, a multi-fault failure model of each component of the wind turbine is established, including: Determine a Copula function based on the joint distribution function and marginal distribution function of multidimensional random variables Satisfies the following relationship: ； in: is the joint distribution function, is the marginal distribution function, For wind turbines n component degradation; if the marginal distribution function is continuous, then the Copula function is a definite function; if the marginal distribution function is a univariate distribution, Is a marginal distribution The joint distribution function of If the wind turbine Components exist h There are multiple coupling relationships between the failure modes, among which the failure modes with multiple coupling relationships are k Then the failure modes with independent relationships are h - k The corresponding joint distribution function is expressed as: ； in: is the actual life of the component with the i-th failure mode with multivariate coupling relationship, is the joint distribution function of component life with multivariate coupling relationship, is the probability that the component life of the failure mode with multi-element coupling relationship is less than the rated life, Rating life of components with failure modes with multi-coupling relationships; Establish k-dimensional Copula function , expressed as ； in: For all k The joint distribution function of the component rated life of the multivariate coupling failure mode is For the i Has k The component life distribution function of the multivariate coupling failure mode, is the cumulative distribution function; At the initial moment, the component is not in a faulty state, so The reliability of each component is: ； in, For the The reliability of each component, is the probability occurrence function, For the Has k Component life of a multi-coupling failure mode, For the The life of each component; The failure mode with multivariate coupling relationship is expressed as Copula form, and the The reliability of each component is: ； in: For the The corresponding component The cumulative distribution function value of the failure mode, For the Components corresponding to and The joint distribution function value of the two failure modes; is the reliability of the failure mode without coupling relationship, is the joint distribution function of the failure modes with k multivariate coupling relationships in the Copula form, Mathematical operation symbol, expressed as the product of the reliabilities of failure modes that do not have a coupling relationship.

3. The method according to claim 2, characterized in that The selection of Copula function and the establishment of the failure correlation model between different components of the wind turbine include: Establish the first Copula function, the second Copula function and the third Copula function; The first Copula function, the second Copula function and the third Copula function are used to analyze the multi-fault failure model of each component of the wind turbine to establish a correlation model of failures between different components of the wind turbine; The distribution function expression and probability density expression of the first Copula function are: ； in: , For different faults; , is the first parameter; is the Archimedean Copula distribution function; is the probability density function, is the distribution function expression of the first Copula function; The distribution function expression and probability density expression of the second Copula function are: ； in: is the second parameter, , ; is the distribution function expression of the second Copula function, is the probability density expression of the second Copula function; The distribution function expression and probability density expression of the third Copula function are respectively: ； in: is the third parameter, ; is the distribution function expression of the third Copula function, is the probability density expression of the third Copula function; The failure correlation model between different components of the wind turbine generator set includes one of a series model, a parallel model and a voting model, and a combination thereof; In the described series model, failure of any one component will lead to failure of the entire system; In the parallel model described, the system fails when all components that make up the system fail; In the voting model, the n There are at least r The system works normally when there are no faults in the components; r ≤ n .

4. A wind turbine reliability assessment device, used to implement the method according to claim 1, characterized in that: The device comprises: The first model building unit is configured to establish a degradation-impact dependent competitive failure model of each component of the wind turbine considering self-healing according to a nonlinear Wiener process and a Markov principle; The second model building unit is configured to build a multi-fault failure model of each component of the wind turbine generator based on a multi-dimensional Copula function; The third model building unit is configured to select the Copula function and establish a correlation model of failures between different components of the wind turbine; The parameter estimation unit is configured to perform parameter estimation on the degradation amount and the Copula function based on the maximum likelihood method and the Monte Carlo method. 5 . A non-transitory computer-readable storage medium storing instructions, which, when executed by a processor, executes the method according to claim 1 .

Images