A method for predicting the lifetime of photovoltaic modules considering individual degradation differences

Abstract

This invention relates to the field of photovoltaic power generation and discloses a method for predicting the lifetime of photovoltaic modules considering individual degradation differences. The method includes: constructing a degradation model based on an inverse Gaussian-gamma mixture distribution for the power degradation increment; introducing a random diffusion parameter variable following a gamma distribution into the inverse Gaussian distribution of the power degradation increment to represent the individual degradation differences of the photovoltaic modules; quantifying and solving the model parameters of the degradation model based on measured data of the photovoltaic modules under test to confirm the optimal model parameters; generating a corresponding module degradation trajectory based on the degradation model under the optimal model parameters and the randomly generated diffusion parameters, and confirming the module lifetime based on a preset failure threshold; generating a lifetime distribution by repeatedly sampling the diffusion parameters. This invention considers the coupled randomness of environmental factors and individual differences during photovoltaic module degradation, improving prediction accuracy.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation technology, and more specifically to a method for predicting the lifespan of photovoltaic modules that takes into account individual degradation differences. Background Technology

[0002] As the core component of a photovoltaic (PV) power generation system, photovoltaic (PV) modules bear the crucial mission of efficiently converting solar energy into electrical energy, serving as the system's "energy converter." Their performance directly determines the system's power generation efficiency and economic benefits, and is a core support for ensuring the stable operation of the PV system. Module lifespan is affected by multiple factors: in the natural environment, high temperature, high humidity, and ultraviolet radiation accelerate the aging process of encapsulation and semiconductor materials; while installation defects, wind and sand erosion, and other problems can lead to physical damage to components, further shortening their service life. Conducting lifespan prediction research is of great significance. By accurately predicting module failure times, maintenance and replacement plans can be planned in advance, playing a crucial supporting role in promoting the sustainable development of the PV industry.

[0003] Currently, significant limitations remain in the field of photovoltaic module lifespan prediction: existing models often ignore the differences between individual modules. Although the overall trend and average rate of module power degradation are roughly similar, the actual power degradation can vary significantly between different modules due to differences in materials, processes, or installation details. Furthermore, as service life increases, environmental factors such as sunlight intensity, ambient temperature, and humidity continuously affect the modules, further leading to a gradual decrease in their output power. If individual differences and the coupled effects of environmental factors and power degradation are not comprehensively considered in the prediction, it is highly likely that the predicted results will deviate significantly from the actual individual lifespan, making it difficult to accurately support maintenance and replacement decisions for individual modules.

[0004] Therefore, how to accurately predict the lifespan of photovoltaic modules by comprehensively considering individual differences and environmental factors is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of the above problems, the present invention is proposed to provide a photovoltaic module lifetime prediction method that takes into account individual degradation differences in order to overcome or at least partially solve the above problems.

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

[0007] A method for predicting the lifetime of photovoltaic modules that takes into account individual degradation differences includes the following steps:

[0008] S1: Based on the inverse Gaussian-gamma mixture distribution, a degradation model is constructed for the power degradation increment. A random diffusion parameter variable following a gamma distribution is introduced into the inverse Gaussian distribution of the power degradation increment to represent the individual degradation differences of photovoltaic modules.

[0009]

[0010] in, It is time Time The power degradation increment, where i is the time step variable. To introduce a comprehensive environmental degradation function with multiple environmental covariates, the function value represents the average degradation rate. It is time Time The change in cumulative degradation intensity; For diffusion parameters, and For different diffusion distribution parameters, Let Gamma be the shape parameter of the distribution. For the rate parameter of the Gamma distribution;

[0011] S2: Based on the measured data of the photovoltaic module under test, the model parameters of the degradation model are quantitatively solved to confirm the optimal model parameters; wherein, the optimal model parameters include the optimal environmental covariate parameters and the optimal diffusion distribution parameters; the optimal environmental covariate parameters and the optimal diffusion distribution parameters are generated simultaneously;

[0012] S3: Based on the degradation model under the optimal model parameters, generate the corresponding component degradation trajectory by combining the randomly generated diffusion parameters, and confirm the component lifetime according to the preset failure threshold; generate the lifetime distribution by repeatedly sampling the diffusion parameters.

[0013] Preferably, in step S1, when constructing the degradation model, multiple environmental covariates are introduced to optimize the average degradation rate, resulting in a comprehensive environmental degradation function:

[0014]

[0015] in, Relative humidity, Irradiance, B and C are the coefficient parameters to be estimated, i.e., environmental covariate parameters.

[0016] Preferably, in step S1, the expression for the cumulative intensity is:

[0017]

[0018] in, To control the degradation rate, t is the degradation time.

[0019] Preferably, in step S2, the maximum likelihood method is used to quantify and solve the model parameters of the degenerate model to obtain the parameters that maximize the likelihood function, which are then used as the optimal model parameters.

[0020] Preferably, the likelihood function is:

[0021]

[0022] in, This is the vector of model parameters that need to be estimated, where n is the number of time steps and Data represents the measured data.

[0023] Preferably, in step S3, the step of generating the degradation trajectory includes:

[0024] The degradation curve is obtained by iterating through each time step and sampling according to the marginal probability density function of the degradation model at each time step.

[0025] Preferably, during the generation of the degradation trajectory, the diffusion parameter follows a gamma distribution based on the optimal model parameters, and the diffusion parameter is randomly generated within the corresponding gamma distribution.

[0026] Preferably, the marginal probability density function is:

[0027]

[0028] Among them, among them, To be in the time interval Average environmental covariate values ​​within, To be in the time interval The observed power degradation increment, The average degradation rate is determined by environmental covariates. The increment of the time effect, i.e., time Time The change in cumulative degradation intensity Let be the shape parameter of the Gamma distribution, and γ be the rate parameter of the Gamma distribution. This is the Gamma function.

[0029] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a photovoltaic module lifetime prediction method that considers individual degradation differences. It introduces an inverse Gaussian-gamma mixture model to achieve comprehensive consideration of individual degradation differences and environmental covariates. Subsequently, parameter estimation methods are used to simultaneously estimate individual randomness parameters and environmental randomness-related parameters. The synchronization of the two enables the final parameter estimation results to take into account the coupling effect of individual differences and environmental factors, thereby obtaining a more accurate model expression and improving the accuracy of lifetime prediction. In addition, the present invention, based on an incremental degradation model combined with a staged sampling trajectory generation method, greatly improves randomness and further improves the accuracy of lifetime distribution, while reducing the data volume requirement. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0031] Figure 1 A schematic diagram of a photovoltaic module lifetime prediction method considering individual degradation differences provided in an embodiment of the present invention;

[0032] Figure 2 This diagram illustrates the degradation mechanism of photovoltaic modules due to key environmental factors.

[0033] Figure 3 Here is a flowchart of the quasi-Newton algorithm;

[0034] Figure 4 The average performance degradation trajectory of the S70L45 monocrystalline silicon photovoltaic module is given by the power degradation amount following the inverse Gaussian-Gamma model.

[0035] Figure 5 The average performance degradation trajectory of the S7L48 monocrystalline silicon photovoltaic module is given by the power degradation rate following an inverse Gaussian-Gamma model.

[0036] Figure 6 The average performance degradation trajectory of the S70L45 monocrystalline silicon photovoltaic module is given, where the power degradation follows a traditional inverse Gaussian model.

[0037] Figure 7 The average performance degradation trajectory of the S7L48 monocrystalline silicon photovoltaic module, whose power degradation follows a traditional inverse Gaussian model;

[0038] Figure 8 This is a lifetime distribution diagram for photovoltaic module 1S70L45;

[0039] Figure 9 This is a lifetime distribution diagram for photovoltaic module 4S7L48. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] like Figure 1 This invention discloses a method for predicting the lifetime of photovoltaic modules that takes into account individual degradation differences, comprising the following steps:

[0042] S1: Based on the inverse Gaussian-gamma mixture distribution, a degradation model is constructed for the power degradation increment. A random diffusion parameter variable following a gamma distribution is introduced into the inverse Gaussian distribution of the power degradation increment to represent the individual degradation differences of photovoltaic modules.

[0043]

[0044] in, It is time Time The power degradation increment, The average degradation rate, It is time Time The change in cumulative degradation intensity; For diffusion parameters, Let Gamma be the shape parameter of the distribution. For the rate parameter of the Gamma distribution.

[0045] S2: Based on the measured data of the photovoltaic module under test, the model parameters of the degradation model are quantitatively solved to confirm the optimal model parameters.

[0046] S3: Based on the degradation model under the optimal model parameters, and combined with randomly generated diffusion parameters, a corresponding component degradation trajectory is generated, and the component lifetime is confirmed according to a preset failure threshold; a lifetime distribution is generated by repeatedly sampling the diffusion parameters. The average lifetime of this type of photovoltaic module can be obtained by calculating the mean of the lifetime distribution.

[0047] The following is a detailed explanation of each step in this embodiment:

[0048] S1 describes the modeling process for the incremental power degradation model of photovoltaic modules. It assumes that the incremental power degradation follows an inverse Gaussian-gamma hybrid model, with the diffusion parameters of the traditional inverse Gaussian process following a gamma distribution. This reflects the individual differences in power degradation of photovoltaic modules. For photovoltaic modules of the same batch and model, under the same environment (same power plant, similar installation angle), the overall trend and average rate of power degradation are roughly the same. However, even if the overall degradation trend is consistent, the actual degree of power degradation of individual modules will still differ significantly due to individual differences.

[0049] (1)

[0050] (2)

[0051] (3)

[0052] The probability density function of the traditional inverse Gaussian distribution is (PDF):

[0053] (4)

[0054] The probability density function of the gamma distribution:

[0055] (5)

[0056] The marginal probability density function of the mixed distribution is obtained by multiplying the probability density function of the traditional inverse Gaussian distribution and the probability density function of the gamma distribution, and integrating with respect to η:

[0057] (6)

[0058] From equation 6, we get:

[0059] (7)

[0060] in, For time Time The power degradation increment, The average degradation rate, For time The cumulative degradation intensity is defined as follows: = , To control the rate of degradation, For diffusion parameters, Let Gamma be the shape parameter of the distribution. For the rate parameter of the Gamma distribution, This is a gamma function.

[0061] To further implement the above technical solution, when constructing the degradation model, multiple environmental covariates are introduced to optimize the average degradation rate, resulting in a comprehensive environmental degradation function:

[0062]

[0063] in, Relative humidity, Irradiance, B and C are both unknown parameters, i.e., parameters to be estimated, and R is the ideal gas constant, which takes the value of .

[0064] In this embodiment, temperature, humidity, and irradiance lead to the following degradation forms: direct degradation of semiconductor material performance, accelerated aging of packaging materials, corrosion of metal components and increased contact resistance; corrosion of metal electrodes, hydrolysis and delamination of packaging materials, back-side aging and insulation failure; photo-oxidation of materials caused by ultraviolet radiation, photo-induced degradation and current fluctuation damage, and aging of solar cells under long-term strong light, such as... Figure 2 As shown.

[0065] The Arrhenius relation is used to simulate the effect of temperature on the degradation rate, and the formula is as follows:

[0066] (8)

[0067] in, The rate constant measures how quickly a chemical reaction proceeds. This is a frequency factor, related to molecular collision frequency and orientation, and its unit is the rate constant. Consistent, Activation energy is the energy required for molecules to participate in a reaction. It is the energy needed for reactant molecules to change from their normal state to a transition state in which a chemical reaction can easily occur. The higher the activation energy, the more difficult the reaction is to proceed. Let be the ideal gas constant, and take the value of . , Thermodynamic temperature, measured in Kelvin (K), is an absolute temperature. =Celsius temperature +273.15.

[0068] Temperature degradation function:

[0069] (9)

[0070] Humidity degradation function:

[0071] (10)

[0072] Irradiance degradation function:

[0073] (11)

[0074] in, The relative humidity is expressed in %. The unit of irradiance is watts per square meter, and B and C are coefficients based on product or material characteristics. These are intermediate parameters obtained based on the frequency factor and activation energy. B and C are parameters that cannot be obtained directly, therefore they are unknown parameters and need to be obtained through parameter estimation.

[0075] Assuming that the degradation effects of temperature, relative humidity, and irradiance are independent of each other, the overall environmental degradation function is:

[0076] (12)

[0077] Simplify parameters

[0078] (13)

[0079] The environmental integrated model is to... Substituting it into the model, its degradation increment The probability density function will be affected by environmental covariates, and the degradation increment Follows an inverse Gaussian distribution:

[0080] (14)

[0081] Combination Follows Gamma distribution Then, considering the influence of environmental covariates, the degradation increment The marginal probability density function is:

[0082] (15)

[0083] in, To be in the time interval Average environmental covariate values ​​within, To be in the time interval The observed power degradation increment, The average degradation rate is determined by environmental covariates. For the increment of time effect, Let be the shape parameter of the Gamma distribution, and γ be the rate parameter of the Gamma distribution. This is the Gamma function.

[0084] S2 represents the parameter estimation process. The maximum likelihood method is used to quantify and solve the model parameters of the degenerate model, obtaining the parameters that maximize the likelihood function, which are then taken as the optimal model parameters.

[0085] For example, the unknown parameters are estimated by combining the maximum likelihood estimation method with the temperature, relative humidity, irradiance, and power degradation increment data of different models of monocrystalline silicon photovoltaic modules at each time point. Component 1 is model S70L45, and component 2 is model S7L48; the parameter estimation steps are as follows:

[0086] S21: Define the likelihood function based on the marginal probability density function of the degenerate model constructed in S1. The role of the log-likelihood function is to quantify the model parameters. The degree of fit to the observed data, for each positive degradation increment Its log-likelihood contribution is:

[0087] (16)

[0088] The likelihood function of the entire model:

[0089] (17)

[0090] Where n is the number of time steps, Data is the observed data, which records the degradation increment for each time period, and f is the likelihood function corresponding to time step i, the logarithmic form of which is Equation 16.

[0091] Log-maximal likelihood function:

[0092] (18)

[0093] in, It is the vector of model parameters that needs to be estimated.

[0094] S22: Construct the optimization objective function, with the goal of finding parameters that maximize the likelihood function. In numerical optimization, the problem is usually transformed into minimizing the negative log-likelihood function.

[0095] The negative log-likelihood is:

[0096] (19)

[0097] S23: Since the log-likelihood function is nonlinear, it cannot be solved directly using analytical methods. Therefore, a numerical optimization algorithm is employed. In this invention, the quasi-Newton method (L-BFGS-B) is used. For example... Figure 3 The flowchart shown is for the algorithm. The following are the specific solution steps of the L-BFGS-B algorithm in the optimization process:

[0098] S231: Initialization, first initialize the unknown parameters. Define the upper and lower bounds of the parameters to ensure that the parameter values ​​are within a reasonable range during the optimization process. Control the parameter settings to set the maximum number of iterations to 10,000 and the relative convergence tolerance, and control the termination conditions of the optimization process.

[0099] S232: Calculate the objective function and gradient. In each iteration, L-BFGS-B calls the negative log-likelihood function to calculate the negative log-likelihood value for the current parameters. The algorithm requires the first derivative (gradient) of the negative log-likelihood function with respect to the parameters. The gradient represents the direction of change of the objective function at the current parameter point and is used to determine the direction of parameter updates.

[0100] S233: Constructing an approximation of the Hessian matrix. The core of the quasi-Newton method is to approximate the inverse of the Hessian matrix by storing a finite number of vector pairs (usually the shifts and gradient changes from the most recent iterations) rather than directly calculating or storing the complete Hessian matrix. In each iteration, L-BFGS updates the Hessian approximation with the following information: parameter changes: in It is the first Parameter values ​​for the next iteration; gradient changes: ,in The gradient of the objective function is used as the basis for updating the formula to construct an approximation of the inverse of the Hessian. .

[0101] S234: Parameter update. In the... In this iteration, L-BFGS-B is used to calculate the search direction. ,in It is an approximation of the Hessian inverse. This is the current gradient; the step size is determined through line search. This makes the negative log-likelihood function value in the direction The value decreased significantly; the updated parameters are as follows: .

[0102] S235: Handling boundary constraints. After each parameter update, L-BFGS-B will... Projecting onto the constraint range, i.e. ,in and They are the first v The lower and upper bounds of each parameter.

[0103] S236: Convergence Check. Checks whether the convergence conditions are met: whether the gradient norm is sufficiently small; whether the parameter change is less than the relative tolerance; whether the maximum number of iterations has been reached. If any of these conditions are met, the algorithm stops and returns the current parameter estimates and the objective function value.

[0104] S237: Output results. After optimization, the optimized parameter values ​​are returned, and the parameters are obtained through inverse transformation. Returns the Hessian matrix of the objective function at the optimal parameter point, which is used to calculate the covariance matrix and correlation matrix of the parameters.

[0105] S24: The algorithm returns the set of parameters that minimizes the negative log-likelihood function, which is the maximum likelihood estimate.

[0106] S25: Evaluate the fit of the environmental covariate-inverse Gaussian-gamma mixture model based on the obtained maximum likelihood estimates.

[0107] S251: The degradation curve is obtained based on the estimated parameters. This curve represents the expected degradation trajectory of the component, i.e., the average degradation curve. The average degradation curve function is:

[0108] (20)

[0109] in, It is in time The cumulative power degradation over time.

[0110] The derivation of the above average degradation curve function is as follows:

[0111] Cumulative power degradation The expectation can be viewed as the sum of the expectations of all tiny degradation increments:

[0112] (twenty one)

[0113] in, The variable representing the degradation increment corresponding to the current time step. So:

[0114] (twenty two)

[0115] As the time step approaches infinitesimal, summation becomes integration, corresponding to:

[0116] (twenty three)

[0117] We set , ,so:

[0118] (twenty four)

[0119] Based on the degradation parameter estimation method proposed in this invention, the degradation parameters of the above-mentioned different types of monocrystalline silicon photovoltaic modules can be estimated.

[0120] Based on the estimated degradation parameters, the expected degradation trajectory of the component is obtained according to Formula 20, such as... Figure 4 The figure shown is a comparison between the average performance degradation trajectory and the actual degradation trajectory of a monocrystalline silicon photovoltaic module (model S70L45) that follows an inverse Gaussian-Gamma model. Figure 5 To determine the average performance degradation trajectory of the S7L48 monocrystalline silicon photovoltaic module, which conforms to the inverse Gaussian-Gamma model. Figure 6 To determine the average performance degradation trajectory of the S70L45 monocrystalline silicon photovoltaic module, which conforms to the traditional inverse Gaussian model. Figure 7 To understand the average performance degradation trajectory of the S7L48 monocrystalline silicon photovoltaic module, which conforms to the traditional inverse Gaussian model, the cumulative degradation predicted by the inverse Gaussian-gamma hybrid model, which considers the individual differences of photovoltaic modules, has a higher degree of fit with the actual degradation. This can be more intuitively seen below through the mean square error.

[0121] S252: Mean Squared Error (MSE) is the average of the squares of the differences between predicted and actual values. A smaller MSE value indicates a smaller overall deviation between the model's predicted and actual values, and higher prediction accuracy. Formula:

[0122] (25)

[0123] The number of samples for each component is n; This represents the actual observed value of the m-th sample; This is the model's predicted value for the m-th sample.

[0124] Observing Table 1, the mean square error of the traditional model, i.e. the traditional inverse Gaussian model, is generally higher than that of the inverse Gaussian-gamma mixture model, indicating that the mixture model has a small overall deviation between the predicted degradation trajectory and the actual degradation trajectory and has high accuracy.

[0125] Table 1. Comparison of MSE between the traditional model and the inverse Gaussian-Gamma mixture model.

[0126]

[0127] S3 represents the lifetime prediction process.

[0128] To determine whether the degradation of a module has reached a pre-set failure threshold and when the failure threshold is reached, the failure threshold for photovoltaic modules is generally defined as 20% of the initial power output of the photovoltaic module in service.

[0129] (26)

[0130] in, Initial output power of photovoltaic modules This is the failure threshold.

[0131] Using an environmental covariate-inverse Gaussian-gamma mixture model with estimated parameters, Monte Carlo simulations generate a large number of independent degradation paths at each small time step. Inside, according to formula 15, one is randomly selected. This process considers the changes in degradation rate over time and random effects. Each simulated path represents the complete degradation history of a component. For each simulated degradation path, the method continuously tracks the cumulative degradation. When the cumulative degradation first reaches a preset failure threshold (e.g., 20%), the corresponding time is recorded. This time is the failure time under that simulated path, also known as the first traverse time. Repeating the above simulation process 10,000 times will yield a sample set of failure times (lifetimes). Statistical analysis is performed on these simulated lifetime values ​​to obtain the failure time distribution.

[0132] In the Monte Carlo process, diffusion parameters are first randomly sampled according to a gamma distribution. The parameters of the gamma distribution have been obtained through S2. The degradation model can be confirmed based on the sampled diffusion parameters. Then, degradation is performed based on the initial power of the photovoltaic module. The degradation process is divided into multiple time steps. The degradation amount, i.e., the degradation increment, at each time step is also confirmed by random sampling. The degradation increment follows the degradation model at each time step. Therefore, sampling within the inverse Gaussian distribution it follows can confirm the increment at each time step, thereby simulating the degradation process until it decays to a preset threshold, confirming the lifetime corresponding to the degradation curve. By repeating the above sampling process, the lifetime distribution can be obtained.

[0133] Table 2. Relative Errors Between Predicted and Actual Lifespans

[0134]

[0135] The following lifetime distributions are all based on the cumulative degradation of photovoltaic modules, following an inverse Gaussian-Gamma model, and do not consider environmental factors. Figure 8 The lifespan distribution of photovoltaic module 1 is shown. The average lifespan is 7.09 years, the median lifespan is 7.10 years, and the standard deviation is 0.4 years. The lifespan data is relatively concentrated. Figure 9 The lifespan distribution of photovoltaic module 2 is shown in Table 2. The average lifespan is 6.29 years, the median lifespan is 6.30 years, and the standard deviation is 0.41 years. The relative error between the predicted lifespan and the actual lifespan is shown in Table 2.

[0136] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0137] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those 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 invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting the lifetime of photovoltaic modules considering individual degradation differences, characterized in that, Includes the following steps: S1: Based on the inverse Gaussian-gamma mixture distribution, a degradation model is constructed for the power degradation increment. A random diffusion parameter variable following a gamma distribution is introduced into the inverse Gaussian distribution of the power degradation increment to represent the individual degradation differences of photovoltaic modules. in, To introduce a comprehensive environmental degradation function with multiple environmental covariates, where T is the absolute temperature, Relative humidity, Irradiance, B and C are the coefficients to be estimated, and R is the ideal gas constant; It is time Time The power degradation increment, where i is the time step variable, and IG represents the inverse Gaussian distribution. It is time Time The change in cumulative degradation intensity; For diffusion parameters, Let Gamma be the shape parameter of the distribution. For the rate parameter of the Gamma distribution; S2: Based on the observation data of the photovoltaic module under test, the model parameters of the degradation model are quantitatively solved to confirm the optimal model parameters; wherein, the optimal model parameters include the optimal environmental covariate parameters and the optimal diffusion distribution parameters; the optimal environmental covariate parameters and the optimal diffusion distribution parameters are generated simultaneously; S3: Based on the degradation model under the optimal model parameters, generate the corresponding component degradation trajectory by combining the randomly generated diffusion parameters, and confirm the component lifetime according to the preset failure threshold; generate the lifetime distribution by repeatedly sampling the diffusion parameters.

2. The photovoltaic module lifetime prediction method considering individual degradation differences according to claim 1, characterized in that, In S1, the expression for the cumulative degradation intensity is: in, To control the degradation rate, t is the degradation time.

3. The photovoltaic module lifetime prediction method considering individual degradation differences according to claim 2, characterized in that, In step S2, the maximum likelihood method is used to quantify and solve the model parameters of the degenerate model to obtain the parameters that maximize the likelihood function, which are then used as the optimal model parameters.

4. The photovoltaic module lifetime prediction method considering individual degradation differences according to claim 3, characterized in that, The likelihood function is: Where L is the likelihood function. This is the vector of model parameters to be estimated, where i is the time step variable, n is the number of time steps, Data represents the observed data, and f is the likelihood function corresponding to time step i. To be in the time interval The average environmental covariate values ​​of the corresponding absolute temperature, relative humidity, and irradiance.

5. The photovoltaic module lifetime prediction method considering individual degradation differences according to claim 2, characterized in that, In step S3, the step of generating the degradation trajectory includes: The degradation trajectory is obtained by iterating through each time step and sampling according to the marginal probability density function of the degradation model at each time step, based on the extracted degradation increment.

6. The photovoltaic module lifetime prediction method considering individual degradation differences according to claim 5, characterized in that, During the generation of the degradation trajectory, the diffusion parameter is determined to follow a gamma distribution based on the optimal model parameters, and the diffusion parameter is randomly generated within the corresponding gamma distribution.

7. The photovoltaic module lifetime prediction method considering individual degradation differences according to claim 5, characterized in that, The marginal probability density function is: in, To be in the time interval The corresponding average environmental covariate values ​​for absolute temperature, relative humidity, and irradiance. To be in the time interval The observed power degradation increment, Let be the shape parameter of the Gamma distribution, and γ be the rate parameter of the Gamma distribution. This is the Gamma function.

Images