Photovoltaic power generation reliability evaluation and power prediction method considering time-varying failure rate

Abstract

The invention provides a photovoltaic power generation reliability evaluation and power prediction method considering a time-varying failure rate, and belongs to the field of renewable energy sources. The photovoltaic power generation reliability evaluation and power prediction method considering the time-varying failure rate comprises the following steps: predicting solar irradiance and temperature per hour from historical data by using Monte Carlo simulation based on a random diffusion process (SDP), and modeling a cloud transient state by using a jump diffusion process integrated in the SDP, and then establishing a reliability evaluation index and photovoltaic generating capacity prediction model of the photovoltaic system considering the time-varying failure rate, thereby more accurately evaluating the reliability of the photovoltaic power generation system and predicting the photovoltaic generating capacity.

Description

Technical Field

[0001] The invention belongs to the technical field of photovoltaic power generation, and relates to a photovoltaic power generation reliability evaluation and power prediction method considering time-varying failure rate. Background Art

[0002] With the increase in the penetration rate of renewable energy, the randomness of solar radiation and other climate factors increases the uncertainty of photovoltaic power generation for photovoltaic power generation systems. In photovoltaic power generation systems, random disturbances include three main sources: renewable energy generation, network events, and load changes. Generally, there are two types of random disturbances: continuous and discrete. Examples of continuous random disturbances are load changes, changes in renewable energy generation, etc. Discrete random disturbances include severe weather events, network failures, cloud events, etc. Random processes are used to simulate random disturbances or uncertainties. Using random processes to build reliability evaluation indicators and power generation prediction models for photovoltaic power generation systems can more accurately handle continuous and discrete uncertainties and make more effective predictions on the reliability of photovoltaic systems.

[0003] At present, a lot of research has been done on the simulation of solar irradiance using stochastic differential equations (SDEs). An improved clear sky index model was proposed based on the use of SDES with jumps to model the clear sky index, and the autoregressive moving average model was used to deal with the movement of clouds, but the above model did not consider the uncertainty of solar irradiance. Later, the clear sky index was modeled by considering different types of cloud events, namely large clouds and small clouds based on the size of cloud impact, to deal with the uncertainty of solar irradiance. However, the research methods that have been conducted on the simulation of solar irradiance using SDEs are limited to exponential autocorrelation. While the autocorrelation of solar irradiance follows a sinusoidal form and exhibits seasonality, the exponential decay autocorrelation used in the clear sky model is not suitable for solar irradiance prediction. Therefore, it is necessary to develop a model for predicting solar irradiance based on the non-exponential decay autocorrelation method of SDE. In addition, SDEs have received attention in the research of new energy prediction, such as the autoregressive moving average (ARMA) and autoregressive integrated moving average (ARIMA) models for short-term prediction of power generation. However, in the reliability evaluation indicators developed for planning work, short-term prediction models are not used. Summary of the invention

[0004] Based on the deficiencies of the prior art, in order to more accurately deal with continuous and discrete uncertainties, a non-exponential decay autocorrelation method is adopted and combined with the impact of cloud transients on solar power generation. The present invention provides a photovoltaic power generation reliability evaluation method considering time-varying failure rate, which is used to evaluate the reliability of photovoltaic power generation systems. The present invention also provides a photovoltaic power generation power prediction method considering time-varying failure rate, which is used to predict photovoltaic power generation power.

[0005] To solve the above technical problems, the present invention is implemented through the following technical solutions: a method for evaluating the reliability of photovoltaic power generation considering time-varying failure rate, comprising: predicting hourly solar irradiance and temperature from historical data by using Monte Carlo simulation based on random diffusion process (SDP), integrating the cloud transient model into the solar irradiance model by using the jump diffusion process integrated in the random diffusion process, and then building a reliability evaluation index for the photovoltaic power generation system considering the time-varying failure rate.

[0006] Further preferably, when using Monte Carlo simulation based on random diffusion process to predict hourly solar irradiance and temperature from historical data, a non-exponential decay autocorrelation method Ornstein-Uhlenbeck type random diffusion process is used to simulate solar irradiance and temperature.

[0007] Further preferably, the cloud transient model comprises: using a Poisson process to model the number of cloud transients occurring during a time duration, and the cloud transient amplitude and duration are generated by random numbers generated by a normal distribution.

[0008] Further preferably, the process of modeling cloud transients using a jump diffusion process integrated in a random diffusion process includes: the jump diffusion process is driven by the jump amplitude, duration and number of jumps within the study period, the cloud transients are modeled as discrete random events through the random diffusion equation, and the cloud transients are integrated into the solar irradiance prediction model by incorporating the jump diffusion term as additional noise into the solar irradiance prediction model.

[0009] Further preferably, the reliability evaluation index of the photovoltaic power generation system considering the time-varying failure rate includes: using Monte Carlo simulation to initialize the simulation year and month, using the failure physics reliability model to obtain the component failure rate, and obtaining the exponential distribution of the failure rate according to the probability density function and the cumulative distribution function. Then, the failure time and repair time of the component are determined, and each cycle is updated based on the new cumulative distribution function of the failure rate until the complete state of the component and the time-varying failure rate are obtained as the reliability evaluation index.

[0010] The technical effects achieved by the present invention are as follows:

[0011] The hourly solar irradiance and temperature are predicted from historical data by using Monte Carlo simulation based on random diffusion process (SDP). Secondly, cloud transients are modeled using the jump diffusion process integrated in SDP. Then, a reliability evaluation index of photovoltaic power generation system considering time-varying failure rate is established. Based on the reliability evaluation index, a photovoltaic power generation prediction model is established using the SMCS algorithm, so as to more accurately evaluate the reliability of photovoltaic power generation system and predict photovoltaic power generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0013] The present invention will be described in detail below in conjunction with the accompanying drawings.

[0014] A reliability evaluation method for photovoltaic power generation considering time-varying failure rate includes: predicting hourly solar irradiance and temperature from historical data by using Monte Carlo simulation based on random diffusion process (SDP), integrating cloud transient model into solar irradiance model by using jump diffusion process integrated in random diffusion process, and then building reliability evaluation index of photovoltaic power generation system considering time-varying failure rate.

[0015] The process of predicting hourly solar irradiance (solar irradiance prediction model) is as follows:

[0016] The one-dimensional random diffusion process (SDP) is defined by the stochastic differential equation:

[0017] dX=A(X,t)*dt+B(X,t)*dW(t),t∈[0,T] (1)

[0018] Where X is a random process, W(t) is a Weiner process (Brownian motion), the terms A(X,t) and B(X,t) are the drift coefficient and diffusion coefficient of the SDP, t is time, and T is the time period.

[0019] For photovoltaic power generation systems, considering the seasonality and frequent fluctuations of solar irradiance, the Ornstein-Uhlenbeck type SDP model is used to simulate solar irradiance. The Ornstein-Uhlenbeck process is used to simulate random variables with mean reversion characteristics, and the drift coefficient and diffusion coefficient of the process are as follows:

[0020] A(X,t)=α*(μ-X),B(X,t)=σ t (2)

[0021] Among them, α is the mean regression parameter, μ is the mean, σ t is the volatility of the random process.

[0022] set up is the average solar irradiance at the tth hour of month m, then the stochastic differential equation for predicting the solar irradiance at the tth hour is:

[0023]

[0024] in, is the mean regression parameter of solar irradiance in month m, is the fluctuation of solar irradiance in hour t of month m; E(t) mis the solar irradiance at time t in month m.

[0025] The evaluation indicators of the drift equation and the diffusion equation are as follows:

[0026] 1) Evaluation of the solar diffusion equation: The diffusion coefficient of solar irradiance prediction is modeled using the Fourier series. The following formula is used to evaluate the fluctuation of solar irradiance at the tth hour of month m:

[0027]

[0028] in and The fluctuation of solar irradiance The Fourier series coefficients of are obtained from historical data using parameter estimation methods. I is the number of days from the starting date to the current date, J is the number of years from the starting year to the current year, i and j are corresponding counting variables, and T E is the time period of the day during which the solar irradiance is considered, Solar irradiance fluctuations on the start day.

[0029] 2) Evaluation of the solar drift equation: Solar irradiance has a periodic diurnal variation. Therefore, the diurnal variation is modeled using Fourier series. The process of estimating the drift equation parameters is as follows.

[0030] Estimation of mean solar irradiance using Fourier series

[0031]

[0032] in, and is the average solar irradiance The Fourier series coefficients of are estimated from historical data using parameter estimation methods. I is the number of days from the starting date to the current date, J is the number of years from the starting year to the current year, and i and j are corresponding counting variables. is the solar irradiance on the starting day.

[0033] Mean reversion parameters The calculation formula is as follows:

[0034]

[0035] Among them, num E Indicates the changing trend of solar irradiance, den E Represents the fluctuation nature of solar irradiance, both of which are used to calculate mean regression parameters The intermediate variable, E(i) m is the solar irradiance value at the i-th time point in month m, is the standard deviation of the solar irradiance value at the i-1th time point in month m, is the average solar irradiance value at the ith time point in month m.

[0036] 3) Monte Carlo simulation based on SDP: Based on historical data, after evaluating the drift equation and diffusion equation, the predicted solar irradiance at time t in month m is predicted:

[0037]

[0038] Hourly solar irradiance obtained from Monte Carlo simulation The calculation is as follows:

[0039]

[0040] Among them, N mcs is the specified number of simulation years, is the solar irradiance at time t in month m of year i in a specified simulation year.

[0041] The process of predicting hourly temperatures (temperature prediction model) is as follows:

[0042] The present invention adopts Ornstein-Uhlenbeck process to predict the hourly temperature of each month. The drift coefficient and diffusion coefficient of SDP reflect the random and seasonal characteristics of temperature, and the temperature of the tth hour is predicted to be:

[0043]

[0044] in, is the temperature mean reversion parameter, is the temperature fluctuation at hour t in month m, is the average temperature at hour t in month m, T(t) m is the temperature at hour t in month m. The drift equation and diffusion equation are evaluated as follows:

[0045] 1) Evaluation of the temperature diffusion equation: The diffusion coefficient for temperature prediction is modeled using the Fourier series. m Temperature fluctuation at hour t for:

[0046]

[0047] in, and For temperature fluctuations The Fourier series coefficients of are estimated from historical data using parameter estimation methods. I is the number of days from the starting date to the current date, J is the number of years from the starting year to the current year, and i and j are corresponding counting variables. T is the temperature period considered during the day, is the temperature diffusion coefficient on the starting day.

[0048] 2) Evaluation of temperature drift equation: Usually, temperature has seasonality and pattern, so the sine function can better fit the temperature model. The method to estimate the drift equation parameters is as follows:

[0049] Calculate the average temperature using the sine function

[0050]

[0051] Where a, b, c, and d are average temperature coefficients, which are obtained from historical data using parameter estimation methods.

[0052] Temperature mean reversion parameters The evaluation is carried out using the following formula:

[0053]

[0054] Among them, num T Indicates the temperature change trend, den T Indicates the fluctuation trend of temperature, which is used to calculate the mean regression parameter The intermediate variable, T(i) m is the temperature at the i-th time point in month m, is the standard deviation of the temperature at the i-th time point in month m, is the average temperature at the i-th time point in month m.

[0055] 3) Monte Carlo simulation based on SDP: Based on historical data, the drift equation and diffusion equation are evaluated. Then, the predicted temperature at hour t of month m is estimated using equation (13):

[0056]

[0057] Hourly solar irradiance obtained from Monte Carlo simulation The calculation formula is as follows:

[0058]

[0059] Among them, N mcs is the specified number of simulation years, is the temperature at time t in month m of year i in a specified simulation year.

[0060] The process of modeling cloud transients is:

[0061] The jump diffusion process usually models discrete random events associated with a random process, and the mathematical representation is as follows:

[0062] dX=A(X,t)*dt+B(X,t)*dW(t)+ξ*dJ(t) (15)

[0063] Among them, ξ is the jump amplitude and J(t) is the jump duration.

[0064] In the SDP environment, cloud transients are modeled as discrete random events. Cloud transients are integrated into SDP, and the jump diffusion term is used as additive noise in Equation (3). The jump diffusion process is driven by the jump amplitude, duration, and number of jumps within the study period, simulating the random jump diffusion process (SJDP) of cloud transients:

[0065]

[0066] Among them, J j (t) takes the value of 1 or 0, and the cloud duration T cloud It is equal to 1 during the period and becomes 0 once the cloud event ends. clouds is the transient number, α E is the mean regression parameter of solar irradiance, σ E (t) is the solar irradiance fluctuation at time t, E(t) avg is the average solar irradiance at time t.

[0067] The number of cloud transients within a time duration T is modeled using a Poisson process with an occurrence rate of λ cloud .

[0068] N clouds ~poisson(λ cloud ) (17) The cloud transient start time is calculated using the following formula:

[0069] τ i =τ i-1 -log(U) / λ cloud (18)

[0070] Among them, τ i is the cloud transient starting time at the i-th moment, τ i-1 is the cloud transient start time of the previous moment;

[0071] The magnitude of cloud transients is modeled using random numbers with a mean μ ξ, and standard deviation σ ξ They are all normally distributed, as shown below:

[0072] ξ~normal(μ ξ ,σ ξ ) (19)

[0073] The transient cloud duration adopts a normal distribution, and the average cloud duration μcloud and standard deviation σ cloud as follows:

[0074] T cloud ~normal(μ cloud ,σ cloud ) (20)

[0075] The photovoltaic reliability evaluation index includes:

[0076] Physics of failure models are often used to model the time-varying failure rates of structures and electronic components. The model considers the effects of different factors such as physical, thermal, process, manufacturing, and electrical on the failure rate. The time-varying failure rate of any electronic component is shown below:

[0077] λ=λ phy *Π PM *Π Process (twenty one)

[0078] Among them, λ phy is the physical factor, PM is a factor that indicates manufacturing quality, Process Factors that represent the component usage process.

[0079] The reliability evaluation index is used to evaluate the time-varying failure rate of photovoltaic power generation system components and photovoltaic power generation output. The failure rate of photovoltaic power generation system components in month m is λ m Formula (22):

[0080]

[0081] in, is the physical factor of month m, Π PM is a factor that indicates manufacturing quality, Process Factors that represent the component usage process.

[0082] Monthly physical factors The calculation formula is:

[0083]

[0084] in Ω con ,Ω fu ,Ω fcon are the repair rates of converter transistor, converter diode, converter capacitor, inverter transistor, PV connector, fuse and fuse connector, respectively. f(·) is the repair rate of the calculated weather variable Ω w The cumulative function of the repair rate of parameters of different components of the photovoltaic power generation system.

[0085] in:

[0086]

[0087] in They are the converter voltage, current, converter transistor temperature, diode temperature, inverter current, inverter transistor temperature and string current at L level in month m respectively. are the number of hours, average temperature, average humidity, number of cycles, average cycle duration temperature difference and maximum temperature of layer L in month m. f(·) is the function that uses these parameters to calculate the repair rate of the corresponding component. The repair rate parameter of each component is also given by the physical failure model Ω=Ω phy *Π PM *Π Process It is concluded that Ω phy ,Π PM , Π Process Physical factors, manufacturing quality factors and usage process factors for each component.

[0088] At hour t in month m, the failure rate of the component is λ m (t) indicates that, considering that the failure rate follows an exponential distribution, the probability density function f(t) and the cumulative distribution function F of the component t (T) is:

[0089]

[0090] Among them, e is a natural constant.

[0091] Generate a random number U f , by inversely transforming the cumulative distribution function obtained in the previous step, the failure time TTF of the component is determined as follows:

[0092] TTF=F t -1 (U f ) (26)

[0093] In addition, the repair time TTR of the component is calculated by another random number U r Directly determine:

[0094]

[0095] μ is the repair rate of the component, which is assumed to be constant over time.

[0096] After each cycle TTF+TTR, new TTF and TTR are generated based on the new cumulative distribution function of the failure rate until the complete status of the component is obtained.

[0097] A photovoltaic power generation prediction method considering time-varying failure rate:

[0098] On the basis of evaluating the monthly failure rate of photovoltaic power generation system components and based on the reliability evaluation index, the sequential MCS (SMCS) algorithm is used to predict the hourly output power of the photovoltaic power generation system:

[0099]

[0100] in is the hardware availability of the photovoltaic power generation system; E0, P m0 , T0, γ P are the irradiance, rate power, temperature, and temperature coefficient of the solar photovoltaic panel under standard test conditions; E(t) is the solar irradiance at time t obtained by the computer. The output power depends on the fault type of the photovoltaic power generation system. P loss,i is the failure rate of the series connected solar panels, Ω S is the fault type set of the series battery panels, P loss,j is the failure rate of the photovoltaic array, Ω A is a set of PV array fault types.

[0101] The actual application process of the present invention refers to Figure 1 , the execution process is as follows:

[0102] First, the hourly solar irradiance and temperature for each month are predicted using SDP-based Monte Carlo simulation;

[0103] The algorithm of the solar irradiance prediction model is as follows:

[0104] 1) The hourly solar irradiance data for each month is imported from historical data, and the matrix Where h is hours and y is years.

[0105] 2) The mean and variance of each hour of each month are calculated from the historical data consisting of y years, using the matrices and express.

[0106] 3) Fitting the matrix using parameter estimation method Hourly mean data, estimate the average solar irradiance Parameters ( and ).

[0107] 4) Use parameter estimation method to estimate the matrix The hourly variance data is fitted to estimate volatility, Parameters ( and ).

[0108] (5) Use equation (5) and equation (4) to calculate the average solar irradiance per hour for the month and volatility

[0109] 6) Use formula (6) to regress the mean parameters of each month Depend on and Estimate. Set n mcs =1,m=1.

[0110] 7) Calculate the predicted solar irradiance per hour using formula (7)

[0111] 8) Repeat steps 7 and 9 for all months of the year (ie until m=12).

[0112] 9) Repeat steps 7 and 8 for the specified number of simulation years, N mcs .

[0113] 10) The hourly solar irradiance obtained by Monte Carlo simulation is:

[0114] The algorithm of the temperature prediction model is as follows:

[0115] 1) From the matrix Import the hourly temperature data for each month into the historical data represented.

[0116] 2) The mean and variance of each hour of each month are calculated from the historical data consisting of y years, using the matrix and express.

[0117] 3) By fitting the matrix The average temperature is estimated by using the parameter estimation method based on the hourly average data in Parameters (a, b, c & d).

[0118] 4) By the matrix The hourly variance data is fitted and the volatility is estimated using parameter estimation method. Parameters ( and ).

[0119] 5) The average solar irradiance per hour per month was calculated using equations (11) and (10) respectively. and volatility

[0120] 6) Use (12) to regress the mean parameters of each month Depend on and Estimate. Set n mcs =1,m=1.

[0121] 7) Calculate the predicted solar irradiance per hour using formula (13):

[0122] 8) Repeat steps 7 and 9 for all months of the year (ie until m=12).

[0123] 9) Repeat steps 7 and 8 for the specified number of simulation years, N mcs .

[0124] 10) The hourly solar irradiance obtained by Monte Carlo simulation is:

[0125] Secondly, the cloud transient is integrated into the solar irradiance prediction model using Monte Carlo simulation based on SJDP. The algorithm is as follows:

[0126] 1) Use the algorithm of the solar irradiance prediction model to import the predicted hourly solar irradiance data E(t) p .

[0127] 2) Assume i=1, j=1, τ0=0, t′=t′0.

[0128] 3) Number of cloud transients occurring during time duration T (N clouds ) is modeled using the Poisson process specified by (17) with an occurrence rate of λ cloud .

[0129] 4) Using formula (18) to calculate the cloud transient start time (τ i ).

[0130] 5) When (τ i <t′ j ):For the time interval, t∈[t′,τ i ), the updated solar irradiance is: E(t) pc =E(t) p ; In τ i Simulate the cloud transient state at any time. Its duration (T cloud ) is calculated using formula (20); for the time interval t∈[τ i ,(τ i +T cloud )], the cloud level (ξ) is simulated by (19), and the updated solar irradiance is as follows: Let i = i + 1, and use (18) to simulate the time τ when the next cloud transient occurs i .

[0131] 6) For the time interval t∈[t′,t′j], the updated solar irradiance is: E(t) pc =E(t) p .

[0132] 7) Let j = j + 1, t' = t' j , proceed to step-5.

[0133] 8) Finally, considering the transient solar irradiance of clouds, E(t) pc Given.

[0134] Then, the sequential Monte Carlo simulation initializes the simulated year and hour of the year m = 1 and t = 1; using Calculate the failure rate of the PV module at the tth hour; obtain the exponential distribution of the failure rate based on the probability density function and the cumulative distribution function; determine the failure time TTF and repair time TTR of the module; update each TTF+TTR cycle based on the new cumulative distribution function of the failure rate until the complete state of the module is obtained, that is, the instantaneous failure rate, which is used as a reliability evaluation indicator.

[0135] Finally, based on the reliability evaluation index, Predict the hourly output power of a photovoltaic system.

[0136] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0137] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A photovoltaic power generation reliability evaluation method considering time-varying failure rate, characterized in that: Monte Carlo simulation based on random diffusion process is used to predict hourly solar irradiance and temperature from historical data. The jump diffusion process integrated in the random diffusion process is used to model cloud transients. Then, a reliability evaluation index of photovoltaic power generation system considering time-varying failure rate is constructed.

2. A photovoltaic power generation reliability evaluation method considering time-varying failure rate according to claim 1, characterized in that: When predicting hourly solar irradiance and temperature from historical data using Monte Carlo simulation based on random diffusion process, a non-exponential decay autocorrelation method Ornstein-Uhlenbeck type random diffusion process is used to simulate solar irradiance and temperature.

3. A photovoltaic power generation reliability evaluation method considering time-varying failure rate according to claim 1, characterized in that: The cloud transient model includes: using a Poisson process to model the number of cloud transients occurring during a time duration, and the cloud transient amplitude and duration are generated by random numbers generated by a normal distribution.

4. A photovoltaic power generation reliability evaluation method considering time-varying failure rate according to claim 1, characterized in that: The process of modeling cloud transients using the jump diffusion process integrated in the random diffusion process includes: the jump diffusion process is driven by the jump amplitude, duration and number of jumps within the study period, cloud transients are modeled as discrete random events through the random diffusion equation, and the cloud transients are integrated into the solar irradiance prediction model by incorporating the jump diffusion term as additional noise into the solar irradiance prediction model.

5. A photovoltaic power generation reliability evaluation method considering time-varying failure rate according to claim 1, characterized in that: The reliability evaluation indicators of the photovoltaic power generation system considering the time-varying failure rate include: using Monte Carlo simulation to initialize the simulation year and month, using the failure physics reliability model to obtain the component failure rate, and obtaining the exponential distribution of the failure rate based on the probability density function and the cumulative distribution function. Then, the failure time and repair time of the component are determined, and each cycle is updated based on the new cumulative distribution function of the failure rate until the complete state of the component and the time-varying failure rate are obtained as the reliability evaluation indicator.

6. A photovoltaic power generation reliability evaluation method considering time-varying failure rate according to claim 1, characterized in that: Hourly solar irradiance and temperature predictions from historical data using Monte Carlo simulations based on a random diffusion process, including: The Ornstein-Uhlenbeck type SDP model is used to simulate solar irradiance; the Ornstein-Uhlenbeck process is used to simulate random variables with mean reversion characteristics, and the drift coefficient and diffusion coefficient of the process are as follows: A(X,t)=α*(μ-X),B(X,t)=σ t Among them, α is the mean regression parameter, μ is the mean, σ t is the volatility of the random process; X is the random process, A(X,t) and B(X,t) are the drift coefficient and diffusion coefficient of SDP, and t is time; set up is the average solar irradiance at the tth hour of month m, then the stochastic differential equation for predicting the solar irradiance at the tth hour is: in, is the mean regression parameter of solar irradiance in month m, is the fluctuation of solar irradiance in hour t of month m; E(t) m is the solar irradiance at time t of month m, W(t) is the Weiner process; The diffusion coefficient for solar irradiance prediction is modeled using Fourier series; Modeling diurnal variations using Fourier series; Monte Carlo simulation based on SDP, based on historical data, after evaluating the drift equation and diffusion equation, predicts the predicted solar irradiance at time t in month m; The process of predicting hourly temperature is as follows: the Ornstein-Uhlenbeck process is used to predict the hourly temperature of each month; the drift coefficient and diffusion coefficient of SDP reflect the random and seasonal characteristics of temperature, so the predicted temperature at hour t is: in, is the temperature mean reversion parameter, is the temperature fluctuation at hour t in month m, is the average temperature at hour t in month m, T(t) m The temperature at hour t in month m is: The diffusion coefficient for temperature prediction was modeled using Fourier series; Calculate the average temperature using the sine function Monte Carlo simulation based on SDP: Based on historical data, the drift equation and the diffusion equation are evaluated; then, the predicted temperature at hour t of month m is estimated; the hourly solar irradiance obtained by the Monte Carlo simulation 7. A photovoltaic power generation reliability evaluation method considering time-varying failure rate according to claim 1, characterized in that: The processes used to model cloud transients are: jump-diffusion processes usually model discrete random events associated with stochastic processes; The cloud transient is integrated into the SDP, and the jump diffusion term is used as additive noise; the jump diffusion process is driven by the jump amplitude, duration, and number of jumps within the study period, simulating the random jump diffusion process of cloud transient: Among them, J j (t) takes the value of 1 or 0, and the cloud duration T cloud It is equal to 1 during the period and becomes 0 once the cloud event ends. clouds is the transient number, E(t) is the solar irradiance at time t, α E is the mean regression parameter of solar irradiance, σ E (t) is the solar irradiance fluctuation at time t, E(t) avg is the average solar irradiance at time t; The number of cloud transients within a time duration T is modeled using a Poisson process with an occurrence rate of λ cloud ; N clouds ~fish(λ cloud ); The cloud transient start time is calculated using the following formula: t i =t i-1 -log(U) / λ cloud ; Among them, τ i is the cloud transient starting time at the i-th moment, τ i-1 is the cloud transient start time of the previous moment; The magnitude of cloud transients is modeled using random numbers with a mean μ ξ , and standard deviation σ ξ They are all normally distributed, as shown below: ξ~normal(μ ξ ,s ξ ); The transient cloud duration adopts a normal distribution, and the average cloud duration μ cloud and standard deviation σ cloud as follows: T cloud ~normal(μ cloud ,s cloud )。 8. A photovoltaic power generation reliability evaluation method considering time-varying failure rate according to claim 1, characterized in that: The photovoltaic reliability evaluation index includes: The time-varying failure rate of any electronic component is given by: λ=λ phy *P PM *P Process ; Among them, λ phy is a physical factor, PM is a factor that indicates manufacturing quality, Process is a factor that represents the use process of components; the reliability evaluation index is used to evaluate the time-varying failure rate of photovoltaic power generation system components and photovoltaic power generation output. The failure rate of photovoltaic power generation system components in month m is λ m : in, is the physical factor of month m, Π PM is a factor that indicates manufacturing quality, Process Factors that represent the process of component use; Monthly physical factors The calculation formula is: in Ω con ,Ω fu ,Ω fcon They are the repair rates of converter transistors, converter diodes, converter capacitors, inverter transistors, photovoltaic connectors, fuses and fuse connectors respectively; where: in: in They are the converter voltage, current, converter transistor temperature, diode temperature, inverter current, inverter transistor temperature and string current at L level in month m respectively; are the number of hours, average temperature, average humidity, number of cycles, average cycle duration temperature difference and maximum temperature of layer L in month m; f(·) is the function for calculating the repair rate of the corresponding component. The repair rate parameter of each component is also given by the physical failure model Ω=Ω phy *Π PM *Π Process It is concluded that Ω phy , Π PM , Π Process Physical factors, manufacturing quality factors and usage process factors for each component; At hour t in month m, the failure rate of the component is λ m (t) indicates that, considering that the failure rate follows an exponential distribution, the probability density function f(t) and the cumulative distribution function F of the component t (T) is: Among them, e is a natural constant; Generate a random number U f , by inversely transforming the cumulative distribution function obtained in the previous step, the failure time TTF of the component is determined as follows: TTF=F t -1 (U f ); In addition, the repair time TTR of the component is calculated by another random number U r Directly determine: μ is the component repair rate, which remains constant over time; After each cycle TTF+TTR, new TTF and TTR are generated based on the new cumulative distribution function of the failure rate until the complete status of the component is obtained.

9. A photovoltaic power generation power prediction method considering time-varying failure rate, characterized in that: On the basis of evaluating the monthly failure rate of the components of the photovoltaic power generation system, based on the appropriate reliability evaluation index of the method described in any one of claims 1 to 8, the sequential MCS algorithm is used to predict the hourly output power of the photovoltaic power generation system: in is the hardware availability of the photovoltaic power generation system; E0, P m0 , T0, γ P are the irradiance, rate power, temperature, and temperature coefficient of the solar photovoltaic panel under standard test conditions; E(t) is the solar irradiance at time t obtained by the computer. The output power depends on the fault type of the photovoltaic power generation system. P loss,i is the failure rate of the series connected solar panels, Ω S is the fault type set of the series battery panels, P loss,j is the failure rate of the photovoltaic array, Ω A is a set of PV array fault types.

Images