A method and system for estimating the probability distribution of capacitor life

By establishing a linear model of capacitor attenuation time and capacitance value attenuation ratio and introducing random error terms to calculate the probability distribution of capacitor life, the problems of high complexity and large calculation amount of existing algorithms are solved, and high precision and uncertainty quantization are achieved Lifespan estimates, suitable for both online and offline testing.

CN112818293BActive Publication Date: 2024-11-19HUAZHONG UNIV OF SCI & TECH +1
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202110100410.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-01-26
Publication Date
2024-11-19
Estimated Expiration
2041-01-26

AI Technical Summary

Technical Problem

The existing capacitor life estimation calculation methods are complex and have a large amount of computation, making it difficult to reduce the demand for training sample data while ensuring estimation accuracy and uncertainty quantification, and there is an impact of individual heterogeneity and noise.

Method used

Establish a linear model of capacitor attenuation time and capacitance value attenuation ratio, introduce a random error term, determine its distribution, calculate the probability distribution of attenuation time, and calculate the confidence interval of life based on the confidence level to provide the capacitor life probability distribution.

Benefits of technology

It reduces the complexity of the algorithm and the amount of computing, provides high-precision life estimation and uncertainty quantification, is suitable for online monitoring and offline testing, and has high industrial application value.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112818293B_ABST
    Figure CN112818293B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and system for estimating the probability distribution of capacitor life, belonging to the field of power electronics technology, the method comprises: establishing a linear model of capacitor decay time and capacitance decay ratio; introducing a random error term in the linear model, and determining the distribution of the random error term; based on the distribution of the random error term, obtaining the probability distribution of decay time corresponding to a certain capacitance decay ratio observation value; based on a given confidence level and the obtained probability distribution of the corresponding decay time, calculating the confidence interval of the individual and mean of the corresponding decay time estimation value; given the capacitance decay ratio when reaching the capacitor life limit, thereby estimating the probability distribution of the capacitor life. In this way, the present invention can not only obtain a relatively accurate life estimation value under small sample data, but also estimate the probability distribution of life, calculate the confidence interval of the life estimation value under different confidence levels, the algorithm is concise and easy to understand, the calculation amount is small, it is easy to implement, and has industrial application value.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of power electronics, and more specifically, relates to a method and system for estimating the probability distribution of capacitor life. Background Art

[0002] Capacitors are important components of power electronics and motor control systems, and their health is of great significance to the safe operation of the system. Traditional capacitor life estimation methods are based on additional hardware or signal injection methods, which are costly and complex. Data-driven algorithms can mine aging features from data and facilitate online monitoring or offline testing. They do not require additional hardware or signal injection, greatly reducing costs and complexity, and have good application prospects.

[0003] However, there are still some problems in data-driven algorithms that need to be promoted and improved. For example, heterogeneity between individuals, measurement noise, and changing operating environments will bring uncertainty to life estimation. In order to make better decisions in actual industrial applications, we hope that the life estimation algorithm can provide uncertainty quantification and indicate the probability distribution and confidence interval of the life estimate. Some feasible algorithms include random data drive, Gaussian process regression, and related vector machines, but these algorithms are highly complex, computationally intensive, or require sufficient sample data. If the demand for training sample data can be reduced, and the computational workload and algorithm complexity can be reduced while ensuring high estimation accuracy and providing uncertainty quantification, it is of great significance for estimating capacitor life using data-driven algorithms. Summary of the invention

[0004] In view of the defects of the prior art and the need for improvement, the present invention provides a method and system for estimating the probability distribution of capacitor life, which aims to solve the technical problems of high complexity and large amount of calculation in the prior art algorithms.

[0005] To achieve the above object, according to one aspect of the present invention, a method for estimating the probability distribution of capacitor life is provided, comprising the following steps:

[0006] S1. Establish a linear model of capacitor decay time and capacitance decay ratio;

[0007] S2. Introducing a random error term into the linear model and determining the distribution of the random error term;

[0008] S3. Based on the distribution of the random error term, a probability distribution of a certain capacitance attenuation ratio observation value corresponding to the attenuation time is obtained;

[0009] S4, based on a given confidence level and the probability distribution of the corresponding decay time obtained in S3, calculating the confidence interval of the individual and mean values ​​of the corresponding decay time estimate;

[0010] S5. Given the capacitance attenuation ratio when the capacitor reaches its life limit, the probability distribution of the capacitor life can be estimated.

[0011] Furthermore, the step S1 comprises:

[0012] S11, drawing a scatter plot of average normalized capacitance data changing over time to obtain capacitance attenuation characteristics;

[0013] S12, establishing a function model of capacitor decay time and capacitance decay ratio according to the decay characteristics, and converting it into a linear model;

[0014] S13, defining the input matrix, output matrix and parameter matrix of the linear model, performing regression using the least squares method, and obtaining the linear model parameters by minimizing the loss function;

[0015] S14, using residual sum of squares RSS, regression sum of squares ESS, and adjusted determination coefficient The linear model is evaluated by the root mean square error RMSE; wherein, m is the number of explanatory variables and n is the number of observations.

[0016] Furthermore, the step S2 comprises:

[0017] S21, introducing a random error term into the linear model;

[0018] S22, using a QQ graph to test and determine the distribution of the random error term;

[0019] S23. Calculate unbiased estimates of the mean and variance of the probability distribution of the random error term.

[0020] Furthermore, the step S3 comprises:

[0021] S31, based on the unbiased estimates of the mean and variance of the probability distribution of the random error term obtained in S23, derive the probability distribution of the attenuation time corresponding to a certain capacitance attenuation ratio observation value;

[0022] S32. Obtain a probability density function and a probability distribution function of the corresponding decay time from the probability distribution of the corresponding decay time.

[0023] Furthermore, the step S4 comprises:

[0024] S41, calculating the estimated value and mean value of the corresponding decay time;

[0025] S42, calculating the error between the estimated value and the true value of the corresponding decay time, combining the given confidence level, and the probability density function and probability distribution function of the corresponding decay time obtained in S32, to obtain the confidence interval of the individual estimated value of the corresponding decay time;

[0026] S43. Calculate the error between the estimated value and the mean of the corresponding decay time, and obtain the confidence interval of the mean of the estimated value of the corresponding decay time by combining the given confidence level and the probability density function and probability distribution function of the corresponding decay time obtained in S32.

[0027] Furthermore, the step S5 comprises:

[0028] S51. Given the capacitance attenuation ratio when the capacitor reaches its life limit, calculate the estimated value and confidence interval of the capacitor life;

[0029] S52, and calculate the B10 life and B63.2 life according to the probability density function of the capacitor life.

[0030] Furthermore, in step S3,

[0031] The corresponding decay time is:

[0032] l * =I0B+ε

[0033] Where, I0 is the given capacitance attenuation ratio observation value, B is the parameter matrix; ε is the random error term, and satisfies the normal distribution with mean 0 and homoscedasticity;

[0034] The probability distribution of the corresponding decay time satisfies:

[0035]

[0036] in, is the variance of the random error term ε;

[0037] The probability density function and distribution function corresponding to the decay time are:

[0038]

[0039]

[0040] in, is the estimated value of the variance of the random error term ε.

[0041] Furthermore, in step S4,

[0042] The estimated values ​​and means of the corresponding decay times are:

[0043]

[0044] E(l|I0)=I0B

[0045] The error between the estimated decay time and the true value follows a normal distribution with mean 0 and different variances:

[0046]

[0047] The error between the decay time estimate and the mean follows a normal distribution with mean 0 and different variances:

[0048]

[0049] Given a confidence level α, the confidence interval of the corresponding decay time estimate is obtained as:

[0050]

[0051] The confidence interval of the mean of the corresponding decay time estimate is obtained as:

[0052]

[0053] According to another aspect of the present invention, there is provided a system for estimating a probability distribution of capacitor life, comprising:

[0054] A linear model building module, used to build a linear model of capacitor decay time and capacitance decay ratio;

[0055] A random error term introduction module, used for introducing a random error term into the linear model and determining the distribution of the random error term;

[0056] A probability distribution acquisition module, used to obtain a probability distribution of a certain capacitance attenuation ratio corresponding to an attenuation time based on the distribution of the random error term;

[0057] A confidence interval calculation module, used to calculate the confidence interval of the individual and mean estimated values ​​of the attenuation time corresponding to the certain capacitance attenuation ratio based on a given confidence level and the probability distribution of the attenuation time obtained by executing the probability distribution acquisition module;

[0058] The life estimation module is used to estimate the probability distribution of capacitor life given the capacitance attenuation ratio when the capacitor reaches its life limit.

[0059] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects:

[0060] (1) The capacitor life estimation method proposed in the present invention is based on data and software algorithms, does not require additional hardware equipment and signal injection, and reduces cost and complexity;

[0061] (2) The software algorithm proposed in the present invention can not only estimate the life of the capacitor, but also provide the confidence interval of the life estimate at different confidence levels, as well as the probability density function and distribution function of the life, which is in line with the research trend of uncertainty quantification in the field of life estimation;

[0062] (3) The software algorithm proposed in the present invention is concise and easy to understand, requires a small amount of data and calculation, is easy to implement, can be conveniently applied to online estimation or offline testing, and can ensure high estimation accuracy, and has high industrial application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A flow chart of a method for estimating a capacitor life probability distribution provided by the present invention;

[0064] Figure 2 A relationship diagram between RSS and ESS and the linear regression model provided by the present invention;

[0065] Figure 3 A scatter plot of the average normalized capacitance data provided by the present invention changing with time;

[0066] Figure 4 A QQ diagram of the test residual distribution provided by the present invention;

[0067] Figure 5 A comparison chart of the lifespan probability distribution estimated by the method provided by the present invention and the median rank method. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0069] The present invention provides a method for estimating the probability distribution of capacitor life, such as Figure 1 As shown, including:

[0070] S1. Establish a linear model of capacitor decay time and capacitance decay ratio;

[0071] Specifically, step S1 includes:

[0072] S11, drawing a scatter plot of average normalized capacitance data changing over time to obtain capacitance attenuation characteristics;

[0073] S12, establishing a function model of capacitor decay time and capacitance decay ratio according to the decay characteristics, and converting it into a linear model;

[0074] S13, defining the input matrix, output matrix and parameter matrix of the linear model, performing regression using the least squares method, and obtaining the linear model parameters by minimizing the loss function;

[0075] S14, using residual sum of squares RSS, regression sum of squares ESS, and adjusted determination coefficient The linear model is evaluated by the root mean square error RMSE; the relationship between residual error, regression bias and linear regression model is as follows Figure 2 shown.

[0076] in, m is the number of explanatory variables, n is the number of observations, usually The larger the value, the stronger the linear relationship between the attenuation time and the capacitance attenuation ratio.

[0077] In addition, RMSE is defined as the square root of the ratio of the square of the deviation between the estimated value and the true value to the number of observations n. The smaller the RMSE, the higher the model estimation accuracy.

[0078] S2. Introducing a random error term into the linear model and determining the distribution of the random error term;

[0079] Specifically, step S2 includes:

[0080] S21, introducing a random error term into the linear model;

[0081] S22, using a QQ graph to test and determine the distribution of the random error term;

[0082] S23. Calculate unbiased estimates of the mean and variance of the probability distribution of the random error term.

[0083] S3. Based on the distribution of the random error term, a probability distribution of a certain capacitance attenuation ratio observation value corresponding to the attenuation time is obtained;

[0084] Specifically, step S3 includes:

[0085] S31, based on the unbiased estimates of the mean and variance of the probability distribution of the random error term obtained in S23, derive the probability distribution of the attenuation time corresponding to a certain capacitance attenuation ratio observation value;

[0086] S32. Obtain a probability density function and a probability distribution function of the corresponding decay time from the probability distribution of the corresponding decay time.

[0087] S4, based on a given confidence level and the probability distribution of the corresponding decay time obtained in S3, calculating the confidence interval of the individual and mean values ​​of the corresponding decay time estimate;

[0088] Specifically, step S4 includes:

[0089] S41, calculating the estimated value and mean value of the corresponding decay time;

[0090] S42, calculating the error between the estimated value and the true value of the corresponding decay time, combining the given confidence level, and the probability density function and probability distribution function of the corresponding decay time obtained in S32, to obtain the confidence interval of the individual estimated value of the corresponding decay time;

[0091] S43. Calculate the error between the estimated value and the mean of the corresponding decay time, and obtain the confidence interval of the mean of the estimated value of the corresponding decay time by combining the given confidence level and the probability density function and probability distribution function of the corresponding decay time obtained in S32.

[0092] It should be noted that if the random error term obeys the normal distribution, for a given capacitance attenuation ratio observation, the corresponding attenuation time obeys the normal distribution with the capacitance attenuation ratio observation as the mean and the variance of the random error term as the variance, so that the error between the estimated attenuation time and the true value and the mean also satisfies the normal distribution.

[0093] S5. Given the capacitance attenuation ratio when the capacitor reaches its life limit, the probability distribution of the capacitor life can be estimated.

[0094] Specifically, step S5 includes:

[0095] S51. Given the capacitance attenuation ratio when the capacitor reaches its life limit, calculate the estimated value and confidence interval of the capacitor life;

[0096] S52, and calculate the B10 life and B63.2 life according to the probability density function of the capacitor life.

[0097] Another aspect of the present invention provides a system for estimating a probability distribution of capacitor life, comprising:

[0098] A linear model building module, used to build a linear model of capacitor decay time and capacitance decay ratio;

[0099] A random error term introduction module, used for introducing a random error term into the linear model and determining the distribution of the random error term;

[0100] A probability distribution acquisition module, used to obtain a probability distribution of a certain capacitance attenuation ratio corresponding to an attenuation time based on the distribution of the random error term;

[0101] A confidence interval calculation module, used to calculate the confidence interval of the individual and mean estimated values ​​of the attenuation time corresponding to the certain capacitance attenuation ratio based on a given confidence level and the probability distribution of the attenuation time obtained by executing the probability distribution acquisition module;

[0102] The life estimation module is used to estimate the probability distribution of capacitor life given the capacitance attenuation ratio when the capacitor reaches its life limit.

[0103] The division of each module in the above-mentioned system for estimating the probability distribution of capacitor life is only for illustration. In other embodiments, the system for estimating the probability distribution of capacitor life can be divided into different modules as needed to complete all or part of the functions of the above-mentioned system for estimating the probability distribution of capacitor life.

[0104] In order to clearly describe the method of the present invention, the following is described in detail with reference to the embodiments:

[0105] In this embodiment, the temperature and humidity of the climate box are set to 85°C and 85% respectively, and 10 1100V / 40μF metallized polypropylene film capacitor samples are subjected to accelerated life test, and the capacitance data of each film capacitor sample is recorded at 19 moments. The collected data is cleaned, and no abnormal data is found. All the collected data is used for modeling.

[0106] S1. Draw a scatter plot of the average normalized capacitance data of 10 samples over time, such as Figure 3 As shown, it can be seen that the capacitance decays exponentially, and the capacitance decay ratio at time γ is defined as:

[0107] F(Θ(γ))%=(Θ(0)-Θ(γ)) / Θ(0)·100%

[0108] Among them, Θ(0) is the initial value of the capacitance, Θ(γ) is the capacitance at the γ moment, and the function model of the attenuation time and capacitance attenuation ratio is further established according to the attenuation characteristics:

[0109] F(Θ(γ))%=λ+μe τγ

[0110] Among them, λ, μ, τ are constants. We know that when γ approaches 0, the attenuation ratio F(Θ(γ))% is close to 0. At the same time, for the convenience of derivation, the function model is further adjusted to:

[0111] F(θ(γ))%=e κ(γ-ν)

[0112] Among them, κ and ν are constants, and taking the logarithm of both sides of the model can be transformed into a linear model:

[0113] l γ =αu γ +β

[0114] Among them, u γ =ln(F(θ(γ))%),l γ =γ, α and β are constants.

[0115] For the jth observation data of the i-th sample, at the monitoring time γ i,j , l γ and u γ Respectively expressed as l i,j = l(γ i,j ) and u i,j =u(γ i,j ). Define the input matrix of the linear regression algorithm as I = (u, 1), where u = (u 1,1 ,...,u 1,19 ,u 2,1 ,...,u 2,19 ,...,u 10,1 ,...,u 10,19 ) T , 1 is a unit vector with the same dimension as u, and the weight of the input matrix, i.e. the estimated value of the parameter matrix, is defined as Target output matrix L = (l 1,1 ,...,l 1,19 ,l 2,1 ,...,l 2,19 ,...,l 10,1 ,...,l 10,19 ) T , use the least squares method for linear regression, and the training goal is to minimize the loss function:

[0116]

[0117] The estimated model parameters are:

[0118]

[0119] Given any capacitance attenuation ratio The decay time is estimated based on the obtained model:

[0120]

[0121] The calculated RSS and ESS values ​​are 52954793.3 and 16559761.9 respectively. The ESS value is greater than the RSS, indicating that the RSS accounts for a small proportion of the total deviation sum of squares. 0.7605, which is close to 1, indicates that there is a strong linear relationship between the decay time and the capacitance decay ratio. The RMSE value is 296, indicating that the model has a high fitting accuracy.

[0122] S2. Rewrite the linear regression model of capacitor decay time and capacitance decay ratio, introducing a random error term, i.e., residual error

[0123] l=IB+ε

[0124] Among them, ε is the residual.

[0125] Make a QQ plot of the residual, such as Figure 4 As shown, it can be seen that the residuals are distributed near the 45-degree line, with a slight deviation at the tail, and it can be seen that the residuals obey a normal distribution with a mean of 0. In this embodiment, it is assumed that the residuals have homoscedasticity, that is,

[0126]

[0127] in is the variance of the residual, and RSS is expressed in matrix form

[0128] RSS=ε T (1 n -I(I T I) -1 I T )ε

[0129] 1 of them n is the n-order unit matrix, and the mean of RSS is

[0130] E(RSS)=σ ε 2 (nk-1)

[0131] This provides an unbiased estimate of the variance of the random error term.

[0132] S3. Given a set of capacitance attenuation ratio observations I0, the corresponding attenuation time is:

[0133] l * =I0B+ε

[0134] It is known that the random error term satisfies the normal distribution with mean 0 and homoscedasticity, so:

[0135]

[0136] Therefore, the probability density function and distribution function of the decay time are:

[0137]

[0138]

[0139] S4. Calculate the estimated value and mean of the decay time:

[0140]

[0141] E(l|I0)=I0B

[0142] The errors between the estimated decay time and the true value and mean are calculated respectively. The estimated errors follow normal distributions with mean 0 and different variances:

[0143]

[0144]

[0145] According to the error between the estimated decay time and the true value, the statistic is constructed:

[0146]

[0147] Given a confidence level α, we get the confidence interval for the individual estimates:

[0148]

[0149] Similarly, the statistics are constructed based on the error between the decay time estimate and the mean, and the confidence interval of the mean of the estimate is obtained as follows:

[0150]

[0151] S5. According to the above theory, an unbiased estimate of the variance of the random error term can be calculated as 87617.79, thereby obtaining the variance of the normal distribution obeyed by the estimated life span. In addition, taking the estimated life span as the mean, the specific probability density function and probability distribution function can be obtained.

[0152] The B10 life span and B63.2 life span were calculated based on the probability density function of life span, which are

[0153]

[0154]

[0155] in It is the capacitance attenuation ratio when a given life limit is reached.

[0156] The life limit of the film capacitor in this embodiment is defined as the capacitance value dropping to 5% of the initial value. Given the capacitance attenuation ratio when the life limit is reached, the life estimate, mean life confidence interval and individual life confidence interval can be calculated, as shown in Table 1. It can be seen that the higher the confidence level, the larger the interval range, and the individual life confidence interval is usually larger than the mean life confidence interval. The calculation results of B10 life and B63.2 life at different confidence levels are shown in Table 2.

[0157] Table 1

[0158]

[0159] Table 2

[0160] life Estimated value 90% Confidence Interval 95% Confidence Interval 99% Confidence Interval B10 1121 [1081,1162] [1073,1170] [1058,1185] B63.2 1600 [1560,1641] [1552,1649] [1537,1664]

[0161] The effectiveness of the proposed method is verified by combining it with traditional life estimation methods.

[0162] The life of the film capacitor in this embodiment is estimated by the traditional life analysis method based on Weibull distribution. The results are shown in Table 3. It can be seen that the life estimation value of the proposed method is slightly different from the result based on Weibull analysis, which proves the effectiveness of the proposed method.

[0163] Table 3

[0164] Estimated value Mean or individual life span 1519 B10 992 B63.2 1686

[0165] The probability density distribution obtained by the median rank method and the probability density distribution of the nominal life span, the 99% confidence interval of the mean life span and the 90% confidence interval of the individual life span estimated by the proposed method are plotted, as shown in Figure 5 As shown in the figure, it can be seen that the probability distribution values ​​estimated by the median-rank method are mostly distributed inside or on both sides of the probability distribution function of the 99% confidence interval of the mean life span, and the probability distribution of the 90% confidence interval of the individual life span is relatively wide, including all possible median-rank estimates. In addition, the probability density distribution estimated by the proposed method is basically consistent with the upward trend of the median-rank method estimate. The comparison once again verifies the effectiveness of the proposed method.

[0166] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for estimating the probability distribution of capacitor life, characterized in that: The following steps are involved: S1. Establish a linear model of capacitor decay time and capacitance decay ratio; S2. Introducing a random error term into the linear model and determining the distribution of the random error term; S3. Based on the distribution of the random error term, a probability distribution of a certain capacitance attenuation ratio observation value corresponding to the attenuation time is obtained; S4, based on a given confidence level and the probability distribution of the corresponding decay time obtained in S3, calculating the confidence interval of the individual and mean values ​​of the corresponding decay time estimate; S5, given the capacitance attenuation ratio when the capacitor reaches the life limit, thereby estimating the probability distribution of the capacitor life; Wherein, step S1 comprises: S11, drawing a scatter plot of average normalized capacitance data changing over time to obtain capacitance attenuation characteristics; S12, establishing a function model of capacitor decay time and capacitance decay ratio according to the decay characteristics, and converting it into a linear model; S13, defining the input matrix, output matrix and parameter matrix of the linear model, performing regression using the least squares method, and obtaining the linear model parameters by minimizing the loss function; S14, using residual sum of squares RSS, regression sum of squares ESS, and adjusted determination coefficient The linear model is evaluated by the root mean square error RMSE; wherein, m is the number of explanatory variables, n is the number of observations; Step S2 includes: S21, introducing a random error term into the linear model; S22, using a QQ graph to test and determine the distribution of the random error term; S23, calculating an unbiased estimate of the mean and variance of the probability distribution of the random error term; Step S3 includes: S31, based on the unbiased estimates of the mean and variance of the probability distribution of the random error term obtained in S23, derive the probability distribution of the attenuation time corresponding to a certain capacitance attenuation ratio observation value; S32, obtaining a probability density function and a probability distribution function of the corresponding decay time from the probability distribution of the corresponding decay time; Step S4 includes: S41, calculating the estimated value and mean value of the corresponding decay time; S42, calculating the error between the estimated value and the true value of the corresponding decay time, combining the given confidence level, and the probability density function and probability distribution function of the corresponding decay time obtained in S32, to obtain the confidence interval of the individual estimated value of the corresponding decay time; S43. Calculate the error between the estimated value and the mean of the corresponding decay time, and obtain the confidence interval of the mean of the estimated value of the corresponding decay time by combining the given confidence level and the probability density function and probability distribution function of the corresponding decay time obtained in S32.

2. The method according to claim 1, characterized in that Step S5 includes: S51. Given the capacitance attenuation ratio when the capacitor reaches its life limit, calculate the estimated value and confidence interval of the capacitor life; S52, and calculate the B10 life and B63.2 life according to the probability density function of the capacitor life.

3. The method according to claim 1, characterized in that In the step S3, The corresponding decay time is: l * =I0B+ε Where, I0 is the given capacitance attenuation ratio observation value, B is the parameter matrix; ε is the random error term, and satisfies the normal distribution with mean 0 and homoscedasticity; The probability distribution of the corresponding decay time satisfies: in, is the variance of the random error term ε; The probability density function and distribution function corresponding to the decay time are: in, is the estimated value of the variance of the random error term ε.

4. The method according to claim 3, characterized in that In the step S4, The estimated values ​​and means of the corresponding decay times are: E(l|I0)=I0B The error between the estimated decay time and the true value follows a normal distribution with mean 0 and different variances: The error between the decay time estimate and the mean follows a normal distribution with mean 0 and different variances: Given a confidence level α, the confidence interval of the corresponding decay time estimate is obtained as: The confidence interval of the mean of the corresponding decay time estimate is obtained as:

5. A system for estimating the probability distribution of capacitor life, characterized in that: The method for executing any one of claims 1 to 4 comprises: A linear model building module, used to build a linear model of capacitor decay time and capacitance decay ratio; A random error term introduction module, used for introducing a random error term into the linear model and determining the distribution of the random error term; A probability distribution acquisition module, used to obtain the probability distribution of the attenuation time corresponding to a certain capacitance attenuation ratio observation value based on the distribution of the random error term; A confidence interval calculation module, used to calculate the confidence interval of the individual and mean values ​​of the corresponding decay time estimation values ​​based on a given confidence level and the probability distribution of the corresponding decay time obtained by executing the probability distribution acquisition module; The life estimation module is used to estimate the probability distribution of capacitor life given the capacitance attenuation ratio when the capacitor reaches its life limit.

Citation Information

Cited By

  • Method and system for detecting residual life of electrolytic capacitor of frequency converter

    CN119596046A