A comprehensive failure class single machine fault probability calculation method

Abstract

The satellite integrated failure type single machine fault probability calculation method divides single machines into random failure parts and wear-out failure parts, the random failure parts need to calculate recoverable fault probability and resident fault probability, and the wear-out failure parts need to calculate resident fault probability.The recoverable fault probability calculation steps are as follows: for each single machine in each sample, the recoverable fault probability of the single machine is calculated; when more than one single machine in the same sample has occurred recoverable fault, the common cause factor is used to correct the recoverable fault probability of the single machine itself; when the single machine has occurred reconstruction, the recoverable fault probability of the single machine after reconstruction is calculated to update the recoverable fault probability of the single machine.The resident fault probability calculation steps are as follows: according to the total number of resident faults occurred in the sample, different resident fault probability calculation methods are selected, and the reconstruction condition is considered, and the resident fault probability is corrected by using the correction coefficient.

Description

Technical Field

[0001] This invention belongs to the field of satellite single-unit risk quantification assessment technology, and relates to a method for calculating the probability of single-unit failures under comprehensive failure categories. Background Technology

[0002] Accurate assessment of individual unit reliability is the foundation of overall satellite reliability assessment. Satellites consist of numerous and diverse individual units with varying failure mechanisms, and relatively limited on-orbit data exists. Currently, reliability assessment methods only exist for analyzing individual units that fail randomly or experience wear and tear. Due to the complex composition and functions of units prone to overall failure, there is no comprehensive assessment method to calculate the probability of failure for all individual units. Summary of the Invention

[0003] The technical problem solved by this invention is to overcome the shortcomings of the prior art and propose a method for calculating the probability of random failure of a single satellite unit, thereby improving the accuracy of satellite unit failure probability estimation.

[0004] The solution provided by this invention is a method for calculating single-machine faults of a comprehensive failure type, comprising the following steps:

[0005] S1. Extract the comprehensive failure category of all satellites. Group the satellites with the same batch, model, and supplier into the same sample. Based on this, further divide the existing samples. As of the current calculation time, group the satellites that have worked into the same sample and the satellites that have not worked into another sample.

[0006] S2. Divide each single machine into a random failure part and a wear-out failure part, and further divide the random failure part within the same sample into the same sample, and divide the wear-out failure part within the same sample into another sample.

[0007] S3. For the failure part within each sample, calculate the recoverable failure probability of the random failure part, which is used as the recoverable failure probability of the single machine in the comprehensive failure class.

[0008] S4. Obtain telemetry parameters that reflect the degradation trend of single-unit power consumption loss in the sample. The telemetry parameters that reflect the degradation trend of single-unit power consumption loss in the sample have the same change law. When the telemetry parameters increase monotonically or decrease monotonically with time, execute steps S5 and S6; otherwise, proceed to step S7.

[0009] S5. For each unit in the sample, calculate the reliability distribution of the random failure part and the wear-out failure part, and multiply the reliability distribution of the random failure part and the wear-out failure part to obtain the reliability distribution of the unit with the comprehensive failure class.

[0010] S6. Substitute the running time t0 of each single machine up to the current calculation time into the reliability distribution R(t) of the single machine in the comprehensive failure category to calculate the reliability R(t0) of the single machine in the comprehensive failure category up to the current calculation time. Based on the reliability R(t0) of the single machine in the comprehensive failure category, calculate the resident failure probability of the single machine in the comprehensive failure category up to the current calculation time.

[0011] S7. For each random failure part and wear-out failure part of a single machine in the sample, calculate the resident failure probability separately. Sum the resident failure probabilities of the random failure part and wear-out failure part of the single machine to obtain the resident failure probability of the single machine in the comprehensive failure category.

[0012] Preferably, the reliability distribution calculation steps for each individual machine random failure component are as follows:

[0013] Based on the total number of resident failures occurring in the sample containing the randomly failed portion of a single machine up to the current calculation time, different methods for calculating the probability of resident failures are selected to calculate the reliability distribution R of the randomly failed portion of a single machine within the sample. 随机 (t);

[0014] Preferably, when the total number of resident failures of the random failure components of all individual machines in the sample up to the current calculation time is 0, the reliability distribution R of the random failure components of the individual machines is calculated using a failure-free small sample analysis model. 随机 (t);

[0015] When the total number of resident failures of the random failure components of all individual machines in the sample up to the current calculation time t0 is 1-2, the reliability distribution R of the random failure components of the individual machines is calculated using the time-truncated model. 随机 (t);

[0016] If the cumulative number of resident failures of all individual machines in the sample at the current calculation time is greater than or equal to 3, the reliability distribution R of the random failure portion of the individual machines is calculated using a nonparametric estimation model. 随机 (t).

[0017] Preferably, in step S5, the reliability distribution R of the single-machine loss failure part is calculated using a log-normal model. 耗损 (t).

[0018] Preferably, in step S7, a stress intensity model is used to calculate the resident failure probability value of the single-unit wear-loss failure part. The formula for the resident failure probability of a single unit with comprehensive failure is as follows:

[0019] The probability of a persistent failure in a single machine under a comprehensive failure category =

[0020] (1-R 随机 (t))+(failure probability value calculated by stress intensity model).

[0021] Preferably, in step S5, the resident failure probability formula for a single machine in the single-machine comprehensive failure category is as follows:

[0022] The probability of a persistent failure in a single unit under a comprehensive failure category is 1 - R. 随机 (t)*R 耗损 (t).

[0023] Preferably, the corrected resident failure probability of a single machine is: λ1 * resident failure probability of a single machine with comprehensive failure, where λ1 is the correction coefficient.

[0024] Preferably, when the single machine is not reconfigured, the correction coefficient λ1 = 1; when the reconfiguration can effectively solve or improve the working state of the single machine in orbit, the correction coefficient λ1 = 0.5 to 1; when the reconfiguration scheme causes the working condition of the specified single machine to deteriorate or the load to increase, the correction coefficient λ1 = 1 to 1.5.

[0025] Preferably, the calculation steps for recoverable faults of random failures of each individual machine within the same random sample are as follows:

[0026] S9.1 For the random failure portion of each individual machine in each sample, calculate the recoverable failure probability of the random failure portion of the individual machine.

[0027] S9.2 When more than one single-machine random failure part in the same sample has experienced a recoverable failure as of the current calculation time, the probability of recoverable failure of the single-machine random failure part is corrected by using the common factor.

[0028] S9.3 If the random failure part of the single machine has been reconstructed up to the current calculation time, calculate the recoverable failure probability of the random failure part of the single machine after reconstruction, so as to update the recoverable failure probability of the random failure part of the single machine.

[0029] Preferably, the recoverable fault probability of a randomly failed single-machine component is calculated using the following formula:

[0030]

[0031] In the formula, MTBF is the mean time between failures (MTBF), and MTTR is the mean time to repair (MTTR).

[0032] Preferably, the method for correcting the recoverable failure probability of a single machine's random failure using a common factor is as follows:

[0033] The corrected probability of recoverable faults in random single-machine failures =

[0034] Before the correction, the probability of a single-machine random failure being self-recoverable was increased by β.

[0035] *The sum of the recoverable failure probabilities of other randomly failed parts within the sample to which the randomly failed part belongs.

[0036] Where β is a common factor.

[0037] Preferably, considering the reconfiguration measures for a single machine, the recoverable fault probability of a randomly failed portion of the single machine is updated as follows:

[0038]

[0039] In the formula, MTBF 重构后 MTTR is the mean time between failures (MTTR) of the randomly failing portion after refactoring. 重构后 This represents the average repair time for the randomly failed parts after reconstruction.

[0040] The advantages of this invention compared to the prior art are:

[0041] (1) This invention considers the different components of a single machine with integrated failure, uses different calculation methods to obtain the failure probability of each component, and integrates the calculation results through a certain method to finally obtain the failure probability of the single machine.

[0042] (2) When using telemetry parameters to calculate the probability of resident failure, calculations can be performed separately according to the characteristics of different telemetry data. In particular, when there are few telemetry parameters that cannot form a sample and there is no obvious degradation trend over time, the stress intensity model can be selected to calculate the probability of resident failure. Attached Figure Description

[0043] Figure 1 This is a process for calculating the probability of single-machine failures across a comprehensive failure category.

[0044] Figure 2 The specific process for calculating the probability of resident failure. Detailed Implementation

[0045] The present invention will be further described below with reference to the embodiments.

[0046] Define integrated failure type single machine, which refers to certain single machines that simultaneously have the characteristics of random failure and wear-out failure.

[0047] like Figure 1 As shown, this invention provides a method for calculating single-machine faults of a comprehensive failure type, the method comprising the following steps:

[0048] S1. Extract the comprehensive failure category of all satellites. Group the satellites with the same batch, model, and supplier into the same sample. Based on this, further divide the existing samples. As of the current calculation time, group the satellites that have worked into the same sample and the satellites that have not worked into another sample.

[0049] S2. Divide each single machine into a random failure part and a wear-out failure part, and further divide the random failure part within the same sample into the same sample, and divide the wear-out failure part within the same sample into another sample.

[0050] S3. For the failure part within each sample, calculate the recoverable failure probability of the random failure part, which is used as the recoverable failure probability of the single machine in the comprehensive failure class.

[0051] S4. Obtain telemetry parameters that reflect the degradation trend of single-unit power consumption loss in the sample. The telemetry parameters that reflect the degradation trend of single-unit power consumption loss in the sample have the same change law. When the telemetry parameters increase monotonically or decrease monotonically with time, execute steps S5 and S6; otherwise, proceed to step S7.

[0052] S5. For each individual machine in the sample, calculate the reliability distribution R for both the random failure portion and the wear-out failure portion. 耗损 (t), multiply the reliability distribution of the random failure part and the wear-out failure part of the single machine to obtain the reliability distribution of the single machine with comprehensive failure;

[0053] S6. Substitute the running time t0 of each single machine up to the current calculation time into the reliability distribution R(t) of the single machine in the comprehensive failure category to calculate the reliability R(t0) of the single machine in the comprehensive failure category up to the current calculation time. Based on the reliability R(t0) of the single machine in the comprehensive failure category, calculate the resident failure probability of the single machine in the comprehensive failure category up to the current calculation time.

[0054] S7. For each random failure part and wear-out failure part of a single machine in the sample, calculate the resident failure probability separately. Sum the resident failure probabilities of the random failure part and wear-out failure part of the single machine to obtain the resident failure probability of the single machine in the comprehensive failure category.

[0055] (1) Random failure part

[0056] 1) Calculation of recoverable failure probability for randomly failed components

[0057] The probability of recoverable failures is mainly calculated based on recoverable failures in quality problems and their corresponding interruption times.

[0058] For the random failure portion of each individual machine within each sample, calculate the recoverable failure probability of the random failure portion of the individual machine.

[0059] When more than one single-machine random failure part in the same sample has experienced a recoverable failure as of the current calculation time, the probability of recoverable failure of the single-machine random failure part is corrected by using the common factor;

[0060] If the random failure part of the single machine has been reconstructed up to the current calculation time, then calculate the recoverable failure probability of the random failure part of the single machine after reconstruction, so as to update the recoverable failure probability of the random failure part of the single machine.

[0061] The specific calculation method is as follows:

[0062] A. Based on the recoverable fault data of the random failure parts of individual units in the comprehensive failure category, the availability of each random failure part during on-orbit operation is calculated, and this is used as the probability of the random failure part itself being recoverable. Therefore, the formula for calculating the probability of the random failure part itself being recoverable in the factory condition is as follows:

[0063]

[0064] In the formula, MTBF is the mean time between failures (MTBF), and MTTR is the mean time to repair (MTTR).

[0065] B. Considering the recoverable failure probability correction during software refactoring

[0066] If a recoverable failure occurs again after the randomly failed part is rebuilt, then its recoverable failure probability needs to be adjusted. The recoverable failure probability of the random failure part on a single machine is updated as follows:

[0067]

[0068] In the formula, MTBF 重构后 MTTR is the mean time between failures (MTTR) after refactoring of the randomly failed components. 重构后 The mean repair time is the time taken to restore the randomly failed parts after reconstruction.

[0069] If no quality issues occur in the randomly failed portion after reconstruction, the recoverable fault probability is adjusted to 0 (MTTR). 重构后 (0).

[0070] C. The recoverable failure probability considering random failures caused by common factors.

[0071] Within the same sample, random failures with the same structure and working environment often share common causes. Therefore, in addition to considering the recoverable failure probability of random failures, we must also take into account the common causes of failures and use common factors to make corrections based on the segmented sample results.

[0072] The method for correcting the self-recoverable failure probability of random failures using common factor is as follows:

[0073] The corrected probability of recovering from random failures =

[0074] Before correction, the probability of a randomly failed part being self-recoverable was increased by β.

[0075] *The sum of the recoverable failure probabilities of other randomly failed parts within the sample to which the randomly failed part belongs.

[0076] Where β is the common factor. In practical applications, expert experience suggests that the value of the common factor β is generally between 0.001 and 0.10.

[0077] Preferably, the type of single machine and sample situation to which the random failure part belongs can be further considered, and the value of β is as follows.

[0078]

[0079] Similarly, by comprehensively considering the common cause failures, the recoverable failure probability of the random failures can be substituted into the recoverable failure probability of the random failures before correction, and the recoverable failure probability after considering the common cause can be calculated.

[0080] 2) Calculation of the resident failure probability of the random failure component

[0081] The probability of resident failures in the random failure component is mainly calculated based on resident failures in quality issues, combined with on-orbit baseline and supplier data. Due to the varying number of actual resident failures within different samples, different analysis methods are employed based on the sample division. The on-orbit baseline refers to the satellite's current on-orbit operational status at the calculation time.

[0082] like Figure 2 As shown, the calculation steps for the resident failure probability of each individual random failure component are as follows:

[0083] Based on the total number of resident failures occurring in the sample containing the randomly failed portion of a single machine up to the current calculation time, different methods for calculating the probability of resident failures are selected to calculate the reliability distribution R of the randomly failed portion of a single machine within the sample. 随机 (t);

[0084] When the total number of resident failures of the random failure components of all individual machines in the sample at the current calculation time is 0, the reliability distribution R of the random failure components of the individual machines is calculated using the failure-free small sample analysis model. 随机 (t);

[0085] When the total number of resident failures of the random failure components of all individual machines in the sample up to the current calculation time is 1-2, the reliability distribution R of the random failure components of the individual machines is calculated using a time-truncated model. 随机 (t);

[0086] If the cumulative number of resident failures of all individual machines in the sample at the current calculation time is greater than or equal to 3, the reliability distribution R of the random failure portion of the individual machines is calculated using a nonparametric estimation model. 随机 (t).

[0087] A. Calculating the probability of resident failure using a failure-free data analysis model.

[0088] If all the randomly failed parts of the individual machines in the sample have not experienced any persistent failures since they started running, they can be regarded as a small sample without failures. The structure of the data without failures is shown in the table below.

[0089] Table 1. Data Structure of Failure-Free Small Samples

[0090] Group number i Cutoff time T The number of samples in the i-th group is n Number of failures S 1 <![CDATA[T1]]> <![CDATA[n1]]> <![CDATA[S1]]> 2 <![CDATA[T2]]> <![CDATA[n2]]> <![CDATA[S2]]> 3 … … … 4 T n S

[0091] Based on the definition of the multilevel Bayesian method and Bayes' theorem, under squared loss, the multilevel failure probability p of the i-th sample group is... ihB The calculation is as follows:

[0092]

[0093] In the formula, s i Let c be the number of non-failures in the i-th sample group, and c be the recommended value of the hyperparameter (engineering) of 5. The failure probability p of the i-th sample group can be obtained from the above formula. ihB Estimated value.

[0094] According to the weighted least squares estimation method, the Weibull two-parameter estimates β and η are obtained as follows:

[0095]

[0096] η = expμ

[0097] In the formula:

[0098]

[0099] In the formula:

[0100]

[0101] In the formula ω i For weight values, T i Let k be the cutoff time for the i-th group, k be the number of groups, and n be the number of groups. i Let be the number of samples in the i-th group.

[0102] x i and y i The formula is as follows:

[0103]

[0104] In the formula p ihB Let T be the failure probability of the i-th sample group. i Let be the truncation time for the i-th group.

[0105] Substituting the obtained Weibull parameter point estimates into the reliability formula yields the reliability R of the original sample following the Weibull sampling function. 随机- No failures (t), as shown below:

[0106]

[0107] Substituting the runtime t0 of the single machine up to the current calculation moment, we can obtain the resident failure probability of the single machine in this sample as 1-R. 随机- No failure (t0).

[0108] B. Calculating the probability of resident failure using a timed truncation model.

[0109] For the random failure sample that has experienced less than three but more than zero resident failures, the reliability distribution parameters can be calculated using the timed truncation test method, and the resident failure probability can be further obtained.

[0110] According to the reliability index model R = e -λt The parameter θ is the mean failure time, which is the reciprocal of the failure rate λ, as shown in the following formula:

[0111]

[0112] In the formula, T r R is the total runtime of all individual machines in the sample, and r is the total number of times a persistent fault occurs in the sample.

[0113] Combining the reliability function, as follows:

[0114]

[0115] Substituting the runtime t0 of the single machine up to the current calculation moment into the above formula, we can obtain the resident failure probability of the random failure portion within the sample that has not experienced a resident failure as 1-R. 随机- Timed truncation (t0).

[0116] C. Calculating the resident failure probability using a nonparametric estimation model.

[0117] For the random failure sample that has experienced more than three resident failures, a non-parametric estimation method can be used to calculate the reliability distribution parameters and further obtain the resident failure probability.

[0118] The total sample size is N, with n failed (abnormal) samples and Nn truncated samples. Using the failure / abnormality information of the samples, calculate the reliability of 1, 2, ..., n failed (abnormal) samples.

[0119] The reliability of a unit that fails is:

[0120] The reliability of two failed units is:

[0121] …

[0122] The reliability of n failed units is:

[0123] The reliability probability of the random failure component estimated by nonparametric methods is expressed in terms of operating time (days). The curve is fitted using Weibull data to calculate the shape and lifetime parameters of the Weibull distribution. The results of the nonparametric analysis of the samples are logarithmed twice, and then fitted using minimum variance.

[0124] By fitting the data, we obtain the slope of the straight line as 'a' and the intercept as 'b', thus yielding the two parameter values ​​of the Weibull distribution:

[0125] β′=a

[0126]

[0127] The reliability calculation formula is as follows:

[0128]

[0129] Substituting the runtime t0 of a single machine up to the current calculation moment into the above formula, we can obtain the probability of a resident failure in the random failure portion of the sample that has not experienced a resident failure as 1-R. 随机- Nonparametric estimation (t0).

[0130] (2) Consumption loss and failure part

[0131] The probability of resident failure of the wear-out and loss-prone parts is mainly calculated based on the telemetry data of the single unit and the corresponding threshold.

[0132] 1) Log-normal model

[0133] When telemetry parameters exhibit a monotonically increasing or decreasing trend over time, a log-normal model can be used for calculation.

[0134] Using the log-normal model, the reliability distribution R of the single-machine loss failure component is calculated. 耗损 (t), the specific steps are as follows:

[0135] A telemetry parameter reflecting the degradation trend of single-unit power consumption loss within the sample is selected as a candidate telemetry parameter. The measured value y of the candidate telemetry parameter is then used in a log-normal model. i and its corresponding single-machine runtime t i For fitting, i = 1 to n, the fitting formula for the candidate telemetry parameters is as follows:

[0136]

[0137] Where b0 and b1 are the fitting parameters, The error is a random variable that follows a standard normal distribution; n is the number of telemetry parameters in the sample.

[0138] Substitute the candidate telemetry parameter boundary values ​​for each individual unit into the candidate telemetry parameter fitting formula to calculate the unit's operating time corresponding to the boundary value. This unit's operating time is defined as the lifetime (Age) of the consumed and failed portion of the corresponding unit. i , i = 1 to n;

[0139] Based on the lifespan Age of the worn-out and failed components of all individual machines in the sample. i The mean μ′ and standard deviation σ′ of the single-unit wear-loss failure life within the sample were calculated using the following formula:

[0140]

[0141] This leads to the reliability model:

[0142]

[0143] The reliability model for the mean μ′ and standard deviation σ′ of the lifespan of the failed portion of a single sample unit is as follows:

[0144]

[0145] 2) Stress intensity model

[0146] For telemetry parameters that do not show a clear degradation trend over time (no monotonically increasing or decreasing trend), a stress intensity model can be used to calculate the probability of resident failure.

[0147] Select a telemetry parameter that reflects the degradation trend of single-unit power consumption loss within the sample as a candidate telemetry parameter;

[0148] Obtain the sampling sequence of candidate telemetry parameters for a single unit;

[0149] Calculate the mean of the candidate telemetry parameter sampling sequence and standard deviation S x :

[0150]

[0151] Where M is the length of the candidate telemetry parameter sampling sequence, x m This is the m-th sampled value in the candidate telemetry parameter sampling sequence.

[0152] If a design lower limit exists for the candidate telemetry parameters, then the first coefficient K1 is calculated based on the design lower limit L:

[0153]

[0154] Based on the first coefficient K1, the given confidence level γ, and the length M of the candidate telemetry parameter sampling sequence, the shortest interval [a1, b1] containing K1 and the corresponding p are obtained by looking up the bilateral tolerance coefficient table of normal distribution QJ1384-88. (11) ,p (12) Then consult the normal distribution quantile table GB4086 to find the value corresponding to p. (11) Corresponding normal distribution quantiles With p (12) Corresponding normal distribution quantiles

[0155] Using linear interpolation, we obtain From the normal distribution table, we obtain the formula for p1:

[0156]

[0157] In the formula, the NormalDist function is used to calculate the cumulative distribution value of the normal distribution;

[0158] If there is a design upper limit for the candidate telemetry parameters, then calculate the second coefficient K2 based on the design upper limit U:

[0159]

[0160] Based on the second coefficient K2, the given confidence level γ, and the sample sequence length M of the candidate telemetry parameters, the shortest interval [a2, b2] containing K2 and the corresponding p are obtained by referring to the table of bilateral tolerance coefficients of the normal distribution QJ1384-88. (21) ,p (22) Then consult the normal distribution quantile table GB4086 to find the value corresponding to p. (21) Corresponding normal distribution quantiles With p (22) Corresponding normal distribution quantiles

[0161] Using linear interpolation, we obtain The formula for p2 is obtained from the normal distribution table:

[0162]

[0163] In the formula, the NormalDist function is used to calculate the cumulative distribution value of the normal distribution;

[0164] If a candidate telemetry parameter has both a lower design limit and a higher design limit, then:

[0165] Reliability of wear-and-tear failure components: R 耗损 = 1 - (p1 + p2).

[0166] If the candidate telemetry parameters only have a design lower limit, then:

[0167] Reliability of wear-and-tear failure components: R 耗损 =1-p1.

[0168] If the candidate telemetry parameters only have a design upper limit, then:

[0169] Reliability of wear-and-tear failure components: R 耗损 =1-p2.

[0170] The probability of a resident failure in the wear-out and loss-related components is: 1-R 耗损 .

[0171] (3) Calculate the resident failure probability of a single machine.

[0172] The recoverable failure probability of a single machine in the comprehensive failure category is obtained from the recoverable failure probability of the random component.

[0173] When the telemetry parameters do not exhibit a monotonically increasing or decreasing relationship with time, the wear and tear failure component adopts a stress intensity model. The formula for the resident failure probability of a single unit in the single-unit comprehensive failure category is as follows:

[0174] Resident failure probability of a single machine in the comprehensive failure category

[0175] =(1-R) 随机 (t))+(failure probability value calculated by stress intensity model)

[0176] When telemetry parameters monotonically increase or decrease with time, the wear and tear failure part adopts a log-normal model. The formula for the resident failure probability of a single unit in the single-unit comprehensive failure category is as follows:

[0177] The probability of a persistent failure in a single unit under a comprehensive failure category is 1 - R. 随机 (t)*R 耗损 (t).

[0178] Considering the impact of single-machine reconfiguration, a correction coefficient λ1 is added to correct the resident failure probability of single machines in the comprehensive failure category. The corrected resident failure probability of a single machine is: λ1 * resident failure probability of single machines in the comprehensive failure category.

[0179] Wherein, λ1 is the correction coefficient. When the single machine does not undergo reconfiguration, the correction coefficient λ1 = 1; when the reconfiguration can effectively solve or improve the working status of the single machine in orbit, the correction coefficient λ1 = 0.5 to 1; when the reconfiguration scheme causes the working condition of the specified single machine to deteriorate or the load to increase, the correction coefficient λ1 = 1 to 1.5.

[0180] Example of stress intensity model calculation

[0181] Calculate the sample mean of telemetry parameters and standard deviation S x get:

[0182]

[0183] With γ = 0.95 selected, when n = 70, the range of K in the table of two-sided tolerance coefficients of normal distribution QJ1384-88 is 1.36499-5.09007. Therefore, when the value of K1 exceeds this range, the reliability can be approximated as 0.99998.

[0184] The upper and lower limits U and L are designed based on the telemetry threshold. Calculate the first coefficient K1 and the second coefficient K2:

[0185]

[0186] We obtain the shortest interval [a1, b1] containing K1 and the corresponding p for a1 and b1. (11) ,p (12) Then consult the normal distribution quantile table GB4086 to find the value corresponding to p. (11) Corresponding normal distribution quantiles With p (12) Corresponding normal distribution quantiles

[0187] Using linear interpolation, we obtain From the normal distribution table, we obtain the formula for p1:

[0188]

[0189] The lower confidence limit for reliability is obtained as: R 耗损 =1-p1=0.998197268.

[0190] The probability of resident failure is: 1-R 耗损 =0.001803.

[0191] Assuming this single machine only has wear-and-tear failures, considering that refactoring can improve the machine's operating state, a correction coefficient λ1 = 0.8 is selected, and then the corrected resident failure probability is obtained: λ1(1-R 耗损 = 0.001442.

[0192] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solutions of the present invention by utilizing the methods and techniques disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall fall within the protection scope of the technical solutions of the present invention.

Claims

1. A method for calculating single-machine faults of a comprehensive failure type, characterized in that... Includes the following steps: S1. Extract the comprehensive failure category of all satellites. Group the satellites with the same batch, model, and supplier into the same sample. Based on this, further divide the existing samples. As of the current calculation time, group the satellites that have worked into the same sample and the satellites that have not worked into another sample. S2. Divide each single machine into a random failure part and a wear-out failure part, and further divide the random failure part within the same sample into the same sample, and divide the wear-out failure part within the same sample into another sample. S3. For the failure part within each sample, calculate the recoverable failure probability of the random failure part, which is used as the recoverable failure probability of the single machine in the comprehensive failure class. S4. Obtain telemetry parameters that reflect the degradation trend of single-unit power consumption loss in the sample. The telemetry parameters that reflect the degradation trend of single-unit power consumption loss in the sample have the same change law. When the telemetry parameters increase monotonically or decrease monotonically with time, execute steps S5 and S6; otherwise, proceed to step S7. S5. For each unit in the sample, calculate the reliability distribution of the random failure part and the wear-out failure part, and multiply the reliability distribution of the random failure part and the wear-out failure part to obtain the reliability distribution of the unit with the comprehensive failure class. S6. Substitute the running time t0 of each single machine up to the current calculation time into the reliability distribution R(t) of the single machine in the comprehensive failure category to calculate the reliability R(t0) of the single machine in the comprehensive failure category up to the current calculation time. Based on the reliability R(t0) of the single machine in the comprehensive failure category, calculate the resident failure probability of the single machine in the comprehensive failure category up to the current calculation time. S7. For each random failure part and wear-out failure part of a single machine in the sample, calculate the resident failure probability separately. Sum the resident failure probabilities of the random failure part and wear-out failure part of the single machine to obtain the resident failure probability of the single machine in the comprehensive failure category.

2. The method for calculating single-machine faults of a comprehensive failure type according to claim 1, characterized in that, The reliability distribution calculation steps for each individual machine's random failure component are as follows: Based on the total number of resident failures occurring in the sample containing the randomly failed portion of a single machine up to the current calculation time, different methods for calculating the probability of resident failures are selected to calculate the reliability distribution R of the randomly failed portion of a single machine within the sample. 随机 (t).

3. The method for calculating a single-machine fault of a comprehensive failure type according to claim 2, characterized in that: When the total number of resident failures of the random failure components of all individual machines in the sample at the current calculation time is 0, the reliability distribution R of the random failure components of the individual machines is calculated using the failure-free small sample analysis model. 随机 (t); When the total number of resident failures of the random failure components of all individual machines in the sample up to the current calculation time is 1-2, the reliability distribution R of the random failure components of the individual machines is calculated using a time-truncated model. 随机 (t); If the cumulative number of resident failures of all individual machines in the sample at the current calculation time is greater than or equal to 3, the reliability distribution R of the random failure portion of the individual machines is calculated using a nonparametric estimation model. 随机 (t).

4. The method for calculating a single-machine fault of a comprehensive failure type according to claim 1, characterized in that: In step S5, the reliability distribution R of the single-machine power loss failure component is calculated using a log-normal model. 耗损 (t).

5. The method for calculating a single-machine fault of a comprehensive failure type according to claim 1, characterized in that: In step S7, a stress intensity model is used to calculate the resident failure probability value of the single-unit wear-loss failure part. The formula for the resident failure probability of a single unit in the single-unit comprehensive failure category is as follows: The probability of a persistent failure in a single machine under a comprehensive failure category = (1-R 随机 (t))+(failure probability value calculated by stress intensity model).

6. The method for calculating a single-machine fault of a comprehensive failure type according to claim 1, characterized in that: In step S5, the formula for the resident failure probability of a single machine in the single-machine comprehensive failure category is as follows: The probability of a persistent failure in a single unit under a comprehensive failure category is 1 - R. 随机 (t)*R 耗损 (t).

7. The method for calculating the probability of single-machine failure under comprehensive failure categories according to claim 1, characterized in that: The corrected resident failure probability of a single machine is: λ1 * resident failure probability of a single machine with comprehensive failures, where λ1 is the correction coefficient.

8. The method for calculating the probability of single-machine failure under comprehensive failure categories according to claim 7, characterized in that: When no reconfiguration occurs on a single machine, the correction factor λ1 = 1; when reconfiguration can effectively solve or improve the working condition of the on-orbit single machine, the correction factor λ1 = 0.5 to 1; when the reconfiguration scheme causes the working condition of the specified single machine to deteriorate or the load to increase, the correction factor λ1 = 1 to 1.

5.

9. The method for calculating single-machine faults of a comprehensive failure type according to claim 1, characterized in that, The calculation steps for the recoverable fault of the random failure portion of each individual machine within the same random sample are as follows: S9.1 For the random failure portion of each individual machine in each sample, calculate the recoverable failure probability of the random failure portion of the individual machine. S9.2 When more than one single-machine random failure part in the same sample has experienced a recoverable failure as of the current calculation time, the probability of recoverable failure of the single-machine random failure part is corrected by using the common factor. S9.3 If the random failure part of the single machine has been reconstructed up to the current calculation time, calculate the recoverable failure probability of the random failure part of the single machine after reconstruction, so as to update the recoverable failure probability of the random failure part of the single machine.

Images