Quality and reliability evaluation method and system based on small sample data

Abstract

The invention relates to the technical field of electronic information, in particular to a quality and reliability evaluation method and system based on small sample data, and the method comprises the steps: obtaining the reliability and quality evaluation data of an evaluation object; based on a maximum likelihood estimation method, calling a classical probability distribution life model to perform reliability evaluation on the evaluation object to obtain a first goodness of fit test result; based on a BAYES method, calling a multi-source information fusion reliability evaluation model to perform reliability evaluation on the evaluation object to obtain a second goodness-of-fit test result; comparing the first goodness-of-fit test result with the second goodness-of-fit test result, and selecting a probability distribution model and distribution parameters with high final goodness-of-fit as a reliability evaluation mathematical model; and extracting a reliability evaluation result of the reliability evaluation mathematical model, and carrying out quality analysis and evaluation on the quality condition of the evaluation object. By adopting the method provided by the invention, a model more suitable for small sample data can be compared, and meanwhile, the accuracy and reliability of analysis and evaluation are improved.

Description

Technical Field

[0001] This invention relates to the field of electronic information technology, and in particular to a method and system for quality and reliability evaluation based on small sample data. Background Technology

[0002] Currently, reliability assessment of avionics equipment mainly involves fitting a distribution model of failure time data and using classical probabilistic and statistical methods such as least squares method and maximum likelihood estimation to obtain characteristic parameters. However, classical probabilistic and statistical methods require a large amount of experimental data.

[0003] Despite decades of development, aviation equipment manufacturers have produced and maintained numerous batches and models of equipment, accumulating a wealth of diverse data. However, factors such as ineffective utilization of reliability information at various stages, low utilization of data from similar products, lack of dynamic links between data at different stages, and relatively independent reliability assessments at each stage mean that the reliability test data accumulated in the early stages of avionics equipment cannot be effectively used in current reliability assessments. Both engineering development information and test data are underutilized. Furthermore, due to the high development cost, high price, and limited number of prototypes, avionics equipment is very expensive to obtain reliability test data. As a result, very little reliability test data can be collected for avionics equipment in the same period of time, resulting in a small sample size. As is well known, the amount of data, i.e. the richness of reliability information, seriously affects the accuracy and precision of reliability assessment results.

[0004] Therefore, how to accurately assess the reliability level of equipment using existing data under conditions of small sample size is an urgent problem to be solved in equipment reliability work. Summary of the Invention

[0005] In view of this, the present invention provides a quality and reliability evaluation method and system based on small sample data, which aims to achieve quality and reliability assessment of multi-source information fusion across stages by using small samples, and further improve the accuracy and reliability of analysis and evaluation.

[0006] To address the above technical problems, the present invention provides a method for evaluating the quality and reliability of data based on small sample data, comprising: Obtain reliability and quality evaluation data for the assessment object; The reliability of the evaluation object is assessed by calling the classical probability distribution lifetime model based on the maximum likelihood estimation method, and the first goodness-of-fit test result is obtained. The reliability assessment of the assessment object is performed by calling the multi-source information fusion reliability assessment model based on the BAYES method, and the second goodness-of-fit test result is obtained. By comparing the results of the first goodness-of-fit test and the second goodness-of-fit test, the probability distribution model and distribution parameters with the highest final goodness-of-fit are selected as the mathematical model for reliability assessment. Extract the reliability assessment results from the reliability assessment mathematical model, and conduct a quality analysis and evaluation of the assessment object based on the reliability assessment results.

[0007] As one implementation method, the step of using the maximum likelihood estimation method to call the classical probability distribution lifetime model to perform a reliability assessment on the evaluation object and obtain the first goodness-of-fit test result includes: Obtain reliability evaluation data of the evaluated object within a preset time period; The classic probability distribution lifetime model is invoked, and its distribution parameters and goodness-of-fit test results are obtained by using the maximum likelihood estimation method. Based on the distribution parameters and the goodness-of-fit test results, the first goodness-of-fit test result and the first average fault interval time are obtained.

[0008] As one implementation method, the step of calling the classical probability distribution lifetime model and using the maximum likelihood estimation method to solve for its distribution parameters and goodness-of-fit test results includes: The probability distribution models are selected as exponential distribution and Welb distribution respectively, and the distribution parameters of the exponential distribution and the Welb distribution are solved by the maximum likelihood estimation method respectively. The goodness-of-fit test was performed on the distribution parameters of the exponential distribution and the Welb distribution using the KS method. Based on the results of the goodness-of-fit test, the probability distribution with a high goodness-of-fit test is determined as the probability distribution of the classical probability distribution lifetime model.

[0009] As one implementation method, the reliability assessment of the assessment object is performed by calling the multi-source information fusion reliability assessment model based on the BAYES method to obtain the second goodness-of-fit test result, including: The reliability and quality data in the reliability and quality evaluation database of the evaluated object are screened and cleaned. The reliability assessment model of multi-source information fusion is invoked, and its distribution parameters and goodness-of-fit test results are obtained by using the BAYES method. Based on the distribution parameters and the goodness-of-fit test results, the second goodness-of-fit test results and the second average fault interval time are obtained.

[0010] As one implementation method, the step of calling the multi-source information fusion reliability assessment model and using the BAYES method to solve for its distribution parameters and goodness-of-fit test results includes: The quality and reliability evaluation database storing the reliability data of the evaluated object is invoked, and similar products of the evaluated object and their weight scores are selected using the BWM method. Based on the similar products and their weight scores, the probability distribution models are selected as exponential distribution and Welb distribution, respectively, and the initial prior distribution of the evaluation object is obtained. Based on the prior information of the initial prior distribution, the posterior calculation formula of the evaluation object is obtained using the BAYES method; The non-analytical solution of the BAYES method is obtained using the MCMC method; The posterior mean of the non-analytical solution is used as the posterior reliability feature parameter for goodness-of-fit testing. Based on the results of the goodness-of-fit test, the probability distribution with a high goodness-of-fit is determined as the probability distribution of the multi-source information fusion reliability assessment model.

[0011] As one implementation method, the selection of similar products and their weighted scores of the evaluation object using the BWM method includes: The rank-sum test method is used to perform a compatibility test on the reliability data of the selected similar products, and similar product models that meet the requirements are screened out. Determine the weight score for each of the selected similar product models.

[0012] As one implementation method, obtaining the initial prior distribution of the evaluation object based on the similar products and their weight scores includes: The conjugate prior distribution is used as the prior distribution of the exponential distribution; Choose a normal or uniform distribution as the prior distribution of the Welb distribution; The prior distributions of the exponential distribution and the Welb distribution are weighted and fused based on the score weights of similar products to obtain the initial prior distribution of the evaluation object.

[0013] As one implementation method, the quality analysis and evaluation of the assessment object based on the reliability assessment results includes: Calculate the batch failure rate, annual zeroing rate, failure percentage of each component of the assessment object, failure percentage of each cause of failure, failure percentage of each geographical environment, fit the curve of failure number change over time, fit the change of failure number of each cause of failure over time, and obtain the statistical analysis results. The quality analysis and evaluation are conducted using the reliability assessment results as the core indicator and combined with the statistical analysis results.

[0014] As one implementation method, the quality analysis and evaluation based on the statistical analysis results specifically includes: By combining the statistical analysis results from different dimensions, the quality characteristics of the evaluation object are analyzed and evaluated. Analyze the quality control status based on changes in failure trends; Based on the distribution of failure types, analyze the key points for product design improvement; The ability to guarantee field use is analyzed based on maintenance efficiency.

[0015] Accordingly, the present invention also provides a quality and reliability evaluation system based on small sample data, applicable to any of the above-described quality and reliability evaluation methods based on small sample data, comprising: The data acquisition module is used to acquire reliability and quality evaluation data of the evaluation object; The first test result module is used to perform a reliability assessment of the evaluation object by calling the classical probability distribution lifetime model based on the maximum likelihood estimation method, and to obtain the first goodness-of-fit test result. The second test result module is used to call the multi-source information fusion reliability assessment model based on the BAYES method to perform a reliability assessment on the assessment object and obtain the second goodness-of-fit test result. The result comparison module is used to compare the first goodness-of-fit test result and the second goodness-of-fit test result, and select the probability distribution model and distribution parameters with high final goodness of fit as the mathematical model for reliability assessment. The evaluation and analysis module is used to extract the reliability evaluation results of the reliability evaluation mathematical model and to perform quality analysis and evaluation on the quality of the evaluated object based on the reliability evaluation results.

[0016] The primary improvement of this invention lies in: acquiring reliability and quality evaluation data of the assessment object; using the maximum likelihood estimation method to call the classical probability distribution lifetime model to perform reliability assessment on the assessment object, obtaining the first goodness-of-fit test result; using the BAYES method to call the multi-source information fusion reliability assessment model to perform reliability assessment on the assessment object, obtaining the second goodness-of-fit test result; comparing the first and second goodness-of-fit test results, selecting the probability distribution model and distribution parameters with the highest final goodness of fit as the reliability assessment mathematical model; extracting the reliability assessment results from the reliability assessment mathematical model, and performing quality analysis and evaluation on the quality of the assessment object. For data with small sample states, considering cross-stage and variable population conditions, calling the more suitable multi-source information fusion reliability assessment model not only makes the calculation process more convenient and faster, but also enables a more effective fit to the failure time distribution of the assessment object, significantly improving the fitting effect and comprehensively enhancing the accuracy and reliability of the reliability assessment. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the steps of a quality and reliability evaluation method based on small sample data provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a device reliability assessment process provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a quality and reliability evaluation system based on small sample data provided in an embodiment of the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the embodiments of the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Information is the foundation of management and the source of decision-making. Reliability data for avionics equipment is multi-type, multi-source, and multi-dimensional, including functional performance test data, reliability test data, production inspection test data, and field use and maintenance data. Managing multi-source data throughout the entire lifecycle of avionics equipment, based on a small sample size, is a crucial aspect of quality and reliability evaluation. Effective information and data are the foundation for quality management, reliability, maintainability, and supportability assessments, and serve as the basis for decision-making.

[0021] In view of this, one embodiment of this application provides as follows: Figure 1 The diagram shows the steps of a quality and reliability evaluation method based on small sample data. Figure 2 The schematic diagram of the equipment reliability assessment process shown below provides a detailed explanation of the evaluation method used in this embodiment.

[0022] S11. Obtain reliability and quality evaluation data of the evaluation object.

[0023] The scope and content of avionics equipment quality and reliability data include, but are not limited to: product identification data (such as factory code, product number, and system to which it belongs), product time data (such as failure time, cumulative operating time, etc.), product spatial data (such as mounting platform, mission profile, geographical environment, stress at the time of failure, and other usage information), and product status data (such as failure phenomenon and failure source, etc.). All relevant data for the assessment object requiring reliability evaluation should be obtained.

[0024] S12. Based on the maximum likelihood estimation method, the classical probability distribution lifetime model is called to conduct a reliability assessment of the evaluation object, and the first goodness-of-fit test result is obtained.

[0025] The basic idea of ​​maximum likelihood estimation (MLE) is that since the sample comes from the population, it can reflect the characteristics of the population to a certain extent. If the observed values ​​of the sample are obtained in a single experiment... , ,… Therefore, it can be said that since this event occurred in one trial, the probability of this event occurring is high. Thus, if the population parameter to be estimated is... It can take many values, because we don't know its truth value in all of us. From the possible values, select the one that maximizes the probability of the sample observation result as the value. The estimated value is denoted as This is the maximum likelihood estimate.

[0026] Furthermore, reliability evaluation data of the evaluation object within a preset time period is obtained; a reliability evaluation mathematical model is determined, and its distribution parameters and goodness-of-fit test results are solved using the maximum likelihood estimation method; based on the distribution parameters and goodness-of-fit test results, a first goodness-of-fit test result and a first mean time between failures are obtained.

[0027] Furthermore, regarding how to determine the mathematical model for reliability assessment and use the maximum likelihood estimation method to solve for its distribution parameters and goodness-of-fit test results, this embodiment selects the exponential distribution as the probability distribution model. and Welb distribution Position parameters are usually taken. If the value is 0, the distribution parameters of the exponential distribution are solved using the maximum likelihood estimation method. and the distribution parameters of the Wilbur distribution. , The goodness-of-fit test was performed on the distribution parameters of the exponential distribution and the Welb distribution using the KS method. Based on the results of the goodness-of-fit test, the probability distribution with a high goodness-of-fit test was determined as the mathematical model for reliability assessment.

[0028] Furthermore, based on the above, assuming the data originates from the exponential distribution and the Welb distribution respectively, the maximum likelihood estimation functions corresponding to the quantitative model under the random truncation case are as follows: Exponential distribution likelihood function: , ,get Point estimation: , .

[0029] Welb distribution: Where r is the number of faulty samples at the end of the statistical period, s is the number of fault-free samples at the end of the statistical period, and n = r + s is the total number of statistical samples at the end of the statistical period. , which is the cumulative statistical time of all samples within the random truncated statistical interval.

[0030] Furthermore, the KS method is used to perform goodness-of-fit tests on the distribution parameters of the exponential distribution and the Welb distribution, respectively, including: For the exponential distribution test: (1) F test: n samples are drawn from the population for a truncated test. r failure samples occur in the test, and the failure time is t 1 ,t 2 ,t 3 ,...,t r ,remember Then the F-test statistic It can be represented as: (2) Test: A truncated experiment is conducted by drawing n samples from the population. r faulty samples occur during the experiment, and the fault time is... t 1 ,t 2 ,t 3 ,...,t r ,remember ,but The test statistic can be expressed as: .

[0031] For the Welb distribution test: (1) F test: n samples are drawn from the population for a truncated test. r failure samples occur in the test, and the failure time is t 1 ,t 2 ,t 3 ,...,t r ,remember , , , ,and Then the F-test statistic It can be represented as: (2) Test: A truncated experiment is conducted by drawing n samples from the population. r faulty samples occur during the experiment, and the fault time is... t 1 ,t 2 ,t 3 ,...,t r ,remember , ,but The test statistic can be expressed as: .

[0032] The exponential distribution describes the case where the failure rate is constant, indicating that the failures of equipment during its lifespan are random. The exponential distribution is generally applicable to complex systems composed of multiple components, electronic systems or components, and systems with a constant failure rate. It is also suitable for calculating the lifespan of components where randomness is independent of time. The Weibull distribution was discovered and proposed by the Swedish physicist Weibull. While studying the strength of chains, Weibull found that when a chain is subjected to tension, the weakest link in the chain breaks first—the weakest link model—and derived the Weibull distribution from this. In a broader sense, it is a model where the failure of any part leads to the failure of the whole—the weakest link model. The Weibull distribution can describe various distribution phenomena by changing its shape, position, and scale parameters, and is easily handled mathematically, making it a "universal distribution."

[0033] S13. Based on the BAYES method, the reliability assessment model of multi-source information fusion is called to perform a reliability assessment on the assessment object, and the second goodness-of-fit test result is obtained.

[0034] The Bayesian method, or Bayesian estimation method, treats the unknown parameters of a distribution model as random variables and uses a probability density function to describe the degree of uncertainty about the parameters. Bayes' theorem states: ,in, This is called the posterior density function. This is called the prior density function. It is the sampling density function of the data. When the experiment ends, the value of y is determined, and at this time the sampling distribution can be regarded as an unknown parameter. The sampling distribution function is called the likelihood function.

[0035] In the above formula Since it is a constant term, for the prior distribution, the posterior distribution and the likelihood function have the following relationship: posterior distribution Prior distribution Likelihood function.

[0036] Bayesian estimation integrates sample information and prior information, and can update prior information with posterior information and observation data, making the estimation results closer to the reality.

[0037] In this embodiment, the reliability and quality data in the reliability and quality evaluation database of the evaluated object are screened and cleaned; a reliability evaluation mathematical model is determined, and its distribution parameters and goodness-of-fit test results are solved using the BAYES method; based on the distribution parameters and goodness-of-fit test results, a second goodness-of-fit test result and a second mean time between failures are obtained.

[0038] Furthermore, regarding how to determine the reliability assessment mathematical model and use the BAYES method to solve for its distribution parameters and goodness-of-fit test results, the specific steps are as follows: A quality and reliability evaluation database storing the reliability data of the assessed object is accessed, and similar products and their weight scores are selected using the BWM (BEST-WORST-METHOD) method. Based on the similar products and their weight scores, the probability distribution models are selected as exponential distribution and Welb distribution, respectively, and the initial prior distribution of the assessed object is obtained. According to the prior information of the initial prior distribution, the posterior calculation formula of the assessed object is obtained using the BAYES method. The non-analytical solution of the BAYES method is obtained using the MCMC method. The posterior mean of the non-analytical solution is used as the posterior reliability feature parameter for goodness-of-fit testing. Based on the results of the goodness-of-fit test, the probability distribution with a high goodness-of-fit test is determined as the reliability assessment mathematical model.

[0039] It should be noted that the quality and reliability evaluation database storing the reliability data of the evaluation object is pre-set. It stores the reliability data of the evaluation object and the relevant reliability data of its similar products. In this embodiment, it is necessary to call the data information of the quality and reliability evaluation database, and then select the similar products of the evaluation object and their weight scores by using the BWM method. This means using the rank-sum test method to perform a compatibility test on the reliability data of the selected similar products, and then screening out the similar product models that meet the requirements, and then determining the weight score of each screened similar product model.

[0040] Furthermore, a conjugate prior distribution is used as the prior distribution of the exponential distribution; a normal distribution or a uniform distribution is selected as the prior distribution of the Welb distribution; the prior distributions of the exponential distribution and the Welb distribution are weighted and fused based on the score weights of similar products to obtain the initial prior distribution of the evaluation object.

[0041] For the exponential distribution The commonly used prior distribution is its conjugate prior distribution: the Gamma distribution, which has the following form: In the formula , for The distribution parameters.

[0042] For the Welb distribution The normal or uniform distribution is chosen as the distribution parameter of the Weibull distribution. and The distribution functions of the two are independent of each other, and their joint distribution is taken as the prior distribution: When normally distributed: , ,have .

[0043] When uniformly distributed:

[0044] Furthermore, in Bayesian estimation models, it is very difficult to obtain the marginal posterior distribution of the parameters to be estimated by removing redundant parameters through high-dimensional integration. If the model is more complex, it may even be impossible to obtain the marginal posterior distribution of the parameters to be estimated. Therefore, this embodiment adopts the MCMC method, which can solve the numerical solution of the Bayes posterior distribution when the lifetime distribution follows the Weibull distribution, thereby obtaining the corresponding reliability assessment results through the posterior distribution.

[0045] S14. Compare the results of the first goodness-of-fit test and the second goodness-of-fit test, and select the probability distribution model and distribution parameters with the highest final goodness-of-fit as the mathematical model for reliability assessment.

[0046] S15. Extract the reliability assessment results from the reliability assessment mathematical model, and conduct quality analysis and evaluation of the assessment object based on the reliability assessment results.

[0047] By comparing the goodness-of-fit test results of the classical likelihood estimation method and the multi-source information fusion BAYES method (i.e., the first goodness-of-fit test result and the second goodness-of-fit test result), the probability distribution model and distribution parameters with the high final goodness-of-fit are selected as the final reliability assessment probability distribution model, and the MTBF is calculated and solved.

[0048] Furthermore, the batch failure rate, annual zero-reset rate, failure percentage of each component of the assessed object, failure percentage of each cause of failure, failure percentage of the assessed object under various geographical conditions, and curves showing the change of failure number over time are fitted. The changes in the number of failures for each cause of failure over time are also fitted to obtain statistical analysis results. Based on the calculation results and time trend graphs of the statistical analysis, with reliability assessment results as the core indicator, quality analysis and evaluation are conducted in conjunction with the aforementioned statistical analysis results. Specifically, this includes: analyzing the quality characteristics of the assessed object based on statistical analysis results from different dimensions; analyzing the quality control status based on failure trend changes; analyzing key points for product design improvement based on the distribution of failure types; and analyzing the field operation support capability based on maintenance efficiency.

[0049] This invention provides a quality and reliability evaluation method based on small sample data. The method involves: acquiring a database of reliability and quality evaluation data for the evaluated object; performing a reliability assessment using a classical probability distribution lifetime model based on maximum likelihood estimation, obtaining a first goodness-of-fit test result; performing a reliability assessment using multi-source information fusion based on the BAYES method, obtaining a second goodness-of-fit test result; comparing the first and second goodness-of-fit test results, selecting the probability distribution model and distribution parameters with the highest final goodness of fit as the final reliability evaluation probability distribution model; using the final reliability evaluation probability distribution model to obtain the reliability evaluation result of the evaluated object; and performing a quality analysis and evaluation of the evaluated object's quality based on the reliability evaluation result. The overall calculation process is more convenient and faster, the model fitting effect is significantly improved, and the overall accuracy of reliability evaluation is enhanced.

[0050] Accordingly, one embodiment of the present invention also provides a quality and reliability evaluation system based on small sample data, such as... Figure 3 As shown, this system is applied to the aforementioned quality and reliability evaluation method based on small sample data, specifically including: The data acquisition module is used to acquire reliability and quality evaluation data of the evaluation object; The first test result module is used to perform a reliability assessment of the evaluation object by calling the classical probability distribution lifetime model based on the maximum likelihood estimation method, and to obtain the first goodness-of-fit test result. The second test result module is used to call the multi-source information fusion reliability assessment model based on the BAYES method to perform a reliability assessment on the assessment object and obtain the second goodness-of-fit test result. The result comparison module is used to compare the first goodness-of-fit test result and the second goodness-of-fit test result, and select the probability distribution model and distribution parameters with high final goodness of fit as the mathematical model for reliability assessment. The evaluation and analysis module is used to extract the reliability evaluation results of the reliability evaluation mathematical model and to perform quality analysis and evaluation on the quality of the evaluated object based on the reliability evaluation results.

[0051] The above describes the quality and reliability evaluation method and system based on small sample data provided by the embodiments of the present invention. The various embodiments in the specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section. It should be noted that those skilled in the art can make several improvements and modifications to the present invention without departing from the principles of the invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

[0052] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can implement the described functions using different methods for each specific application, but such implementation should not be considered beyond the scope of the invention. The steps of the methods or algorithms described in connection with the embodiments disclosed herein can be implemented directly in hardware, software modules executed by a processor, or a combination of both. Software modules can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art.

Claims

1. A quality and reliability evaluation method based on small sample data, characterized in that, include: Obtain reliability and quality evaluation data for the assessment object; The reliability of the evaluation object is assessed by calling the classical probability distribution lifetime model based on the maximum likelihood estimation method, and the first goodness-of-fit test result is obtained. The reliability assessment of the assessment object is performed by calling the multi-source information fusion reliability assessment model based on the BAYES method, and the second goodness-of-fit test result is obtained. By comparing the results of the first goodness-of-fit test and the second goodness-of-fit test, the probability distribution model and distribution parameters with the highest final goodness-of-fit are selected as the mathematical model for reliability assessment. Extract the reliability assessment results from the reliability assessment mathematical model, and conduct a quality analysis and evaluation of the assessment object based on the reliability assessment results.

2. The quality and reliability evaluation method based on small sample data according to claim 1, characterized in that, The reliability assessment of the evaluation object based on the maximum likelihood estimation method and the classical probability distribution lifetime model is obtained, including the following: Obtain reliability evaluation data of the evaluated object within a preset time period; The classic probability distribution lifetime model is invoked, and its distribution parameters and goodness-of-fit test results are obtained by using the maximum likelihood estimation method. Based on the distribution parameters and the goodness-of-fit test results, the first goodness-of-fit test result and the first average fault interval time are obtained.

3. The quality and reliability evaluation method based on small sample data according to claim 2, characterized in that, The process of calling the classical probability distribution lifetime model and using the maximum likelihood estimation method to solve for its distribution parameters and goodness-of-fit test results includes: The probability distribution models are selected as exponential distribution and Welb distribution respectively, and the distribution parameters of the exponential distribution and the Welb distribution are solved by the maximum likelihood estimation method respectively. The goodness-of-fit test was performed on the distribution parameters of the exponential distribution and the Welb distribution using the KS method. Based on the results of the goodness-of-fit test, the probability distribution with a high goodness-of-fit test is determined as the probability distribution of the classical probability distribution lifetime model.

4. The quality and reliability evaluation method based on small sample data according to claim 1, characterized in that, The reliability assessment of the assessment object is performed by calling the multi-source information fusion reliability assessment model based on the BAYES method, and the second goodness-of-fit test result is obtained, including: The reliability and quality data in the reliability and quality evaluation database of the evaluated object are screened and cleaned. The reliability assessment model of multi-source information fusion is invoked, and its distribution parameters and goodness-of-fit test results are obtained by using the BAYES method. Based on the distribution parameters and the goodness-of-fit test results, the second goodness-of-fit test results and the second average fault interval time are obtained.

5. The quality and reliability evaluation method based on small sample data according to claim 4, characterized in that, The process of calling the multi-source information fusion reliability assessment model and using the BAYES method to solve for its distribution parameters and goodness-of-fit test results includes: The quality and reliability evaluation database storing the reliability data of the evaluated object is invoked, and similar products of the evaluated object and their weight scores are selected using the BWM method. Based on the similar products and their weight scores, the probability distribution models are selected as exponential distribution and Welb distribution, respectively, and the initial prior distribution of the evaluation object is obtained. Based on the prior information of the initial prior distribution, the posterior calculation formula of the evaluation object is obtained using the BAYES method; The non-analytical solution of the BAYES method is obtained using the MCMC method; The posterior mean of the non-analytical solution is used as the posterior reliability feature parameter for goodness-of-fit testing. Based on the results of the goodness-of-fit test, the probability distribution with a high goodness-of-fit is determined as the probability distribution of the multi-source information fusion reliability assessment model.

6. The quality and reliability evaluation method based on small sample data according to claim 5, characterized in that, The selection of similar products and their weighted scores for the evaluation object using the BWM method includes: The rank-sum test method is used to perform a compatibility test on the reliability data of the selected similar products, and similar product models that meet the requirements are screened out. Determine the weight score for each of the selected similar product models.

7. The quality and reliability evaluation method based on small sample data according to claim 5, characterized in that, The step of obtaining the initial prior distribution of the evaluation object based on the similar products and their weight scores includes: The conjugate prior distribution is used as the prior distribution of the exponential distribution; Choose a normal or uniform distribution as the prior distribution of the Welb distribution; The prior distributions of the exponential distribution and the Welb distribution are weighted and fused based on the score weights of similar products to obtain the initial prior distribution of the evaluation object.

8. The quality and reliability evaluation method based on small sample data according to claim 1, characterized in that, The quality analysis and evaluation of the assessed object based on the reliability assessment results includes: Calculate the batch failure rate, annual zeroing rate, failure percentage of each component of the assessment object, failure percentage of each cause of failure, failure percentage of each geographical environment, fit the curve of failure number change over time, fit the change of failure number of each cause of failure over time, and obtain the statistical analysis results. The quality analysis and evaluation are conducted using the reliability assessment results as the core indicator and combined with the statistical analysis results.

9. The quality and reliability evaluation method based on small sample data according to claim 8, characterized in that, The quality analysis and evaluation based on the statistical analysis results specifically includes: By combining the statistical analysis results from different dimensions, the quality characteristics of the evaluation object are analyzed and evaluated. Analyze the quality control status based on changes in failure trends; Based on the distribution of failure types, analyze the key points for product design improvement; The ability to guarantee field use is analyzed based on maintenance efficiency.

10. A quality and reliability evaluation system based on small sample data, applied to the quality and reliability evaluation method based on small sample data as described in any one of claims 1-9, characterized in that, include: The data acquisition module is used to acquire reliability and quality evaluation data of the evaluation object; The first test result module is used to perform a reliability assessment of the evaluation object by calling the classical probability distribution lifetime model based on the maximum likelihood estimation method, and to obtain the first goodness-of-fit test result. The second test result module is used to call the multi-source information fusion reliability assessment model based on the BAYES method to perform a reliability assessment on the assessment object and obtain the second goodness-of-fit test result. The result comparison module is used to compare the first goodness-of-fit test result and the second goodness-of-fit test result, and select the probability distribution model and distribution parameters with high final goodness of fit as the mathematical model for reliability assessment. The evaluation and analysis module is used to extract the reliability evaluation results of the reliability evaluation mathematical model and to perform quality analysis and evaluation on the quality of the evaluated object based on the reliability evaluation results.

Images