System reliability analysis method based on multi-index staged degradation model

Through the multi-index phased degradation model, combined with the Wiener process and Vine-Copula function, the problem of difficult to consider the relationship between multiple failure modes and multiple indicators in the existing technology is solved, and the accurate reliability evaluation of complex systems is achieved, and the analysis is improved. Accuracy and operation and maintenance support.

CN119026319APending Publication Date: 2024-11-26CHONGQING JIAOTONG UNIV
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202410953759.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-16
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The existing reliability analysis methods mainly target individual performance degradation indicators in single failure mode, and cannot effectively consider the coupling relationship between multiple failure modes and multiple indexes, resulting in low accuracy in reliability evaluation and inability to meet the actual needs of complex equipment systems.

Method used

The system reliability analysis method based on multi-index phased degradation model is adopted. By determining key performance degradation indicators, a phased degradation model of the Wiener process is established, and a Vine-Copula function is used to build a system multi-index phased degradation model, combined with Bayes-Gibbs Step-by-step parameter estimation method, calculate multi-failure reliability model to improve analysis accuracy.

Benefits of technology

It effectively improves the accuracy of system reliability analysis, provides accurate data support for the operation and maintenance of equipment systems, and can more accurately reflect the relationship between the multi-stage degradation laws and multiple failure modes of complex systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026319A_ABST
    Figure CN119026319A_ABST
Patent Text Reader

Abstract

The invention provides a system reliability analysis method based on a multi-index staged degradation model. The method comprises the following steps: S1, determining key performance degradation indexes of a to-be-evaluated system; s2, establishing a stage degradation model of the Wiener process based on the key performance degradation index; s3, based on the stage degradation model of the Wiener process, establishing a system multi-index stage degradation model by adopting a Vine-Copula function; s4, determining a multi-stage change point position of the multi-index stage degradation model of the to-be-evaluated system, substituting the change point position into the multi-index stage degradation model to obtain a multi-failure reliability model, and determining a to-be-evaluated parameter; and S5, determining an evaluation parameter by adopting a Bayes-Gibbs step-by-step parameter estimation method, and substituting the evaluation parameter into the multi-failure reliability model to calculate and obtain a reliability index of the to-be-evaluated system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to an equipment reliability analysis method, and in particular to a system reliability analysis method based on a multi-index staged degradation model. Background Art

[0002] With the continuous development of science and technology, equipment systems in the fields of manned spaceflight, energy equipment, rail transit, etc. are rapidly developing towards integration, diversification, and intelligence. Equipment in these fields often bears significant safety responsibilities and high operating costs. Therefore, scientific prediction of its health status can effectively improve the reliability and safety of the overall system, optimize the rational use of maintenance resources, and strengthen the performance and life management of equipment systems. However, due to the many factors that affect system reliability, system reliability analysis that only considers the degradation of a single performance indicator does not conform to engineering practice.

[0003] Existing reliability modeling and analysis methods mainly include two types of methods: failure data-based and degradation data-based. Reliability technology based on failure data is widely used in mass-produced products with short service life. It usually requires a large number of experiments to obtain sufficient failure samples. Based on the tested failure sample life, the distribution function parameters can be estimated, and then the reliability can be directly calculated through the distribution function. The use of random processes for degradation data analysis and reliability modeling is an important method for reliability assessment of products with high reliability, small samples, and missing failure data. It is currently very common in the field of reliability assessment; for example: Gao proposed a multiphase degradation model based on Wiener random process to describe the sudden change phenomenon of fault threshold and drift parameters, and derived an analytical expression for system reliability; Sun et al. combined the acceleration model with the Gamma random process and proposed an improved reliability analysis method based on the accelerated degradation test of rubber O-rings; Morita proposed a reliability analysis method based on the inverse Gaussian process model with a fragile term.

[0004] However, the above research analyzes the reliability of products or systems with a single performance degradation index under a single failure mode, which is inconsistent with the actual engineering situation. Usually, equipment systems include many key power transmission parts. Due to the diversity of functions and complexity of components, they often have multiple failure modes, and there is a coupling relationship between these failure modes. In reliability assessment, multiple failure modes correspond to multiple degradation indicators, and any degradation failure of any of these indicators will cause system failure. If the existence of multiple indicators is not taken into account, the reliability assessment of the system will be too conservative, resulting in low accuracy of reliability assessment and failure to guide actual production.

[0005] In the prior art, a lot of work has been done on reliability research based on performance index degradation models. However, researchers have found in the actual research process that in real situations, due to the complex performance degradation mechanism of products or systems, their degradation process presents a staged evolutionary characteristic. Taking the rail transit gear transmission system as an example, the gear transmission system is subject to dynamic changes in the operating environment, load conditions, and the influence of multi-source uncertainties such as random errors, alternating loads, flexible track beams, and speed fluctuations. Its performance degradation process will inevitably present staged characteristics. At this point, the traditional single-stage performance degradation model cannot accurately reflect the real degradation law and reliability of the system. Therefore, how to comprehensively consider the correlation between multiple indicators and the staged degradation law under the system's multiple failure modes and construct a reasonable degradation model to accurately evaluate the reliability of the system has become a difficult point in the research.

[0006] Therefore, in order to solve the above technical problems, it is urgent to propose a new technical means. When conducting reliability assessment on the system to be evaluated, key degradation indicators are determined from the system to be evaluated, and the reliability of the entire system to be evaluated is analyzed from the multiple stages of the key degradation indicators and their mutual influence, so as to effectively improve the accuracy of the reliability analysis and provide accurate data support for the subsequent operation and maintenance of the system to be evaluated. Summary of the invention

[0007] In view of this, the purpose of the present invention is to provide a system reliability analysis method based on a multi-indicator staged degradation model, to determine the key degradation indicators from the system to be evaluated, and to perform reliability analysis of the entire system to be evaluated from the multiple stages of the key degradation indicators and their mutual influence, thereby effectively improving the accuracy of the reliability analysis and providing accurate data support for the subsequent operation and maintenance of the system to be evaluated.

[0008] The present invention provides a system reliability analysis method based on a multi-index staged degradation model, comprising the following steps:

[0009] S1. Determine the key performance degradation indicators of the system to be evaluated;

[0010] S2. Establish a stage degradation model of the Wiener process based on key performance degradation indicators;

[0011] S3. Based on the staged degradation model of Wiener process, the Vine-Copula function is used to establish a multi-index staged degradation model of the system;

[0012] S4. Determine the multi-stage change point positions of the multi-index stage degradation model of the system to be evaluated, substitute the change point positions into the multi-index stage degradation model to obtain a multi-failure reliability model, and determine the parameters to be evaluated;

[0013] S5. The evaluation parameters are determined by using the Bayes-Gibbs step-by-step parameter estimation method, and the evaluation parameters are substituted into the multi-failure reliability model to calculate the reliability index of the system to be evaluated.

[0014] Further, in step S2, establishing a staged degradation model of the Wiener process based on key performance degradation indicators specifically includes:

[0015] Establish a nonlinear degradation model for the Wiener process:

[0016] X(t)=X(0)+μΛ(t;γ)+σB(t);

[0017] Where: μ is the drift coefficient, Λ(t;γ) is the time scale transformation function, σ is the diffusion coefficient, and B(·) is the standard Brownian motion;

[0018] Based on the nonlinear degradation model of Wiener process, a multi-stage degradation model is established for index i:

[0019]

[0020] Where: X i (t) is the degradation amount of performance index i (i=1,2,...,n) at time t; X i,k (t) is the degradation amount of performance index i at the kth (k=1,2,...,m) stage at time t; τ i,k is the change point of performance index i from the kth stage to the k+1th stage, τ i,0 =0,τ i,m =∞;X i (τ i,k-1 ) is the initial degradation amount of performance index i at the kth stage; is the characteristic function, when t∈(τ i,k-1 ,τ i,k ]hour, otherwise μ i,k and σ i,k are the drift coefficient and diffusion coefficient of index i at stage k, respectively, characterizing the degradation rate and dispersion of each stage; is the time scale transformation function of index i at stage k, which characterizes the nonlinear characteristics of index i degradation at stage k;

[0021] Determine the degradation increment of performance indicator i in time period j:

[0022] △X i,k (t j )=X i,k (t j )-X i,k (t j-1);

[0023] The total degradation of performance index i during the entire time period T is:

[0024]

[0025] In the Wiener process, the degradation increment △X i,k At the changing point τ i,k The front and back follow the normal distribution with different parameters, specifically:

[0026]

[0027] Degenerate △X i,k The cumulative distribution function of the increment is:

[0028]

[0029] Degenerate △X i,k The probability density function of the increment is:

[0030]

[0031] Where: △t=t j -t j-1 , when t j ≤τ i,k When the degradation increment △X ij The mean is The variance is Normal distribution; when t j >τ i,k When the degradation increment △X ij The mean is The variance is Normal distribution of

[0032] Among them: mean for:

[0033]

[0034] variance for:

[0035]

[0036] Time scale transformation function for:

[0037]

[0038] Further, step S3 specifically includes:

[0039] Determine the joint distribution function F(△X 1,k,△X 2,k ,L,△X n,k ), where: joint distribution function F(△X 1,k ,△X 2,k ,L,△X n,k ) is the joint distribution function of an n-dimensional random variable, and the distribution functions of the degradation index increment are F1(△X 1,k ),F2(△X 2,k ),L,F n (△X n,k );

[0040] According to the definition of Copula function and Sklar's theorem, there exists a Copula function that makes the following equation true:

[0041] F(△X 1,k ,△X 2,k ,L,△X n,k )=C(F1(△X 1,k ),F2(△X 2,k ),L,F n (△X n,k );θ);

[0042] Where: C(·) is the Copula function, θ is the Copula function parameter;

[0043] When F1(△X 1,k ),F2(△X 2,k ),L,F n (△X n,k ) is a continuous function and its inverse function exists, then the Copula function is uniquely determined, and its joint probability density function is expressed as:

[0044] f(△X 1,k ,△X 2,k ,L,△X n,k )=c(F1(△X 1,k ),L,F n (△X n,k );θ)f1(△X 1,k )Lf n (△X n,k );

[0045] f1(△X 1,k )f2(△X 2,k )Lf n (△X n,k ) is △X 1,k ,△X 2,k ,L,△X n,k The probability density function of 1,k ),F2(△X2,k ),L,F n (△X n,k );θ) is △X 1,k ,△X 2,k ,L,△X n,k The Copula density function of , and:

[0046] Using the cascading structure of vines using graphical analysis methods, the multi-variable Copula functions are decomposed into a series of binary paired Copula functions through a pre-set tree structure. Each paired Copula captures the C-Vine dependency structure between the corresponding two variables, and then all binary paired Copula functions are multiplied to form a C-Vine-Copula.

[0047] Each edge of the C-Vine dependency structure corresponds to a pair Copula function, and the joint probability density function of the pair Copula function is:

[0048]

[0049] Based on the stage characteristics of the degradation process of the system to be evaluated, the change points of the i degradation indicators divide the test time period into i+1 segments, and θ takes different values ​​in each segment, namely θ1, θ2, L, θ i+1 ,So:

[0050]

[0051] The multi-index staged degradation model is constructed as follows:

[0052]

[0053] Further, in step S4, the multiple failure reliability model is specifically:

[0054] The degradation process of n performance indicators in the system to be evaluated is X i,k (t), the failure threshold is D i , i=1,2...n, when the degradation of any performance indicator of the system exceeds its failure threshold, the system fails. Let T i = inf{t≥0|X i (t)≥D i} represents the time when indicator i first reaches the failure threshold, and the cumulative failure distribution function F of indicator i i,k (t) can be expressed as:

[0055]

[0056] in:

[0057] The reliability function of index i can be expressed as:

[0058] R i,k (t) = 1-F i,k (t);

[0059] The multiple failure mode reliability function of the system to be evaluated is:

[0060]

[0061] The degradation process of the system to be evaluated includes n performance indicators and multiple change stages k. The n performance degradation indicators of the system to be evaluated correspond to n change points τ 1,1 ,...,τ n,1 ; when t∈[τ k-1 ,τ j ) stage, the system to be evaluated is in [τ k-1 ,τ j The reliability of the stage is:

[0062]

[0063]

[0064] Where: μ k =(μ 1,k ,L,μ n,k ),σ k =(σ 1,k ,L,σ n,k ), the conditional joint probability density function of n performance indicators at the change point

[0065]

[0066] Then, the reliability of the system to be evaluated when there is a change point is:

[0067]

[0068] Among them: the parameters to be evaluated of the system to be evaluated are:

[0069] Θ=(μ i,k ,σ i,k ,γ i,k ,τ i,k ,θ i,1 ,θ i,2 ,...,θ i,k )i=1,2,...,n;k=1,2,...,m;i is the i-th key performance degradation indicator of the system to be evaluated; k is the different change stages of the key performance degradation indicator of the system to be evaluated.

[0070] Beneficial effects of the present invention: Through the present invention, key degradation indicators are determined from the system to be evaluated, and the reliability analysis of the entire system to be evaluated is performed from the multiple stages of the key degradation indicators and their mutual influence, thereby effectively improving the accuracy of the reliability analysis and providing accurate data support for the subsequent operation and maintenance of the system to be evaluated. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0072] Figure 1 It is a flow chart of the present invention.

[0073] Figure 2 It is a flow chart of estimating parameters to be evaluated of the present invention.

[0074] Figure 3 This is a phase change point analysis diagram of the degradation index of a specific example of the present invention.

[0075] Figure 4 It is the reliability curve of the specific example of the present invention under different evaluation models.

[0076] Figure 5 These are curves of different correlations of specific examples of the present invention. DETAILED DESCRIPTION

[0077] The present invention is further described in detail below:

[0078] The present invention provides a system reliability analysis method based on a multi-index staged degradation model, comprising the following steps:

[0079] S1. Determine the key performance degradation indicators of the system to be evaluated;

[0080] Among them: Determining the key performance degradation indicators of the system first requires the comprehensive use of domain knowledge, historical data analysis and reliability engineering methods. Through discussions and experience with domain experts, key factors that may affect system performance are identified, and the importance of these factors is verified by analyzing past failure data and operation records.

[0081] Applying reliability engineering tools such as failure mode and effects analysis, the main failure modes that may lead to system degradation and their associated key performance indicators can be systematically identified.

[0082] After determining the key performance indicators, a variety of analytical methods can be used to analyze the relationship between indicators, such as statistical correlation analysis to evaluate the linear correlation between different indicators, causal inference technology to explore possible causal relationships, time series analysis to reveal the trend of indicators over time, and machine learning models to discover complex nonlinear relationships and predict performance degradation patterns. These analytical methods can not only help understand how the various indicators affect each other, but also provide a theoretical basis and scientific basis for predicting system health, formulating effective maintenance strategies, and optimizing operations.

[0083] S2. Establish a stage degradation model of the Wiener process based on key performance degradation indicators;

[0084] S3. Based on the staged degradation model of Wiener process, the Vine-Copula function is used to establish a multi-index staged degradation model of the system;

[0085] S4. Determine the multi-stage change point positions of the multi-index stage degradation model of the system to be evaluated, substitute the change point positions into the multi-index stage degradation model to obtain a multi-failure reliability model, and determine the parameters to be evaluated;

[0086] S5. The evaluation parameters are determined by using the Bayes-Gibbs step-by-step parameter estimation method, and the evaluation parameters are substituted into the multi-failure reliability model to calculate the reliability index of the system to be evaluated. Through the above method, the key degradation index is determined from the system to be evaluated, and the reliability analysis of the entire system to be evaluated is performed from the multiple stages of the key degradation index and the influence between them, so as to effectively improve the accuracy of the reliability analysis and provide accurate data support for the subsequent operation and maintenance of the system to be evaluated.

[0087] In this embodiment, in step S2, establishing a staged degradation model of the Wiener process based on key performance degradation indicators specifically includes:

[0088] Establish a nonlinear degradation model for the Wiener process:

[0089] X(t)=X(0)+μΛ(t;γ)+σB(t);

[0090] Where: μ is the drift coefficient, Λ(t;γ) is the time scale transformation function, which reflects the nonlinear characteristics of the degradation process. Usually Λ(t;γ)=t γ , σ is the diffusion coefficient, B(·) is the standard Brownian motion;

[0091] Based on the nonlinear degradation model of Wiener process, a multi-stage degradation model is established for index i:

[0092]

[0093] Where: X i (t) is the degradation amount of performance index i (i=1,2,...,n) at time t; X i,k (t) is the degradation amount of performance index i at the kth (k=1,2,...,m) stage at time t; τ i,k is the change point of performance index i from the kth stage to the k+1th stage, τ i,0 =0 , τ i,m =∞ ; X i (τ i,k-1 ) is the initial degradation amount of performance index i at the kth stage; is the characteristic function, when t∈(τ i,k-1 ,τ i,k ]hour, otherwise μ i,k and σ i,k are the drift coefficient and diffusion coefficient of index i at stage k, respectively, characterizing the degradation rate and dispersion of each stage; is the time scale transformation function of index i at stage k, which characterizes the nonlinear characteristics of index i degradation at stage k;

[0094] Determine the degradation increment of performance indicator i in time period j:

[0095] △X i,k (t j )=X i,k (t j )-X i,k (t j-1 );

[0096] The total degradation of performance index i during the entire time period T is:

[0097]

[0098] In the Wiener process, the degradation increment △X i,k At the changing point τ i,k The front and back follow the normal distribution with different parameters, specifically:

[0099]

[0100] Degenerate △X i,k The cumulative distribution function of the increment is:

[0101]

[0102] Degenerate △X i,k The probability density function of the increment is:

[0103]

[0104] Where: △t=t j -t j-1 , when t j ≤τ i,k When the degradation increment △X ij The mean is The variance is Normal distribution; when t j >τ i,k When the degradation increment △X ij The mean is The variance is Normal distribution of

[0105] Among them: mean for:

[0106]

[0107] variance for:

[0108]

[0109] Time scale transformation function for:

[0110]

[0111] In this embodiment, step S3 specifically includes:

[0112] Determine the joint distribution function F(△X 1,k ,△X 2,k ,L,△X n,k ), where: joint distribution function F(△X 1,k ,△X 2,k ,L,△X n,k ) is the joint distribution function of an n-dimensional random variable, and the distribution functions of the degradation index increment are F1(△X 1,k ),F2(△X 2,k ),L,F n (△X n,k );

[0113] According to the definition of Copula function and Sklar's theorem (both Copula function definition and Sklar's theorem are common knowledge and will not be elaborated here), there exists a Copula function that makes the following equation true:

[0114] F(△X 1,k ,△X 2,k ,L,△X n,k )=C(F1(△X1,k ),F2(△X 2,k ),L,F n (△X n,k );θ);

[0115] Where: C(·) is the Copula function, θ is the Copula function parameter;

[0116] When F1(△X 1,k ),F2(△X 2,k ),L,F n (△X n,k ) is a continuous function and its inverse function exists, then the Copula function is uniquely determined, and its joint probability density function is expressed as:

[0117] f(△X 1,k ,△X 2,k ,L,△X n,k )=c(F1(△X 1,k ),L,F n (△X n,k );θ)f1(△X 1,k )Lf n (△X n,k );

[0118] f1(△X 1,k )f2(△X 2,k )Lf n (△X n,k ) is △X 1,k ,△X 2,k ,L,△X n,k The probability density function of 1,k ),F2(△X 2,k ),L,F n (△X n,k );θ) is △X 1,k ,△X 2,k ,L,△X n,k The Copula density function of , and:

[0119] Using the cascading structure of vines using graphical analysis methods, the multi-variable Copula functions are decomposed into a series of binary paired Copula functions through a pre-set tree structure. Each paired Copula captures the C-Vine dependency structure between the corresponding two variables, and then all binary paired Copula functions are multiplied to form a C-Vine-Copula.

[0120] The C-Vine dependency structure is an n-dimensional C-Vine-Copula structure, which contains n-1 trees T l,l=1,2,L,n-1; each tree has n+1-l nodes and nl edges. Each tree has a main node, which is connected to other nodes. Each edge formed corresponds to a pair Copula function. Then, each edge of the C-Vine dependency structure corresponds to a pair Copula function. The joint probability density function of the pair Copula function is:

[0121]

[0122] Based on the stage characteristics of the degradation process of the system to be evaluated, the change points of the i degradation indicators divide the test time period into i+1 segments, and θ takes different values ​​in each segment, namely θ1, θ2, L, θ i+1 ,So:

[0123]

[0124] The multi-index staged degradation model is constructed as follows:

[0125]

[0126] In this embodiment, in step S4, the multiple failure reliability model is specifically:

[0127] The degradation process of n performance indicators in the system to be evaluated is X i,k (t), the failure threshold is D i , i=1,2...n, when the degradation of any performance indicator of the system exceeds its failure threshold, the system fails. Let T i = inf{t≥0|X i (t)≥D i} represents the time when indicator i first reaches the failure threshold, and the cumulative failure distribution function F of indicator i i,k (t) can be expressed as:

[0128]

[0129] in:

[0130] The reliability function of index i can be expressed as:

[0131] R i,k (t) = 1-F i,k (t);

[0132] The multiple failure mode reliability function of the system to be evaluated is:

[0133]

[0134] The degradation process of the system to be evaluated includes n performance indicators and multiple change stages k. The n performance degradation indicators of the system to be evaluated correspond to n change points τ 1,1 ,...,τ n,1 ; when t∈[τ k-1 ,τ j ) stage, the system to be evaluated is in [τ k-1 ,τ j The reliability of the stage is:

[0135]

[0136]

[0137] Where: μ k =(μ 1,k ,L,μ n,k ),σ k =(σ 1,k ,L,σ n,k ), the conditional joint probability density function of n performance indicators at the change point

[0138]

[0139] Then, the reliability of the system to be evaluated when there is a change point is:

[0140]

[0141] Among them: the parameters to be evaluated of the system to be evaluated are:

[0142] Θ=(μ i,k ,σ i,k ,γ i,k ,τ i,k ,θ i,1 ,θ i,2 ,θ i,3 ,θ i,k )i=1,2,...,n;k=1,2,...,m;i is the i-th key performance degradation indicator of the system to be evaluated; k is the different change stages of the key performance degradation indicator of the system to be evaluated.

[0143] For the solution of the parameters to be evaluated, the Bayes method and Gibbs sampling are combined (these two methods are existing technologies and their processes are not described here), and the Win BUGS software is used to estimate the parameters, and iterative optimization is performed in multidimensional conditions. Combined with prior knowledge, the posterior distribution of the model parameters is obtained. Specifically: The first step is to solve the parameter α in the degradation process = (μ i,k ,σ i,k ,γ i,k ,τ i,k)i=1,2,...,n;k=1,2,...m,The second step is to solve the parameter θ=(θ i,1 ,θ i,2 ,...,θ i,n )i=1,2,...n, and the model is tested at each step. The specific process is as follows Figure 2 As shown in the figure: According to the continuous iterative solution of this process, the required parameters for building the model can be obtained. After the estimated model parameters are determined, they are substituted into the constructed multi-failure mode reliability model, and the multi-failure mode reliability analysis of the system is performed based on the model. The system reliability curve under different performance indicators of the system and the reliability solution curve under multi-stage and multi-degradation indicators are obtained. The reliability curve can be used to evaluate and predict the reliability of the system.

[0144] In a specific embodiment, the first step is to solve the parameter α in the degradation process = (μ i,k ,σ i,k ,γ i,k ,τ i,k )i=1,2,3;k=1,2,the second step is to solve the parameters of each degradation process of Copula θ=(θ i,1 ,θ i,2 ,θ i,3 ,θ i,4 )i=1,2,3, the increment of each degradation index satisfies: In the first stage, it can be expressed as: The second stage can be expressed as: Then the log-likelihood function can be expressed as:

[0145]

[0146] The parameters of each degradation process are shown in Table 1.

[0147]

[0148] Based on the different correlations between λ1, λ2 and λ3, the Clayton Copula function is used to fit the joint distribution between λ1 and λ2; the Frank Copula function is used to fit the joint distribution between λ1 and λ3.

[0149] The log-likelihood function of the multi-index staged degradation model can be expressed as:

[0150]

[0151] Where: c is the Copula density function. According to the analysis, c 13 is the Frank Copula function, c 12 and c231 Both are Clayton Copula functions.

[0152] The parameter estimation results are shown in Table 2.

[0153] Pair Copula <![CDATA[θ i,1 ]]> <![CDATA[θ i,2 ]]> <![CDATA[θ i,3 ]]> <![CDATA[θ i,4 ]]> <![CDATA[c 12 ]]> 5.4106 13.9290 7.4722 12.0320 <![CDATA[c 13 ]]> 61.4490 50.0189 30.5763 85.0241 <![CDATA[c 231 ]]> 10.4289 15.6516 9.0304 7.8939

[0154] The initial value of the wear depth of the helical gear in the example is set to 0. When the wear depth exceeds 5% of the tooth thickness, failure will occur, so the failure threshold D1 is 1mm; the initial value of the bearing clearance of the sample gearbox is 2μm. When the bearing clearance reaches 0.6mm, it is considered as a bearing failure, that is, the failure threshold D2 is 0.6mm; the initial value of the lubricating oil consumption of the sample gearbox is 0. When the oil seepage exceeds 2L, it is considered as a failure, that is, the failure threshold D3 is 2L. Substituting the estimated values ​​of the degradation model parameters and the failure threshold into the formula, the reliability function curves of the three failure modes can be obtained. Combined with the system reliability analysis method based on a multi-index staged degradation model proposed in this patent, the reliability evaluation of the transmission system under the multi-stage degradation indicators of this example can be obtained as follows: Figure 4 As shown: The reliability evaluation of the transmission system based on the piecewise correlation of multiple degradation indicators is as follows: Figure 5 As shown:

[0155] Through the comparison of the above examples, when the product has the characteristics of multi-indicator combination and staged degradation, a reasonable modeling analysis is carried out by comprehensively considering the two characteristics. The system reliability under the conditions of constant correlation and segmented correlation of the three indicators is higher than the reliability when they are independent of each other. In addition, the system reliability under the segmented correlation of the three indicators is higher than the system reliability under the constant correlation, which shows that the segmented correlation modeling has a significant impact on the system reliability assessment. It is different that the three indicators do have local correlation. The reliability assessment results under the segmented correlation conditions of the indicators proposed in this patent are more credible. If the reliability assessment is still based on the constant correlation, the system reliability will be at risk of being underestimated, which will lead to reduced equipment utilization and increased maintenance costs. Compared with the independent model and the constant correlation model between indicators, the Vine-Copula model based on segmented correlation reasonably describes the correlation problem between the two indicators and takes into account the differences in the degree of correlation in different time periods, eliminating the potential impact of correlation on the reliability assessment results, thereby improving the system reliability assessment results and improving the accuracy of reliability assessment.

[0156] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A system reliability analysis method based on a multi-index staged degradation model, characterized by: The following steps are involved: S1. Determine the key performance degradation indicators of the system to be evaluated; S2. Establish a stage degradation model of the Wiener process based on key performance degradation indicators; S3. Based on the staged degradation model of Wiener process, the Vine-Copula function is used to establish a multi-index staged degradation model of the system; S4. Determine the multi-stage change point positions of the multi-index stage degradation model of the system to be evaluated, substitute the change point positions into the multi-index stage degradation model to obtain a multi-failure reliability model, and determine the parameters to be evaluated; S5. The evaluation parameters are determined by using the Bayes-Gibbs step-by-step parameter estimation method, and the evaluation parameters are substituted into the multi-failure reliability model to calculate the reliability index of the system to be evaluated.

2. The system reliability analysis method based on the multi-index staged degradation model according to claim 1 is characterized by: In step S2, establishing a staged degradation model of the Wiener process based on key performance degradation indicators specifically includes: Establish a nonlinear degradation model for the Wiener process: X(t)=X(0)+μΛ(t;γ)+σB(t); Where: μ is the drift coefficient, Λ(t;γ) is the time scale transformation function, σ is the diffusion coefficient, and B(·) is the standard Brownian motion; Based on the nonlinear degradation model of Wiener process, a multi-stage degradation model is established for index i: Where: X i (t) is the degradation amount of performance index i (i=1,2,...,n) at time t; X i,k (t) is the degradation amount of performance index i at the kth (k=1,2,...,m) stage at time t; τ i,k is the change point of performance index i from the kth stage to the k+1th stage, τ i,0 =0,τ i,m =∞;X i (τ i,k-1 ) is the initial degradation amount of performance index i at the kth stage; is the characteristic function, when t∈(τ i,k-1 ,τ i,k ]hour, otherwise μ i,k and σ i,k are the drift coefficient and diffusion coefficient of index i at stage k, respectively, characterizing the degradation rate and dispersion of each stage; is the time scale transformation function of index i at stage k, which characterizes the nonlinear characteristics of index i degradation at stage k; Determine the degradation increment of performance indicator i in time period j: △X i,k (t j )=X i,k (t j )-X i,k (t j-1 ); The total degradation of performance index i during the entire time period T is: In the Wiener process, the degenerate increment △X i,k At the change point τ i,k The front and back follow the normal distribution with different parameters, specifically: Degenerate △X i,k The cumulative distribution function of the increment is: Degenerate △X i,k The probability density function of the increment is: Where: △t=t j -t j-1 , when t j ≤τ i,k When the degradation increment △X ij The mean is The variance is Normal distribution; when t j >τ i,k When the degradation increment △X ij The mean is The variance is Normal distribution of Among them: mean for: variance for: Time scale transformation function for:

3. The system reliability analysis method based on the multi-index staged degradation model according to claim 2 is characterized by: Step S3 specifically includes: Determine the joint distribution function F(△X 1,k ,△X 2,k ,L,△X n,k ), where: joint distribution function F(△X 1,k ,△X 2,k ,L,△X n,k ) is the joint distribution function of an n-dimensional random variable, and the distribution functions of the degradation index increment are F1(△X 1,k ),F2(△X 2,k ),L,F n (△X n,k ); According to the definition of Copula function and Sklar's theorem, there exists a Copula function that makes the following equation true: F(△X 1,k ,△X 2,k ,L,△X n,k )=C(F1(△X 1,k ),F2(△X 2,k ),L,F n (△X n,k );θ); Where: C(·) is the Copula function, θ is the Copula function parameter; When F1(△X 1,k ),F2(△X 2,k ),L,F n (△X n,k ) is a continuous function and its inverse function exists, then the Copula function is uniquely determined, and its joint probability density function is expressed as: f(△X 1,k ,△X 2,k ,L,△X n,k )=c(F1(△X 1,k ),L,F n (△X n,k );θ)f1(△X 1,k )Lf n (△X n,k ); f1(△X 1,k )f2(△X 2,k )Lf n (△X n,k ) is the probability density function of △X1,k,△X2,k,L,△Xn,k; c(F1(△X 1,k ),F2(△X 2,k ),L,F n (△X n,k );θ) is △X 1,k ,△X 2,k ,L,△X n,k The Copula density function of , and: Using the cascading structure of vines using graphical analysis methods, the multi-variable Copula functions are decomposed into a series of binary paired Copula functions through a pre-set tree structure. Each paired Copula captures the C-Vine dependency structure between the corresponding two variables, and then all binary paired Copula functions are multiplied to form a C-Vine-Copula. Each edge of the C-Vine dependency structure corresponds to a pair Copula function, and the joint probability density function of the pair Copula function is: Based on the stage characteristics of the degradation process of the system to be evaluated, the change points of the i degradation indicators divide the test time period into i+1 segments, and θ takes different values ​​in each segment, namely θ1, θ2, L, θ i+1 ,So: The multi-index staged degradation model is constructed as follows:

4. The system reliability analysis method based on the multi-index staged degradation model according to claim 3 is characterized by: In step S4, the multiple failure reliability model is specifically: The degradation process of n performance indicators in the system to be evaluated is X i,k (t), the failure threshold is D i , i=1,2...n, when the degradation of any performance indicator of the system exceeds its failure threshold, the system fails. Let T i = inf{t≥0X i (t)≥D i } represents the time when indicator i first reaches the failure threshold, and the cumulative failure distribution function F of indicator i i,k (t) can be expressed as: in: The reliability function of index i can be expressed as: R i,k (t)=1-F i,k (t); The multiple failure mode reliability function of the system to be evaluated is: The degradation process of the system to be evaluated includes n performance indicators and multiple change stages k. The n performance degradation indicators of the system to be evaluated correspond to n change points τ 1,1 ,...,τ n,1 ; when t∈[τ k-1 ,τ j ) stage, the system to be evaluated is in [τ k-1 ,τ j The reliability of the stage is: Where: μ k =(μ 1,k ,L,μ n,k ),σ k =(σ 1,k ,L,σ n,k ), the conditional joint probability density function of n performance indicators at the change point Then, the reliability of the system to be evaluated when there is a change point is: Among them: the parameters to be evaluated of the system to be evaluated are: Θ=(μ i,k ,σ i,k ,γ i,k ,τ i,k ,θ i,1 ,θ i,2 ,θ i,3 ,θ i,k )i=1,2,...,n;k=1,2,...,m;i is the i-th key performance degradation indicator of the system to be evaluated; k is the different change stages of the key performance degradation indicator of the system to be evaluated.

Citation Information

Cited By

  • Lubricating grease accelerated degradation modeling and reliability evaluation method

    CN119740408A

  • Organic coating service life prediction method based on distribution before point change and random degradation combined modeling

    CN120430068A