Power connector contact service life analysis method based on hybrid random process

Abstract

The invention relates to a power connector contact life analysis method based on a hybrid random process, and the method comprises the steps: carrying out the modeling of a degradation path of a power connector contact based on historical data, constructing a hybrid random process, and constructing an improved hybrid degradation model; establishing an acceleration model in the accelerated degradation process; obtaining an integral accelerated degradation model expression based on variables related to stress in the improved hybrid degradation model and an acceleration model; performing two-step estimation on unknown parameters of the overall accelerated degradation model; and deducing the service life distribution of the contacts of the power connector from a mixed random process. The method is more suitable for the actual application condition, the degradation model is more accurate, the actual performance of the equipment under different stress conditions can be reflected more truly, and the credible prediction capability of the residual life of the equipment is remarkably improved; important technical reference is provided for subsequent reliability analysis, and a feasible solution is provided for practical engineering application.

Description

Technical Field

[0001] The invention relates to the technical field of calculation, extrapolation or counting, and in particular to a method for analyzing the life of a power connector contact based on a mixed random process. Background Art

[0002] Existing technologies face some challenges in reliability testing of high-reliability products, especially under normal stress conditions, where the degradation process of these products is often slow and requires long-term operation to collect sufficient failure data. Traditional reliability testing methods have limited applicability to such products, so accelerated degradation testing has gradually become an important means to ensure the reliability and safe operation of system equipment. Accelerated degradation testing, combined with the health management framework of reliability testing, provides a key solution for life analysis of long-life products.

[0003] Generally, accelerated degradation tests rely on random processes to construct degradation models, which have the advantages of high accuracy and easy modeling. However, a single degradation mechanism may lead to a poor match with the actual degradation process in modeling, resulting in the problem of model inapplicability. The influence of accelerated stress makes the multi-modal performance in the degradation process cannot be ignored. Therefore, how to establish an accurate accelerated degradation test model and conduct system life analysis under normal stress has become an issue that needs to be explored urgently.

[0004] The core of accelerated degradation test is to build acceleration model and performance degradation model. Acceleration model is used to describe the relationship between product degradation rate and applied stress, and usually needs to be combined with performance degradation model to form a complete accelerated degradation test framework. Acceleration model has various forms, including Arrhenius model, electric stress model and power law model. In terms of performance degradation model, according to different modeling methods, it can be mainly divided into several types: model based on failure physical process, model based on degradation trajectory, model based on degradation distribution, and model based on random process. Model based on failure physical process requires in-depth study of the failure mechanism of equipment and establishment of corresponding degradation model through analysis of physical and chemical reaction process. Although this method has high accuracy, it also requires a full understanding of failure mechanism, and the modeling process is relatively complicated. The method based on degradation trajectory uses actual degradation data for fitting, which can more accurately reflect the degradation state of equipment, but has high requirements on data quality and fitting method. Model based on degradation distribution considers that the degradation distribution at different time points is the same, but the model parameters change with time. The application is relatively simple, but it may not be able to capture the specific degradation trajectory. Finally, the degradation model based on random processes is closer to reality and can better describe the equipment degradation process affected by internal and external environments.

[0005] In summary, accelerated degradation testing, as an effective method to improve the reliability and safety of system equipment, has become a key way to analyze the life of high-reliability products. By combining the acceleration model with the performance degradation model, an important theoretical foundation has been laid for the performance degradation prediction and health management of high-reliability products. However, the application of existing methods in complex engineering systems still faces certain challenges, so further research and improvement are particularly important. Future work can focus on optimizing model construction, improving data processing capabilities, and enhancing the adaptability of the model to better meet the testing needs of high-reliability products. Summary of the invention

[0006] The invention solves the problems existing in the prior art and provides a method for analyzing the life of a power connector contact based on a mixed random process.

[0007] The technical solution adopted by the present invention is a method for analyzing the life of a power connector contact based on a mixed random process, the method comprising the following steps:

[0008] S1 models the degradation path of power connector contacts based on historical data, constructs a mixed random process, and constructs an improved mixed degradation model;

[0009] S2 establishes an accelerated model in the accelerated degradation process;

[0010] S3 obtains the overall accelerated degradation model expression based on the stress-related variables in the improved hybrid degradation model in S1 and the accelerated model in S2;

[0011] S4 uses a two-step method to estimate the unknown parameters of the overall accelerated degradation model;

[0012] S5 derives the power connector contact life distribution from a mixed random process.

[0013] Preferably, in S1, the degradation path of the hybrid degradation model is Y(t), satisfying,

[0014] Y(t) = Y 1 (t) + Y 2 (t) + Y 3 (t) (1)

[0015] Among them, Y 1 (t), Y 2 (t) and Y 3 (t) are random variables of Wiener process, Gamma process and Inverse Gaussian process respectively;

[0016] The probability density function of Y(t) satisfies,

[0017]

[0018] Among them, Y 1 (t), Y 2 (t) and Y 3 (t) are denoted by Y 1 (t)~N(μ 1 Λ(t),σ 2 Λ(t)), Y 2 (t)~Ga(αΛ(t),β) and Y 3 (t)~IG(μ 2 Λ(t),λΛ(t) 2 ), Λ(t) is a non-negative increasing function, μ 1 Λ(t) and σ 2 Λ(t) represents Y 1 The mean and variance of Y(t), αΛ(t) and β represent the 2 (t) Shape parameter and scale parameter of the distribution, μ 2 Λ(t) and Yes 3 (t); β and λ are parameters to be determined.

[0019] Preferably, in S2, the acceleration model is

[0020] lnR=η 0 +η 1 L(S) (3)

[0021] Among them, R is the degradation rate, S is the accelerated stress;

[0022] When η 0 , η 1 When , L(S) are different, the acceleration model corresponds to the Arrhenius model, power law model or electric stress model respectively.

[0023] Preferably, L(S) is normalized to obtain,

[0024]

[0025] Among them, S 0 , S and S H These are normal, current, and maximum stress levels.

[0026] Preferably, in S3, the stress-parameter change formula of the overall accelerated degradation model expression is:

[0027] lnμ 1 =a 1 +b 1 L(S) (5)

[0028] lnσ=u+vL(S) (6)

[0029] lnα=a 2 +b 2 L(S) (7)

[0030] lnμ 2 =a 3 +b 3 L(S) (8)

[0031] Among them, a 1 ,b 1 ,a 2 ,b 2 ,u,v,a 3 ,b 3 Parameters to be determined.

[0032] Preferably, S4 comprises the following steps:

[0033] S4.1 Mean parameter μ for a mixed random process 1 Λ(t),αβΛ(t),μ 2 Λ(t) is estimated by least squares to obtain the trend expression of the degradation process, and the dynamic characteristics of the weight are explained by time transformation;

[0034] S4.2 uses the Metropolis-Hastings method for parameter estimation, and combines the mean parameter estimation results obtained in S4.1 to obtain the overall accelerated degradation model.

[0035] Preferably, in S4.1, the path models of the Wiener process, the Gamma process, and the Inverse Gaussian process are respectively,

[0036] M 1 =μ 1 Λ(t)

[0037] M 2 =αβΛ(t) (9)

[0038] M 3 =μ 2 Λ(t)

[0039] The degradation path model of the mixed random process is:

[0040] M=M 1 +M 2 +M 3 =μ 1 Λ(t)+αβΛ(t)+μ 2 Λ(t) (10)

[0041] Estimate the degradation path model parameters through historical data;

[0042] The time scale is transformed as a power function of time t to describe the mean dynamics of the degradation process

[0043] Λ(t)=t c (11)

[0044] Where c is an unknown parameter not affected by stress, c>0;

[0045] Measurement error μ kj -M kj The square of Minimize the sum of

[0046]

[0047] Among them, μ kj is the mean of historical accelerated degradation data, M kj is the mean expression of the degradation path model at time j under the kth stress. Substitute equations (5), (7), (8) and (11) into equation (10) and use the least squares method to directly estimate the unknown parameter θ, θ = {a 1 ,a 2 ,a 3 ,b 1 ,b 2 ,b 3 ,β,c}.

[0048] Preferably, in S4.2, assuming that the target distribution is π(·), the joint posterior distribution is obtained based on Bayesian theory as follows:

[0049] π(θ R |x kij ,τ kij )∝L(θ R |x kij ,τ kij )p(θ R ) (13)

[0050] Among them, x kij is the real degradation data, i.e., the jth degradation value of unit i under the kth stress level, τ kij is the time increment, τ kij =Λ(t kij )-Λ(t ki(j-1) ); p(θ R ) is the joint prior distribution of the remaining unknown parameters, satisfying

[0051]

[0052] Assuming that the prior distribution of all parameters satisfies the Gaussian distribution and the other parameters are in the same form, the new state is generated by the kinetic equation K(θ′|θ) using the MH sampling method, and each kinetic equation is accepted or rejected relative to the acceptance probability.

[0053]

[0054] In the MH sampling method, a Markov chain is defined and converges to the target distribution. If θ in equation (15) R is the current state of the chain, then θ′ R is the corresponding proposed next state.

[0055] Preferably, in S5, γ represents the critical failure threshold of the degradation process described by the proposed hybrid model, and the lifespan T is defined as the time when the degradation path Y(t) first crosses the failure threshold γ.

[0056] T=inf{t:t≥0|Y(t)=γ} (16)

[0057] Based on the fact that Λ(t) is a non-negative increasing function, the cumulative distribution function of the first arrival time of the model satisfies

[0058]

[0059] in, and Φ(·) represent the standard normal probability density function PDF and the standard normal cumulative distribution function CDF, respectively. n = γ / αβ.

[0060] Preferably, the reliability function of the degradation model is expressed as R T (t) = 1-F T (t).

[0061] The present invention relates to a method for analyzing the life of a power connector contact based on a mixed random process. The method comprises the following steps: modeling the degradation path of a power connector contact based on historical data, constructing a mixed random process, and constructing an improved mixed degradation model; establishing an acceleration model in an accelerated degradation process; obtaining an overall accelerated degradation model expression based on stress-related variables and an acceleration model in the improved mixed degradation model; performing a two-step method to estimate unknown parameters of the overall accelerated degradation model; and deriving the life distribution of the power connector contact from the mixed random process.

[0062] The beneficial effects of the present invention are that it can well deal with the problems of model mismatch, failure mechanism change with stress and poor modeling accuracy existing in the accelerated degradation test of complex engineering systems. The model is more in line with the actual application situation, making the degradation model more accurate and able to more truly reflect the actual performance of the equipment under different stress conditions; by combining multiple random processes, the model can flexibly adapt to the changes of various failure mechanisms, thereby significantly improving the reliable prediction ability of the remaining life of the equipment; it provides an important technical reference for subsequent reliability analysis, and can help engineers make more informed decisions when formulating maintenance and replacement strategies; it not only expands the boundaries of existing research, but also provides a practical solution for actual engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 is a flow chart of the method of the present invention;

[0064] Figure 2 It is the waveform diagram of stress relaxation data along with stress degradation;

[0065] Figure 3 Dynamic estimation and historical expectation of degradation trend of stress relaxation data;

[0066] Figure 4 It is a CDF schematic diagram under the mixed random process of the present invention;

[0067] Figure 5 It is the reliability curve under the mixed random process of the present invention. DETAILED DESCRIPTION

[0068] The present invention is further described in detail below in conjunction with embodiments, but the protection scope of the present invention is not limited thereto.

[0069] The present invention relates to a method for analyzing the life of a power connector contact based on a mixed random process, the method comprising the following steps:

[0070] (1) Model the degradation path of power connector contacts based on historical data, construct a hybrid random process, and construct an improved hybrid degradation model;

[0071] (2) Establish an acceleration model in the accelerated degradation process;

[0072] (3) Based on the stress-related variables in the improved hybrid degradation model in (1) and the acceleration model in (2), the overall accelerated degradation model expression is obtained;

[0073] (4) Two-step estimation of unknown parameters of the overall accelerated degradation model;

[0074] (5) The contact life distribution of power connectors is derived from a mixed random process.

[0075] The following is a description of the accelerated degradation data set due to stress relaxation. This data set simulates the stress relaxation degradation data of a certain type of electrical connector. The data set includes operating data within 0-3000 hours, and a total of 6 samples under 3 groups of accelerated degradation data of different stresses. Figure 2 Shown are stress relaxation data along with stress degradation waveforms.

[0076] (1) Model the degradation path of power connector contacts based on historical data, construct a hybrid random process, and construct an improved hybrid degradation model;

[0077] The degradation path is modeled through the commonly used Wiener process, Gamma process and Inverse Gaussian process. The three single random processes are linearly combined to construct a mixed random process and a new mixed degradation model. The degradation path of the mixed degradation model is Y(t), which satisfies:

[0078] Y(t) = Y 1 (t) + Y 2 (t) + Y 3 (t) (1)

[0079] Among them, Y 1 (t), Y 2 (t) and Y 3 (t) are random variables of Wiener process, Gamma process and Inverse Gaussian process respectively;

[0080] The probability density function (PDF) of Y(t) satisfies,

[0081]

[0082] Among them, Y 1 (t), Y 2 (t) and Y 3 (t) are denoted by Y 1 (t)~N(μ 1 Λ(t),σ 2 Λ(t)), Y 2 (t)~Ga(αΛ(t),β) and Y 3 (t)~IG(μ 2 Λ(t),λΛ(t) 2 ), Λ(t) is a non-negative increasing function, which is also Y 1 (t), Y 2 (t) and Y 3(t) approximate description;

[0083] μ 1 Λ(t) and σ 2 Λ(t) represents Y 1 The mean and variance of Y(t), αΛ(t) and β represent the 2 (t) Shape parameter and scale parameter of the distribution, μ 2 Λ(t) and Yes 3 (t); β and λ are parameters to be determined; here β and λ are both positive numbers.

[0084] (2) Establish an acceleration model in the accelerated degradation process;

[0085] Through the commonly used stress models, namely Arrhenius model, power law model and electric stress model, an acceleration model in the accelerated degradation process is established to describe the relationship between the degradation rate and the accelerated stress, and a unified expression is constructed;

[0086] The acceleration model is

[0087] lnR=η 0 +η 1 L(S) (3)

[0088] Among them, R is the degradation rate, S is the accelerated stress;

[0089] When η 0 , η 1 When , L(S) are different, the acceleration model corresponds to the Arrhenius model, power law model or electric stress model respectively;

[0090] Specifically,

[0091] When η 0 =ln(δ 0 ), η 1 =-δ 1 , L(S)=1 / S, the above expression becomes the Arrhenius model;

[0092] When η 0 =ln(δ 0 ), η 1 =δ 1 , L(S)=ln(S), the above expression becomes a power law model;

[0093] When η 0 =ln(δ 0 ), η 1 =δ 1 , L(S)=S, the above expression becomes the electrical stress model.

[0094] To further quantify the effect of stress, L(S) is normalized to obtain:

[0095]

[0096] Among them, S 0 , S and S H They are normal, current and maximum stress levels;

[0097] That is, S 0 =0, S H =1, L(S)∈[0,1].

[0098] (3) Based on the stress-related variables in the improved hybrid degradation model in (1) and the acceleration model in (2), the overall accelerated degradation model expression is obtained;

[0099] The stress-parameter variation formula of the overall accelerated degradation model expression is:

[0100] lnμ 1 =a 1 +b 1 L(S) (5)

[0101] lnσ=u+vL(S) (6)

[0102] lnα=a 2 +b 2 L(S) (7)

[0103] lnμ 2 =a 3 +b 3 L(S) (8)

[0104] Among them, a 1 ,b 1 ,a 2 ,b 2 ,u,v,a 3 ,b 3 is the parameter to be determined;

[0105] Substituting equations (5) to (8) into equation (2), we can obtain a unified expression for the accelerated degradation model that takes stress into account.

[0106] (4) Two-step estimation of unknown parameters of the overall accelerated degradation model;

[0107] The unknown parameters of the accelerated degradation model based on mixed random process are estimated in two steps. Firstly, the least squares estimation is performed on the mean parameter of the mixed random process to obtain the trend expression of the degradation process, and the time transformation is used to explain the dynamic characteristics of the weight.

[0108] The following steps are involved:

[0109] (4-1) For the mean parameter μ of the mixed random process 1 Λ(t),αβΛ(t),μ 2 Λ(t) is estimated by least squares to obtain the trend expression of the degradation process, and the dynamic characteristics of the weight are explained by time transformation;

[0110] (4-2) The Metropolis-Hastings method is used for parameter estimation, and the mean parameter estimation results obtained in (4-1) are combined to obtain the overall accelerated degradation model.

[0111] In (4-1), the path model of the degradation trend can be constructed through the model of the mixed random process. Since the model of the mixed random process is in the form of linear addition, its degradation path can be expressed as the linear addition of each sub-degradation process path model. According to the properties of the Wiener process, Gamma process, and Inverse Gaussian process, its path model can be expressed as:

[0112]

[0113] Among them, M 1 ,M 2 ,M 3 They are the degradation path models of Wiener process, Gamma process and Inverse Gaussian process respectively.

[0114] According to formula (1), the degradation path model of the mixed random process is:

[0115] M=M 1 +M 2 +M 3 =μ 1 Λ(t)+αβΛ(t)+μ 2 Λ(t) (10)

[0116] The degradation path model parameters are estimated through historical data, and the accelerated degradation data are obtained from the historical data.

[0117] Under actual conditions, the parameter estimation of the degradation path model should take into account the influence of stress, because accelerated stress will affect the physical and chemical degradation process of the equipment. According to engineering experience, the parameter trajectory is generally linear, concave or convex, so the power function of time t can be used for time scale transformation to describe the mean dynamics of the degradation process.

[0118] Λ(t)=t c (11)

[0119] Where c is an unknown parameter not affected by stress, c>0;

[0120] In order to estimate the unknown parameters, the measurement error μ kj -M kj The square of The sum is minimized, based on the smaller the distance value, the smaller the M kij The closer the value is to the mean of historical data, the better the result is (12).

[0121]

[0122] Among them, μ kj is the mean of historical accelerated degradation data, M kj is the mean expression of the degradation path model at time j under the kth stress. Substitute equations (5), (7), (8) and (11) into equation (10) and use the least squares method to directly estimate the unknown parameter θ, θ = {a 1 ,a 2 ,a 3 ,b 1 ,b 2 ,b 3 ,β,c}, based on c, the dynamic form of the degradation process can be determined by formula (11).

[0123] In the present invention, the estimated parameters can be plotted as Figure 3 Dynamic estimates and historical expectations of degradation trends for stress relaxation data are shown.

[0124] In (4-2), assuming the target distribution π(·), the joint posterior distribution is obtained based on Bayesian theory:

[0125] π(θ R |x kij ,τ kij )∝L(θ R |x kij ,τ kij )p(θ R ) (13)

[0126] Among them, x kij is the real degradation data, i.e., the jth degradation value of unit i under the kth stress level, τ kij is the time increment, τ kij =Λ(t kij )-Λ(t ki(j-1) ); p(θ R ) is the joint prior distribution of the remaining unknown parameters, satisfying

[0127]

[0128] Assuming that the prior distribution of all parameters satisfies the Gaussian distribution and the other parameters are in the same form, the new state is generated by the kinetic equation K(θ′|θ) using the MH sampling method, and each kinetic equation is accepted or rejected relative to the acceptance probability.

[0129]

[0130] In the MH sampling method, a Markov chain is defined and converges to the target distribution. If θ in equation (15) R is the current state of the chain, then θ′ R is the corresponding proposed next state.

[0131] In the present invention, after estimating the parameters of the degradation trend model, the estimated values ​​are brought into the likelihood function, and the likelihood function is expressed as follows:

[0132]

[0133] Among them, τ PQR represents the jth time increment function of the ith degraded element under the kth stress, that is, According to the acceleration model, the parameter μ 1 ,α,μ 2 You can use a 1 ,b 1 ,a 2 ,b 2 ,a 3 ,b 3 Represents, and has been estimated in (4-1), the remaining unknown model parameters θ R It can be defined as θ R ={σ,β,λ}, where σ is also affected by stress, lnσ=u+vL(S), update θ R ={u,v,β,λ}.

[0134] The corresponding operations of the MH algorithm are:

[0135] S1 Input: Dataset {x kij ,τ kij}; target distribution π(θ R ); dynamic equation K(θ′ R |θ R ).

[0136] S2 Initialization state θ R (0);

[0137] S3 performs cyclic calculation: ir=1,2,3,···

[0138] S3.1 Sampling θ from the kinetic equation K(·) R (ir);

[0139] S3.2 calculate the acceptance rate ρ;

[0140] S3.3 Sample r from Uniform(0,1);

[0141] S3.4 If ρ≤r, then let θ R (ir+1)=θ R (ir); otherwise, let θ R (ir+1)=θ R (ir-1).

[0142] S4 ends the loop.

[0143] (5) The contact life distribution of power connectors is derived from a mixed random process.

[0144] The life analysis and reliability function are calculated according to the first hit time distribution. γ represents the critical failure threshold of the degradation process described by the proposed hybrid model. The life T is defined as the time when the degradation path Y(t) crosses the failure threshold γ for the first time, that is, the first hit time (FHT).

[0145] T=inf{t:t≥0|Y(t)=γ} (16)

[0146] Based on the fact that Λ(t) is a non-negative increasing function, the cumulative distribution function (CDF) of the first arrival time of the model satisfies

[0147]

[0148] in, and Φ(·) represent the standard normal probability density function PDF and the standard normal cumulative distribution function CDF, respectively. n = γ / αβ.

[0149] The reliability function of the degradation model is expressed as R T (t) = 1-F T (t).

[0150] Substituting the estimated weight parameters and model parameters into formula (16) and taking their derivatives and substituting them into formula (17), we can draw the following diagrams: Figure 4 As shown in the CDF based on the first arrival time distribution, Figure 5 Reliability curves under different degradation processes are shown.

[0151] In order to implement the above-mentioned embodiment, a computer-readable storage medium is proposed, on which a power connector contact life analysis program based on a hybrid random process is stored. When the power connector contact life analysis program based on a hybrid random process is executed by a processor, the power connector contact life analysis method based on a hybrid random process is implemented.

[0152] In order to implement the above-mentioned embodiment, a computer device is also proposed, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the power connector contact life analysis method based on the mixed random process as described above is implemented.

[0153] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for analyzing the life of power connector contacts based on a mixed random process, characterized in that: The method comprises the following steps: S1 models the degradation path of power connector contacts based on historical data, constructs a mixed random process, and constructs an improved mixed degradation model; S2 establishes an accelerated model in the accelerated degradation process; S3 obtains the overall accelerated degradation model expression based on the stress-related variables in the improved hybrid degradation model in S1 and the accelerated model in S2; S4 uses a two-step method to estimate the unknown parameters of the overall accelerated degradation model; S5 derives the power connector contact life distribution from a mixed random process.

2. The method for analyzing the life of a power connector contact based on a hybrid random process according to claim 1, characterized in that: In S1, the degradation path of the mixed degradation model is Y(t), which satisfies, Y(t) = Y1(t) + Y2(t) + Y3(t) (1) Among them, Y1(t), Y2(t) and Y3(t) are random variables of Wiener process, Gamma process and Inverse Gaussian process respectively; The probability density function of Y(t) satisfies, Among them, Y1(t), Y2(t) and Y3(t) are denoted as Y1(t)~N(μ1Λ(t),σ 2 Λ(t)), Y2(t)~Ga(αΛ(t),β) and Y3(t)~IG(μ2Λ(t),λΛ(t) 2 ), Λ(t) is a non-negative increasing function, μ1Λ(t) and σ 2 Λ(t) represents the mean and variance of Y1(t), αΛ(t) and β represent the shape parameter and scale parameter of the Y2(t) distribution, and μ2Λ(t) and are the mean and variance of Y3(t); β and λ are parameters to be determined.

3. The method for analyzing the life of a power connector contact based on a hybrid random process according to claim 2, characterized in that: In S2, the acceleration model is lnR=η0+η1L(S) (3) Among them, R is the degradation rate, S is the accelerated stress; When η0, η1, and L(S) are different, the acceleration model corresponds to the Arrhenius model, the power law model, or the electric stress model, respectively.

4. The method for analyzing the life of a power connector contact based on a hybrid random process according to claim 3, characterized in that: Normalizing L(S), we get Among them, S0, S and S H These are normal, current, and maximum stress levels.

5. The method for analyzing the life of a power connector contact based on a hybrid random process according to claim 3, characterized in that: In S3, the stress-parameter change formula of the overall accelerated degradation model expression is: lnμ1=a1+b1L(S) (5) lnσ=u+vL(S) (6) lnα=a2+b2L(S) (7) lnμ2=a3+b3L(S) (8) Among them, a1, b1, a2, b2, u, v, a3, b3 are parameters to be determined.

6. The method for analyzing the life of a power connector contact based on a hybrid random process according to claim 5, characterized in that: S4 includes the following steps: S4.1 The least squares estimation is performed on the mean parameters μ1Λ(t), αβΛ(t), and μ2Λ(t) of the mixed random process to obtain the trend expression of the degradation process and explain the dynamic characteristics of the weights by time transformation; S4.2 uses the Metropolis-Hastings method for parameter estimation, and combines the mean parameter estimation results obtained in S4.1 to obtain the overall accelerated degradation model.

7. The method for analyzing the life of a power connector contact based on a hybrid random process according to claim 6, characterized in that: In S4.1, the path models of Wiener process, Gamma process and Inverse Gaussian process are, The degradation path model of the mixed random process is: M=M1+M2+M3=μ1Λ(t)+αβΛ(t)+μ2Λ(t) (10) Estimate the degradation path model parameters through historical data; The time scale is transformed as a power function of time t to describe the mean dynamics of the degradation process Λ(t)=t c (11) Where c is an unknown parameter not affected by stress, c>0; Measurement error μ kj -M kj The square of Minimize the sum of Among them, μ kj is the mean of historical accelerated degradation data, M kj is the mean expression of the degradation path model at time j under the kth stress. Substitute equations (5), (7), (8), and (11) into equation (10) and use the least squares method to directly estimate the unknown parameters θ, θ = {a1, a2, a3, b1, b2, b3, β, c}.

8. The method for analyzing the life of a power connector contact based on a hybrid random process according to claim 6, characterized in that: In S4.2, assuming the target distribution π(·), the joint posterior distribution is obtained based on Bayesian theory: π(θ R |x kij ,t kij )∝L(θ R |x kij ,t kij )p(θ R ) (13) Among them, x kij is the real degradation data, i.e., the jth degradation value of unit i under the kth stress level, τ kij is the time increment, τ kij =Λ(t kij )-Λ(t ki(j-1) ); p(θ R ) is the joint prior distribution of the remaining unknown parameters, satisfying Assuming that the prior distribution of all parameters satisfies the Gaussian distribution and the other parameters are in the same form, the new state is generated by the kinetic equation K(θ′|θ) using the MH sampling method, and each kinetic equation is accepted or rejected relative to the acceptance probability. In the MH sampling method, a Markov chain is defined and converges to the target distribution. If θ in equation (15) R is the current state of the chain, then θ′ R is the corresponding proposed next state.

9. The method for analyzing the life of a power connector contact based on a hybrid random process according to claim 6, characterized in that: In S5, γ represents the critical failure threshold of the degradation process described by the proposed hybrid model, and the lifespan T is defined as the time when the degradation path Y(t) first crosses the failure threshold γ. T=inf{t:t≥0|Y(t)=γ} (16) Based on the fact that Λ(t) is a non-negative increasing function, the cumulative distribution function of the first arrival time of the model satisfies in, and Φ(·) represent the standard normal probability density function PDF and the standard normal cumulative distribution function CDF, respectively. n = γ / αβ.

10. The method for analyzing the life of a power connector contact based on a hybrid random process according to claim 1, characterized in that: The reliability function of the degradation model is expressed as R T (t) = 1-F T (t).

Images