Research article

Inference and optimal design for the k-level step-stress accelerated life test based on progressive Type-I interval censored power Rayleigh data


  • Received: 18 August 2023 Revised: 14 November 2023 Accepted: 22 November 2023 Published: 01 December 2023
  • In this paper, a new generalization of the one parameter Rayleigh distribution called the Power Rayleigh (PRD) was employed to model the life of the tested units in the step-stress accelerated life test. Under progressive Type-I interval censored data, the cumulative exposure distribution was considered to formulate the life model, assuming the scale parameter of PRD has the inverse power function at each stress level. Point estimates of the model parameters were obtained via the maximum likelihood estimation method, while interval estimates were obtained using the asymptotic normality of the derived estimators and the bootstrap resampling method. An extensive simulation study of k=4 levels of stress in different combinations of the life test under different progressive censoring schemes was conducted to investigate the performance of the obtained point and interval estimates. Simulation results indicated that point estimates of the model parameters are closest to their initial true values and have relatively small mean squared errors. Accordingly, the interval estimates have small lengths and their coverage probabilities are almost convergent to the 95% significance level. Based on the Fisher information matrix, the D-optimality and the A-optimality criteria are implemented to determine the optimal design of the life test by obtaining the optimum inspection times and optimum stress levels that improve the estimation procedures and give more efficient estimates of the model parameters. Finally, the developed inferential procedures were also applied to a real dataset.

    Citation: Hatim Solayman Migdadi, Nesreen M. Al-Olaimat, Omar Meqdadi. Inference and optimal design for the k-level step-stress accelerated life test based on progressive Type-I interval censored power Rayleigh data[J]. Mathematical Biosciences and Engineering, 2023, 20(12): 21407-21431. doi: 10.3934/mbe.2023947

    Related Papers:

    [1] Manal M. Yousef, Rehab Alsultan, Said G. Nassr . Parametric inference on partially accelerated life testing for the inverted Kumaraswamy distribution based on Type-II progressive censoring data. Mathematical Biosciences and Engineering, 2023, 20(2): 1674-1694. doi: 10.3934/mbe.2023076
    [2] Wael S. Abu El Azm, Ramy Aldallal, Hassan M. Aljohani, Said G. Nassr . Estimations of competing lifetime data from inverse Weibull distribution under adaptive progressively hybrid censored. Mathematical Biosciences and Engineering, 2022, 19(6): 6252-6275. doi: 10.3934/mbe.2022292
    [3] M. Nagy, Adel Fahad Alrasheedi . The lifetime analysis of the Weibull model based on Generalized Type-I progressive hybrid censoring schemes. Mathematical Biosciences and Engineering, 2022, 19(3): 2330-2354. doi: 10.3934/mbe.2022108
    [4] M. Nagy, M. H. Abu-Moussa, Adel Fahad Alrasheedi, A. Rabie . Expected Bayesian estimation for exponential model based on simple step stress with Type-I hybrid censored data. Mathematical Biosciences and Engineering, 2022, 19(10): 9773-9791. doi: 10.3934/mbe.2022455
    [5] H. T. Banks, R. A. Everett, Neha Murad, R. D. White, J. E. Banks, Bodil N. Cass, Jay A. Rosenheim . Optimal design for dynamical modeling of pest populations. Mathematical Biosciences and Engineering, 2018, 15(4): 993-1010. doi: 10.3934/mbe.2018044
    [6] Anlu Yuan, Tieyi Zhang, Lingcong Xiong, Zhipeng Zhang . Torque control strategy of electric racing car based on acceleration intention recognition. Mathematical Biosciences and Engineering, 2024, 21(2): 2879-2900. doi: 10.3934/mbe.2024128
    [7] Yarong Yu, Liang Wang, Sanku Dey, Jia Liu . Estimation of stress-strength reliability from unit-Burr Ⅲ distribution under records data. Mathematical Biosciences and Engineering, 2023, 20(7): 12360-12379. doi: 10.3934/mbe.2023550
    [8] Jihong Pang, Chaohui Zhang, Xinze Lian, Yichao Wu . Reliability assessment of permanent magnet brake based on accelerated bivariate Wiener degradation process. Mathematical Biosciences and Engineering, 2023, 20(7): 12320-12340. doi: 10.3934/mbe.2023548
    [9] Walid Emam, Khalaf S. Sultan . Bayesian and maximum likelihood estimations of the Dagum parameters under combined-unified hybrid censoring. Mathematical Biosciences and Engineering, 2021, 18(3): 2930-2951. doi: 10.3934/mbe.2021148
    [10] Roman Czapla, Vladimir V. Mityushev . A criterion of collective behavior of bacteria. Mathematical Biosciences and Engineering, 2017, 14(1): 277-287. doi: 10.3934/mbe.2017018
  • In this paper, a new generalization of the one parameter Rayleigh distribution called the Power Rayleigh (PRD) was employed to model the life of the tested units in the step-stress accelerated life test. Under progressive Type-I interval censored data, the cumulative exposure distribution was considered to formulate the life model, assuming the scale parameter of PRD has the inverse power function at each stress level. Point estimates of the model parameters were obtained via the maximum likelihood estimation method, while interval estimates were obtained using the asymptotic normality of the derived estimators and the bootstrap resampling method. An extensive simulation study of k=4 levels of stress in different combinations of the life test under different progressive censoring schemes was conducted to investigate the performance of the obtained point and interval estimates. Simulation results indicated that point estimates of the model parameters are closest to their initial true values and have relatively small mean squared errors. Accordingly, the interval estimates have small lengths and their coverage probabilities are almost convergent to the 95% significance level. Based on the Fisher information matrix, the D-optimality and the A-optimality criteria are implemented to determine the optimal design of the life test by obtaining the optimum inspection times and optimum stress levels that improve the estimation procedures and give more efficient estimates of the model parameters. Finally, the developed inferential procedures were also applied to a real dataset.



    To estimate the reliability of highly reliable products under the normally used conditions, the tested units have a long time for failure, and only few failure times are available to make inference about their life model. To overcome this problem, the life testing experiment is conducted using the accelerated life tests (ALTs). This can be done by exposing the tested units to successively higher levels of stress including the effect of temperature, pressure, and voltage. Practically, this procedure will provide the requested lifetime's data within a faster acceptable time and with relatively low costs.

    In reliability frame works, two types of ALTs are mainly considered: The simple accelerated life test (SALT) by exposing the tested units to only one single change in stress level or exposing the tested units to more than one higher level of stress with the step-stress accelerated life test (SSALT). Using the ordinary Type-I and Type-II censored data, SALT and SSALT are devoted to estimate the parameters of many life models. Point and interval estimation are obtained for the exponential lifetime model by Balakrishnan and Han [1], Balakrishnan and Han [2] and Haghighi [3]. Lu and Rudy [4], McSorley et al. [5] and Komori [6] have considered inference in the Weibull life model. Chung and Bai [7] have explored SSALT for the log-normal distribution. Related researches are conducted also by Ebrahem and Al-Masri [8] and Bleed and Hasan [9] for the Logistic distribution, Saxena et al. [10] for the Rayleigh distribution and Ahmadi et al. [11] for the generalized half-normal distribution.

    When it is difficult to observe the exact lifetimes by continuous monitoring of the tested units using Type-I and Type-II censoring schemes, one feasible way to conduct the life test is by the intermittent inspection of the tested units in pre-determined time intervals. In this case, the only available data are the numbers of failed units within these intervals. In addition, if for some constraints the life testing experiment requests removing some units during the life test before the final termination point, then we will have what is called the progressively Type-I interval-censored data (PT-ICD). The PT-ICD consists of the numbers of failed and the numbers of the removed units within the inspection intervals. This procedure will decrease the cost and reduce the efforts requested for conducting the life testing experiment.

    Based on SSALT conducted under PT-ICD, many studies are focused on either estimation of the model parameters or in determining the optimum the life test plans. Wu et al. [12] explored the SSALT for the Exponential life model. Planning of the SSALTs are also considered in [13] for the log-normal distribution, in [14] for the Weibull distribution and in [15] for the Burr type XII distribution. Optimum designs of SSALTs are also implemented in [16] for Pareto distribution, in [17] for the Weibull Poisson distribution and in [18] for the Weibull distribution.

    As it has a growing interest in many engineering and medical applications, inference in SSALT under different censoring schemes has become a main topic in research. Examples of such research explorations include Budhiraja and Pradhan [19] for estimation of the cost model parameters and Bayesian optimum life testing plans analyzed by Roy and Pradhan [20]. Wang [21] addressed the group effects in the data analysis of SSALT with random removals. EL-Sagheer and Khder [22] have also considered estimation in a k-stage partially SSALT when the life model is the generalized Pareto distribution. Parameter estimation in partially SSALT under different types of censored data was also justified by Kamal [23].

    Based on a variety of progressive censored data, inferences of SSALT are also conducted in other lifetime models by Zhuang et al. [24], Alotaibi et al. [25], Bai et al. [26], Almarashi [27], Nassar and Elshahhat [28] and Alam et al. [29].

    The power Rayleigh distribution (PRD), introduced by Bhat and Ahmad [30], is a new extension of the standard Rayleigh distribution. To date, only a few researchers have considered this distribution for modeling survival and reliability data. To utilize this distribution in a broader range of engineering and medical applications as a lifetime model, we assume that the SSALT is conducted under PT-ICD when the life model is the PRD. This paper aims to make inferences about the parameters of the derived model. To increase the efficiency of the estimation procedures, we also approach finding both the optimum inspection times and the optimum stress levels. Unlike most of the previous studies that concentrate in equidistant inspection intervals, this work considers an arbitrary selection of time cut points.

    The rest of this paper is organized as follows: The model and the test procedures are introduced in the next section. The maximum likelihood estimators (MLEs) of the derived model parameters are obtained in Section 3. The Fisher information matrix is obtained in Section 4. Asymptotic and bootstrap confidence intervals of the model parameters are obtained in Section 5. An extensive simulation study is conducted in Section 6 to evaluate the performance of the obtained estimates. Two optimality criteria are employed in Section 7 to obtain both the optimum inspection times and the optimum stress levels. The estimation procedures are investigated in a real dataset in Section 8. Finally, Section 9 explores the general conclusions of this work.

    Using the power transformation method, Bhat and Ahmad [30] have derived a new generalization of the one parameter Rayleigh distribution called the PRD with probability density (PDF), cumulative distribution function (CDF) and reliability functions given respectively by

    f(x;θ,β)=βθ2x2β1ex2β2θ2;x>0,θ,β>0  (2.1)
    F(x;θ,β)=1ex2β2θ2;x>0,θ,β>0 (2.2)
    R(x;θ,β)=ex2β2θ2;x>0,θ,β>0 (2.3)

    The hazard rate function (HRF) of the PRD is given by

    h(x;θ,β)=βθ2x2β1;x>0,θ,β>0 (2.4)

    which is constant when the shape parameter is β=0.5, increasing if β<0.5 and decreasing if β>0.5.

    This flexibility in the hazard function indicated that PRD can be effectively used in place of the traditional Weibull, Gamma, log-normal, and other distributions to model life data.

    Additionally, PRD can be used as a general case of the standard Rayleigh distribution, which is employed in modeling the lifetimes of resisters, transformers, and capacitors (see Ateeq et al. [31] and Kundu and Raqab [32]). The PDF and the HRF of the PRD with different combinations of (θ, β) are presented in Figure 1.

    Figure 1.  The PDF and the HRF of the PRD with different combinations of (θ, β).

    Recent inference of the PRD in the analysis of survival data is explored by Tolba et al. [33] and Migdadi et al. [34].

    Consider the life-testing experiment in which n units are placed in a k-level SSALT based on progressively Type-I interval censored data. The units are run at stress level S1 until the time τ1, and within the interval (0,τ1] the number n1 of failed units and the number r1 of the removed units are recorded. Now, starting at the time τ1, the remaining (nn1r1) survival units are subject to a higher level of stress S2 and run until the time τ2, at which point the number n2 of failed units and the number r2 of the removed units within the interval (τ1,τ2] are recorded and so on. The model assumptions are described in the following constraints:

    1. There are successively k levels of higher stresses S1<S2<<Sk and k1 predetermined inspection times: τ1<τ2<<τk1<τk=, where the normal used stress is assumed to be S0=0.

    2. The termination test time is the point τk1 at which the remaining tested units are either failed or removed from the test. This implies the inspection intervals are: (0,τ1],(τ1,τ2],,(τk1,].

    3. The probabilities of the removed units rj, πj, j=1,2,,k1 are randomly determined by the experimenter with the constraint πk=1.

    4. The lifetimes of the n tested units are identically distributed according to the PRD indexed by the scale parameter θj, which has at the level stress Sj, j=1,2,,k the inverse power function given by θj=cSpj,c>0, where the parameter c is the constant of proportionality and the parameter p>0 is the power of the applied stress.

    5. The shape parameter β of the PRD is assumed to be independent of the stress levels.

    Figure 2 presents the k-Level SSAT based on PT-ICD.

    Figure 2.  Schematic presentation of the the k-Level SSAT based on PT-ICD.

    From the previous assumptions and since failures have occurred according to the cumulative exposure model, this implies, as shown by Nelson [35], that the CDF of the lifetimes of the tested units is given by

    F(t)={F(t;θ1)0<tτ1F(tτ1+u1;θ2)τ1<tτ2F(tτk1+uk1;θk)τk1t<, (2.5)

    where ui is the solution to the equation:

    F(uj,θj+1)=F(τj+uj1,θj).

    From Eq (2.2), the general solution of uj is given by:

    uj=(θj+1θj)12β(τj+uj1),j=1,2,,k.

    Now, substituting for θj=cSpj, j=1,2,,k in Eq (2.5) implies

    uj=(Sj+1Sj)pβ(τj+uj1).

    Therefore, The CDF and the PDF of the lifetimes of the tested units are given, respectively, by:

    F(t)={1et2β2c2S2p10<tτ11e(tτj1+j1i=1(SjSi)pβτi)2β2c2S2pjτj1<tτj,j=2,,k (2.6)
    f(t)={2βt2β1c2S2p1et2β2c2S2p1 0<tτ1β(tτj1+j1i=1(SjSi)pβτi)2β1)c2S2pje(tτj1+j1i=1(SjSi)pβτi)2β2c2S2pjτj1<t<τj. (2.7)

    Based on PT-ICD and following Wu et al. [36] and Aly [37], the likelihood function for the parameters (c,p,β) has the following form:

    L(c,p,β)kj=1(F(τj)F(τj1))nj(1F(τj))rj. (3.1)

    Substituting for F(τj),j=1,2,,k, from equation (2.1), the likelihood function can be written as:

    L(c,p,β)L1(c,p,β)×L2(c,p,β),

    where:

    L1(c,p,β)=(1eτ12β2c2S2p1)n1er1τ2β12c2S2p1
    L2(c,p,β)=kj=2[e(τj1τj2+j2i=1(SjSi)pβτi)2β2c2S2pj1 e(τjτj1+j1i=1(SjSi)pβτi)2β2c2S2pj]njerj(τjτj1+j1i=1(SjSi)pβτi)2β2c2S2pj.

    This implies that the log likelihood function is given by:

    ln(L(c,p,β))=n1ln(1eτ12β2c2S2p1)r1τ2β12c2S2p1kj=2rj(τjτj1+j1i=1(SjSi)pβτi)2β2c2S2pj+kj=2njln[e(τj1τj2+j2i=1(SjSi)pβτi)2β2c2S2pj1e(τjτj1+j1i=1(SjSi)pβτi)2β2c2S2pj] (3.2)
    Set: Y1=τ2β12c2S2p1 ,Yj=(τjτj1+j1i=1(SjSi)pβτi)2β2c2Spj,j=1,,k,

    then, ln(L(c,p,β)) can be simplified as:

    ln(L(c,p,β ))=n1ln(1eY1 )r1Y1kj=2rjYj+kj=2njln((eYj1eYj). (3.3)

    Taking the first partial derivatives of Eq (3.3) with respect to c, p and β and then equating each to zero, we have the following normal equations:

    ln(L(c,p,β))c=2c(n1Y1eY1(1eY1)r1Y1kj=2rjYj+kj=2nj(YjeYjYj1eYj1)eYj1eYj)=0 (3.4)
    ln(L(c,p,β))p=2pln(S1)(n1Y1eY11eY1r1Y1)+2kj=2rj(Zjhjln(Sj))Yj+2kj=2nj(Zjhjln(Sj))YjeYj(Zj1hj1ln(Sj1))Yj1eYj1eYj1eYj=0 (3.5)
    ln(L(c,p,β))β=2βln(τ1)(n1Y1eY11eY1r1Y1)2kj=2rj(ln(hj)pZjβhj)Yj+2kj=2nj(ln(hj)pZjβhj)YjeYj(ln(hj1)pZj1βhj1)Yj1eYj1eYj1eYj=0, (3.6)

    where

    hj=(τjτj1+j1i=1( SjSi)pβτi),Zj=j1i=1ln(SjSi)( SjSi)pβτi,j=2,,k.

    The system of nonlinear Eqs (3.4)–(3.6) cannot be solved analytically. Therefore, the Newton-Raphson numerical iteration method can be employed to approximate the MLEs of the unknown parameters (c, p, β) denoted as (ˆc, ˆp, ˆβ).

    To quantify the information contained in the data obtained from a life test plan in the long run, the Fisher information matrix is frequently used. The Fisher information matrix is defined by the negative expectation of the second derivatives of the log-likelihood function as:

    F=E[2ln(L(c,p,β))c22ln(L(c,p,β))cp2ln(L(c,p,β))cβ2ln(L(c,p,β))cp2ln(L(c,p,β))p22ln(L(c,p,β))pβ2ln(L(c,p,β))cβ2ln(L(c,p,β))pβ2ln(L(c,p,β))β2]. (4.1)

    To find the elements of F, we have to find the expectations of nj, rj, j=1,2,,k and the second derivatives of the log-likelihood function.

    Based on the multinomial distribution, the numbers of failures are distributed as:

    njnj1,,n1,rj1,,r1Bin(nj1i=1(njrj),Pj),j=1,,k,

    where Pj is the conditional probability of failure in the interval (τj1, τj) given of survival at τj1 given by

    Pj=P(τj1<TτjT>τj1)=F(τj)F(τj1)1F(τj1)
    ={1eτ12β2c2S2p1,j=1WjWj11Wj1,j=2,,k},

    where Wj=e(τjτj1+j1i=1(SjSi)pβτi)2β2c2S2pj,j=2,,k.

    This implies that the numbers of removals are distributed as:

    rjnj,,n1,rj1,,r1Bin(nji=1nij1i=1ri,πj).

    Therefore,

    E(n1)=n(1eτ12β2c2S2p1) (4.2)
    E(r1)=(nE(n1))π1. (4.3)

    By induction,

    E(nj)={nj1i=1(E(ni)E(ri))}(1WjWj1),j=2,,k (4.4)
    E(rj)={nji=1E(ni)j1i=1E(ri)}πj, j=2,,k. (4.5)

    Since ln(L(c,p,β))=ln(L1(c,p,β))+ln(L2(c,p,β)), we set

    U=τ2β12c2S2p1 then ln(L1(c,p,β))=n1ln(1eU)+r1U.

    This implies

    2L1c2=n1(1eU)2{(eU(1eU)(2Uc2+(Uc)2))+(eU(Uc)2)}+r12Uc2. (4.6)

    Hence, 2L1p2, 2L1β2 are obtained by substituting p, β in place of c in equation (4.6). To find the second derivatives of U with respect to (c, p, β), see Appendix A.1, and to find the second derivatives of ln(L2(c,p,β)) with respect to c, p, β, set:

    fj=(τjτj1+j1i=1( SjSi)pβτi)2β2c2S2pj,j=1,,k, Pj=efj1efj ,j=2,,k.

    This implies

    L2=kj=2njln(Pj)+kj=2rjfj,

    and therefore,

     2L2c2=kj=2nj(2Pjc2Pj(Pjc)2P2j)+kj=2rj2fjc2 (4.7)
    2L2cp=kj=2nj(2PjcpPj(Pjc)(Pjp)P2j)+kj=2rj2fjcp, (4.8)

    where 2L2p2, 2L2β2 can be obtained by substituting p, β in place of c in equation (4.7), respectively. Also, 2L2cβ, 2L2pβ can also be obtained by replacing β in place of p and in place of c in equation (4.8), respectively.

    To find the second derivatives of Pj, j=2,,k and fj, j=1,,k with respect to c, p and β, see Appendix A.2.

    Now, substituting for E(nj) and E(rj), j=1,,k from the Eqs (4.4)–(4.5) and the corresponding second derivatives of the log-likelihood function from the Eqs (4.6)–(4.8) in Eq (4.1), we will have the elements of the Fisher information matrix (F).

    Another desired inference is to obtain interval estimates for the model parameters. In this section, two methods are employed for constructing confidence intervals (CIs) for the parameters c, p and β. The first method is based on the asymptotic normality of the MLEs, while the second method is based on the bootstrap resampling technique.

    The estimated variance-covariance matrix ˆV for the parameters (c, p, β) is given by:

    ˆV=[ˆv11ˆv12ˆv13ˆv21ˆv22ˆv23ˆv31ˆv32ˆv33]=(F(ˆc,ˆp,ˆβ))1,

    where F(ˆc,ˆp,ˆβ) is the Fisher information matrix given the estimated values of (ˆc,ˆp,ˆβ). Consequently, based on the large sample theory of the MLEs (see Lawless [38]) the approximate (asymptotic) 100(1α)% confidence intervals (ACIs) for the parameters (c,p,β) are obtained respectively as:

    ˆc±zα2ˆv11,ˆp±zα2ˆv22,ˆβ±zα2ˆv33,

    where zα2 is the (1α)% quantile of the standard normal distribution and ˆv11, ˆv22 and ˆv33 are the diagonal elements of ˆV.

    The bootstrap resampling technique is a common method to construct confidence intervals for the unknown parameters and provides more efficient estimation than conventional asymptotic intervals for relatively small sample sizes. For more details, refer to Hall [39] and Efron and Tibshirani [40]. Given initial values of the parameters (c,p,β), initial stress levels, initial inspection times and a progressive censoring scheme (PCs), the procedure to obtain interval estimates for the parameters (c,p,β) by this method is described in the following steps:

    Step 1: Compute (ˆc,ˆp,ˆβ)

    Step 2 Using (ˆc,ˆp,ˆβ), generate a bootstrap resample from equation (2.6).

    Step 3: Calculate the bootstrap estimates (ˆc1,ˆp1,ˆβ1).

    Step 4: Repeat steps two to three up to k times to obtain: ˆcB1,ˆcB2,,ˆcBk,ˆpB1,ˆpB2,,ˆpBk,ˆβB1,ˆβB2,,ˆβBk.

    Step 5: Rearrange the bootstrap estimates in ascending order to have 100(1α)% percentile confidence intervals for the parameters (c,p,β), respectively, as:

    (cB[kα2],cB[k(1α2)]), (pB[kα2],pB[k(1α2)]), (βB[kα2],βB[k(1α2)]).

    This study aims to evaluate the performance of the obtained point and interval estimates of the model parameters. For this purpose, a Monte Carlo simulation of 1000 repetitions is conducted for the SSALT with k=4 levels under PT-ICD assuming PRD is the underlying life model of the tested units. The performance of the ML point estimates is measured in terms of their average absolute biases (ABs) and their mean squared errors (MSEs), while the performance of interval estimates is measured in terms of their average lengths (ALs) and their coverage probabilities (CPs). Different settings in the simulation processes are considered as listed in Table 1. The simulation process is performed using MATLAB9.7R2019 software code through the following steps:

    Table 1.  The indices used in the simulation study.
    Sample size (n) Parameters true values Stress levels inspection times PSC
    30 β= 1.2 S1=0.3 τ1=0.4 PSC1:
    (π1=π2= π3=0,π4=1)
    50 c= (2,2.5) S2=0.5 τ2=1.0 PSC2:
    (π1= π2=π3=0.05π4=1)
    100 p= (0.3,0.8) S3=1.0 τ3=1.25 PSC3:
    (π1=π2=π3= 0.10,π4=1)
    S4=1.3 τ4=1.5

     | Show Table
    DownLoad: CSV

    Step 1: Generate a random sample of size n from the uniform distribution U(0,1).

    Step 2: Using the inverse-CDF method and the random sample generated in step one, generate a random sample of size n from model (2.6) based on the true parameter values, initial stress values Sj, τj, j=1,2,,4 and initial inspection times τj, j=1,2,,4.

    Step 3: Based on the indicated progressively censored scheme (PSC), compute the number of failures nj and the number of removals rj, j=1,2,,4.

    Step 4: Obtain the MLEs of the parameters (c,p,β).

    Step 5: Based on Step 4, obtain the Fisher information matrix.

    Step 6: Obtain the 95% asymptotic confidence intervals.

    Step 7: Repeat Steps (1–4) 1000 times to obtain the 95% bootstrap confidence intervals.

    Step 8: Repeat Steps (1–7) 1000 times and compute ABs and MSEs of the MLEs for the parameters (c,p,β), as well as ALs and CPs of the interval estimates.

    The ABs and the MSEs of the MLE point estimates are listed in Table 2 and the ALs and the CPs of the 95% asymptotic and bootstrap interval estimates are listed in Tables 34.

    Table 2.  The ABs and the MSEs of point estimates under different PSCs.
    Parameters c=2 p=0.3 β=1.2
    n PSC AB MSE AB MSE AB MSE
    30 PSC1 0.2083 0.0433 0.0472 0.0023 0.1158 0.0138
    PSC2 0.2456 0.0602 0.0658 0.0045 0.1502 0.0226
    PSC3 0.2781 0.0775 0.0739 0.0057 0.1686 0.0287
    50 PSC1 0.1964 0.0401 0.0392 0.0019 0.1005 0.0122
    PSC2 0.2382 0.0587 0.0566 0.0035 0.1339 0.0207
    PSC3 0.2543 0.0745 0.0605 0.0048 0.1456 0.0232
    100 PSC1 0.1603 0.0367 0.0312 0.0011 0.0965 0.0108
    PSC2 0.2022 0.0508 0.0517 0.0028 0.1288 0.0175
    PSC3 0.2268 0.0572 0.0580 0.0035 0.1391 0.0206
    c=2 p=0.8 β=1.2
    30 PSC1 0.2105 0.0462 0.0574 0.0039 0.1223 0.0143
    PSC2 0.2472 0.0631 0.0763 0.0061 0.1576 0.0231
    PSC3 0.2803 0.0804 0.0841 0.0073 0.1756 0.0292
    50 PSC1 0.1987 0.0439 0.0494 0.0035 0.1078 0.0127
    PSC2 0.2402 0.0616 0.0668 0.0051 0.1407 0.0212
    PSC3 0.2564 0.0774 0.0707 0.0064 0.1525 0.0237
    100 PSC1 0.1621 0.0396 0.0414 0.0027 0.1033 0.0113
    PSC2 0.2046 0.0537 0.0619 0.0044 0.1351 0.0182
    PSC3 0.2285 0.0601 0.0682 0.0051 0.1462 0.0211
    c=2.5 p=0.3 β=1.2
    30 PSC1 0.2188 0.0494 0.0483 0.0032 0.1257 0.0218
    PSC2 0.2561 0.0663 0.0669 0.0054 0.1594 0.0306
    PSC3 0.2886 0.0836 0.0758 0.0066 0.1778 0.0367
    50 PSC1 0.2069 0.0471 0.0403 0.0028 0.1097 0.0202
    PSC2 0.2487 0.0648 0.0577 0.0044 0.1431 0.0287
    PSC3 0.2648 0.0806 0.0616 0.0057 0.1548 0.0312
    100 PSC1 0.1708 0.0428 0.0323 0.0023 0.1057 0.0188
    PSC2 0.2127 0.0569 0.0528 0.0037 0.1381 0.0255
    PSC3 0.2373 0.0633 0.0591 0.0044 0.1483 0.0286
    c=2.5 p=0.8 β=1.2
    30 PSC1 0.3135 0.0515 0.0601 0.0051 0.1316 0.0252
    PSC2 0.3502 0.0684 0.0796 0.0073 0.1669 0.0340
    PSC3 0.3833 0.0857 0.0868 0.0085 0.1849 0.0401
    50 PSC1 0.3017 0.0492 0.0521 0.0047 0.1171 0.0236
    PSC2 0.3432 0.0669 0.0695 0.0063 0.1503 0.0321
    PSC3 0.3594 0.0827 0.0734 0.0076 0.1618 0.0346
    100 PSC1 0.2651 0.0449 0.0441 0.0039 0.1126 0.0222
    PSC2 0.3076 0.0590 0.0646 0.0056 0.1444 0.0289
    PSC3 0.3315 0.0654 0.0709 0.0063 0.1555 0.0320

     | Show Table
    DownLoad: CSV
    Table 3.  The ALs and the CPs of the 95% asymptotic confidence intervals.
    Parameters c=2 p=0.3 β=1.2
    n PSC AL CP AL CP AL CP
    30 PSC1 0.8156 0.94 0.1875 0.93 0.4601 0.95
    PSC2 0.9617 0.97 0.2622 0.96 0.5888 0.96
    PSC3 1.0912 0.98 0.2951 0.97 0.6635 0.97
    50 PSC1 0.7849 0.93 0.1704 0.93 0.4326 0.94
    PSC2 0.9497 0.96 0.2313 0.95 0.5635 0.96
    PSC3 1.0699 0.98 0.2708 0.97 0.5968 0.96
    100 PSC1 0.7509 0.93 0.1296 0.92 0.4070 0.94
    PSC2 0.8835 0.95 0.2068 0.94 0.5181 0.95
    PSC3 0.9375 0.96 0.2313 0.95 0.5621 0.96
    c=2 p=0.8 β=1.2
    30 PSC1 0.8425 0.95 0.2446 0.95 0.4670 0.96
    PSC2 0.9846 0.97 0.3059 0.98 0.5936 0.97
    PSC3 1.1115 0.98 0.3346 0.98 0.6674 0.98
    50 PSC1 0.8213 0.94 0.2317 0.95 0.4401 0.96
    PSC2 0.9729 0.97 0.2797 0.97 0.5687 0.97
    PSC3 1.0905 0.98 0.3133 0.98 0.6013 0.97
    100 PSC1 0.7800 0.93 0.2035 0.94 0.4152 0.96
    PSC2 0.9083 0.96 0.2598 0.96 0.5269 0.97
    PSC3 0.9616 0.97 0.2797 0.97 0.5673 0.97
    c=2.5 p=0.3 β=1.2
    30 PSC1 0.8669 0.94 0.2217 0.94 0.5774 0.96
    PSC2 1.0042 0.98 0.2880 0.97 0.6841 0.97
    PSC3 1.1277 0.98 0.3184 0.98 0.7492 0.98
    50 PSC1 0.8464 0.94 0.2074 0.94 0.5558 0.96
    PSC2 0.9928 0.97 0.2600 0.96 0.6625 0.97
    PSC3 1.1072 0.98 0.2959 0.97 0.6908 0.97
    100 PSC1 0.8069 0.94 0.1879 0.93 0.5362 0.96
    PSC2 0.9303 0.96 0.2384 0.95 0.6245 0.97
    PSC3 0.9812 0.97 0.2601 0.96 0.6614 0.97
    c=2.5 p=0.8 β=1.2
    30 PSC1 0.8883 0.95 0.2792 0.97 0.6214 0.97
    PSC2 1.0237 0.98 0.3340 0.98 0.7216 0.98
    PSC3 1.1459 0.98 0.3604 0.99 0.7836 0.99
    50 PSC1 0.8683 0.95 0.2680 0.96 0.6014 0.97
    PSC2 1.0125 0.98 0.3103 0.98 0.7012 0.98
    PSC3 1.1257 0.99 0.3408 0.98 0.7279 0.98
    100 PSC1 0.8295 0.94 0.2441 0.95 0.5833 0.96
    PSC2 0.9508 0.96 0.2925 0.97 0.6654 0.97
    PSC3 1.0012 0.98 0.3103 0.98 0.7001 0.97

     | Show Table
    DownLoad: CSV
    Table 4.  The ALs and the CPs of the 95% bootstrap confidence intervals.
    Parameters c=2 p=0.3 β=1.2
    n PSC AL CP AL CP AL CP
    30 PSC1 0.8049 0.95 0.1766 0.93 0.4291 0.94
    PSC2 0.9510 0.97 0.2513 0.96 0.5578 0.96
    PSC3 1.0805 0.98 0.2842 0.97 0.6325 0.97
    50 PSC1 0.7742 0.93 0.1595 0.92 0.4016 0.94
    PSC2 0.9390 0.96 0.2204 0.94 0.5325 0.96
    PSC3 1.0592 0.98 0.2599 0.96 0.5656 0.96
    100 PSC1 0.7402 0.93 0.1184 0.88 0.3760 0.93
    PSC2 0.8728 0.95 0.1959 0.93 0.4871 0.94
    PSC3 0.9268 0.96 0.2204 0.94 0.5311 0.96
    c=2 p=0.8 β=1.2
    30 PSC1 0.8317 0.95 0.2337 0.95 0.4355 0.94
    PSC2 0.9738 0.97 0.2950 0.97 0.5621 0.96
    PSC3 1.1007 0.98 0.3237 0.98 0.6359 0.97
    50 PSC1 0.8105 0.94 0.2208 0.94 0.4086 0.93
    PSC2 0.9621 0.97 0.2688 0.96 0.5372 0.96
    PSC3 1.0797 0.98 0.3024 0.98 0.5698 0.96
    100 PSC1 0.7692 0.93 0.1926 0.93 0.3837 0.92
    PSC2 0.8975 0.95 0.2489 0.95 0.4954 0.95
    PSC3 0.9502 0.96 0.2688 0.96 0.5358 0.96
    c=2.5 p=0.3 β=1.2
    30 PSC1 0.8566 0.95 0.1137 0.88 0.5467 0.96
    PSC2 0.9939 0.97 0.1805 0.93 0.6534 0.97
    PSC3 1.1174 0.98 0.2104 0.94 0.7185 0.98
    50 PSC1 0.8361 0.95 0.0994 0.85 0.5251 0.96
    PSC2 0.9825 0.97 0.1520 0.97 0.6318 0.97
    PSC3 1.0969 0.98 0.1879 0.93 0.6601 0.97
    100 PSC1 0.7966 0.93 0.0799 0.89 0.5055 0.95
    PSC2 0.9200 0.96 0.1304 0.85 0.5938 0.96
    PSC3 0.9709 0.97 0.1527 0.87 0.6307 0.97
    c=2.5 p=0.8 β=1.2
    30 PSC1 0.8778 0.95 0.2686 0.96 0.5908 0.96
    PSC2 1.0132 0.98 0.3235 0.97 0.6910 0.97
    PSC3 1.1354 0.98 0.3502 0.99 0.7535 0.98
    50 PSC1 0.8578 0.95 0.2576 0.96 0.5702 0.96
    PSC2 1.0020 0.98 0.2997 0.97 0.6706 0.96
    PSC3 1.1152 0.98 0.3304 0.98 0.6973 0.97
    100 PSC1 0.8196 0.94 0.2337 0.95 0.5527 0.95
    PSC2 0.9403 0.96 0.2821 0.97 0.6341 0.97
    PSC3 0.9905 0.97 0.2959 0.97 0.6675 0.97

     | Show Table
    DownLoad: CSV

    From the simulation results listed in Tables 24, the following concluding remarks are clearly observed:

    1. The obtained point estimates have satisfactory performance since they have relatively small ABs and MSEs under the three PCs.

    2. The ABs and the MSEs of the point estimates are uniformly decreasing with an increase in the sample size and increasing with an increase in their indicated initial values.

    3. The ABs and the MSEs of the point estimates are increasing as the number of removals is increasing. This is expected because point estimates obtained under the conventional Type-II (PC1) censored scheme have more information about the failure process than point estimates obtained under the PC2 and PC3 (progressively censored schemes).

    4. The ALs of the asymptotic and bootstrap 95% confidence intervals are also decreasing with an increase in the sample size and increasing with an increase in their indicated initial values.

    5. The bootstrap 95% confidence intervals have smaller ALs than their corresponding 95% asymptotic confidence intervals. (The intuitive reason for this result is that the bootstrap technique can be considered a modification of the ML method).

    6. Generally, the CPs of the obtained interval estimates exceed the nominal 95% significance level, but for interval estimates with extremely small ALs, their CPs are strictly less than the 95% significance level because, mathematically, the conditional CPs of the interval estimates are decreasing with a decrease in their ALs.

    This section explores two criteria to determine the optimum inspection times and the optimum stress levels that lead to the most precision in the estimation of the model parameters. Both criteria are based on the Fisher information matrix since the determinant of the Fisher information matrix is proportional to the reciprocal of the volume of the joint confidence region of the parameters (c,p,β). Therefore, one way to minimize the asymptotic variances of (ˆc,ˆp, ˆβ) is to maximize the determinant of the Fisher information matrix with respect to τ=(τ1,τ2,,τk)t and =(S1,S2,,Sk)t. In reliability contexts, such a criterion is called the D-optimality criterion.

    Given the proposed values of the progressively censored scheme π=(π1,π2,,πk), and the initial values of stress levels, applying this criterion yields the optimal values of τ=(τ1,τ2,,τk)t as solutions to:

    |ˆF|τ=0, (7.1)

    where |ˆF| is the determinant of the Fisher information matrix F computed at (ˆc,ˆp, ˆβ).

    Consequently, after determining the optimum inspection times τ=(τ1,τ2,,τk)t, the optimum stress levels: S=(S1,S2,,Sk)t are obtained simultaneously as the solutions of the equation:

    |ˆF|S=0. (7.2)

    Another criterion, which is called the A-optimality criterion, is based on the trace of the Fisher information matrix. Applying this criterion, the optimum inspection times are determined by maximizing the sum of the diagonal elements of the Fisher information matrix with respect to τ=(τ1,τ2,,τk)t and S=(S1,S2,,Sk)t through solving the following two nonlinear equations:

    (ki=1ˆFii)τ=0 (7.3)
    (ki=1ˆFii)S=0. (7.4)

    More details about the D-optimality and the A-optimality criteria are explored in Gouno et al. [41]. Figure 3 presents the optimum inspection times and optimum stress levels when the sample size n=50 with different true values of the model parameters. Table 5 lists the ABs and the MSEs of the point estimates when the true parameter values are (c=2,p=0.3,β=1.2), obtained using the optimum inspection times and optimum stress levels by applying the simulation process in the previous section when the sample size n=50.

    Figure 3.  Optimum inspection times and stress levels at different values of c and p.
    Table 5.  The ABs and the MSEs of the point estimates obtained using the optimum inspection times and stress levels when n=50.
    Parameters c=2 p=0.3 β=1.2
    Criterion PSC AB MSE AB MSE AB MSE
    D-Optimality PSC1 0.1436 0.0208 0.0274 0.0008 0.0876 0.0093
    A-Optimality 0.1572 0.0249 0.0309 0.0012 0.0963 0.0102
    D-Optimality PSC2 0.1876 0.0352 0.0465 0.0017 0.1185 0.0104
    A-Optimality 0.2031 0.0416 0.0526 0.0023 0.1277 0.0176
    D-Optimality PSC3 0.2081 0.0442 0.0538 0.0026 0.1292 0.0188
    A-Optimality 0.2264 0.0518 0.0589 0.0038 0.1387 0.0219

     | Show Table
    DownLoad: CSV

    From Figure 3 and the results in Table 5, it is clear that:

    1. The optimum inspection times and optimum stress levels increase as the true values of the model parameters increase.

    2. The optimum inspection times obtained by PC1 are strictly greater than those obtained by PC2 and PC3. This implies that removing some units during the life test will shorten the termination time of the life testing experiment.

    3. The optimum inspection times and optimum stress levels obtained by the D-optimality criterion are greater than their corresponding values obtained by the A-optimality criterion.

    4. It is noted that the ABs and the MSEs of the point estimates obtained using the optimum inspection times and optimum stress levels by the D-optimality criterion are strictly lower than their corresponding values obtained by the A-optimality criterion.

    5. Additionally, the ABs and the MSEs of the point estimates obtained using the optimum inspection times and optimum stress levels by both criteria are lower than their corresponding values in the simulation study. This confirms that the optimum inspection times and optimum stress levels significantly improve the point estimates of the model parameters.

    In this section, the developed estimation procedures are applied to a real-life data. This dataset consists of 66 observations representing the breaking stress of carbon fibers of 50 mm length (measured in GPa), as reported in Al-Aqtash et al. [42] and reanalyzed by Bhat and Ahmad [30]. Assuming the PRD modeling for this data, the MLEs of the parameters θ and β are ˆθ=4.850, ˆβ=1.721 and the asymptotic 95% confidence intervals for θ and β are (2.818,6.883) and (1.396,2.045), respectively. The estimated value of Kolmogorov-Smirnov test statistics is (K-S) = 0.08 with a p-value = 0.768. This confirms validity of the PRD to this dataset.

    Consider a SSALT with k=5 levels of stress. This data is divided into 5 groups, each consisting of 10 observations by the initial inspection times τ1=1.81,τ2=2.46,τ3=2.75,τ4=3,τ5=3.3, as it observed in Table 6.

    Table 6.  The interval grouped presentation of the breaking stress data.
    Groups
    Group 1 0.39 0.85 1.08 1.25 1.47 1.57 1.61 1.61 1.69 1.80
    Group 2 1.84 1.87 1.89 2.03 2.03 2.05 2.12 2.35 2.41 2.43
    Group 3 2.48 2.50 2.53 2.55 2.55 2.56 2.59 2.67 2.73 2.74
    Group 4 2.79 2.81 2.82 2.85 2.87 2.88 2.93 2.95 2.96 2.97
    Group 5 3.09 3.11 3.11 3.15 3.15 3.19 3.22 3.22 3.27 3.28

     | Show Table
    DownLoad: CSV

    Therefore, the termination time of the life test is τ5=3.3, thus we will have 16 units still working and removed from the step stress test after the fifth stage.

    Suppose the scale parameter of the PRD, θj,j=1,2,,5 at each stress level has the inverse power low given by θj=θj=cSpj,c>0. Setting the initial value of the first stress level (S1=1), initial values of Sj,j=2,3,...,5 are indicated such that:

    S(j+1)Sj=θ(j+1)θj,j=2,3,4,5.

    This gives: S2=1.9197,S3=2.6985,S4=3.2406,S5=5.9292.

    Using the progressively censored schemes:

     PSC1: (π1=π2=π3=0.10,π4=1) PSC2: (π1=π2=π3=0.20,π4=1)

    Employing the D-Optimality and the A-Optimality criteria, MLEs of the model parameters and 95% asymptotic and bootstrap CIs with optimum inspection times and optimum stress levels are listed in Table 7.

    Table 7.  Estimation results of the breaking stress data.
    parameters Optimum inspection times Optimum stress levels
    ˆc ˆp ˆβ
    PC1
    D-optimality MLE 0.180 1.514 1.712 τ1=1.79 s1=0.98
    Asymptotic CI (0.144, 0.231) (1.227, 2.116) (1.365, 2.051) τ2=2.44 s2=2.65
    Bootstrap CI (0.151, 0.236) (1.309, 2.097) (1.385, 1.997) τ3=2.75 s3=2.89
    τ4=2.98 s4=3.16
    τ5=3.27 s5=5.74
    A-optimality MLE 0.168 1.503 1.709 τ1=1.72 s1=0.94
    Asymptotic CI (0.143, 0.238) (1.206, 2.172) (1.355, 2.11) τ2=2.41 s2=2.61
    Bootstrap CI (0.146, 0.244) (1.291, 2.176) (1.377, 2.062) τ3=2.73 s3=2.78
    τ4=2.91 s4=3.08
    τ5=3.19 s5=5.64
    PC2
    D-optimality MLE 0.176 1.509 1.707 τ1=1.70 s1=0.96
    Asymptotic CI (0.140, 0.241) (1.176, 2.208) (1.306, 2.102) τ2=2.35 s2=2.49
    Bootstrap CI (0.141, 0.253) (1.275, 2.188) (1.342, 2.092) τ3=2.68 s3=2.73
    τ4=2.85 s4=3.96
    τ5=3.18 s5=5.17
    A-optimality MLE 0.166 1.497 1.688 τ1=1.63 s1=0.93
    Asymptotic CI (0.137, 0.249) (1.165, 2.211) (1.292, 2.114) τ2=2.27 s2=2.44
    Bootstrap CI (0.136, 0.242) (1.282, 2.157) (1.337, 2.102) τ3=2.66 s3=2.68
    τ4=2.80 s4=3.87
    τ5=3.14 s5=5.08

     | Show Table
    DownLoad: CSV

    From the above table, it is clearly observed that the obtained estimates are closest to their proposed initial values, agreed with the simulation results, and gave efficient estimates for the model parameters.

    This paper dealt with the k-level SSALT using PT-ICD as an extension of the traditional Rayleigh distribution. The PRD was considered to model the lifetimes of the tested units. The cumulative exposure distribution is considered to construct the lifetime model, assuming that the scale parameter of the PRD has the inverse power function of the stress level. Point estimates of the formulated model were obtained via the MLE approach. Additionally, interval estimates of the model parameters were obtained using the approximated variance-covariance matrix and the bootstrap resampling method.

    To evaluate the performance of the obtained estimates, a simulation study of different combinations of life testing experiments was conducted under three PCs. The results of this study indicate that all obtained estimates have good characteristics and are efficient for estimating the model parameters.

    In order to increase the accuracy of the obtained estimates, the test procedures were improved by obtaining the optimum inspection times and the optimum stress levels using the derived Fisher information matrix with two optimality criteria: The D-optimality criterion and the A-optimality criterion. The concluding results confirmed that the optimum inspection times and the optimum stress levels increase the efficiency of the obtained estimates with priority to the D-optimality criterion.

    The developed estimation procedures also investigated in a real stress data. In this investigation, concluding results confirmed that the obtained estimates are closest to their initial practical setting and gave an efficient estimation for the model parameters.

    In this work, the obtained results also indicate that both point and interval estimates obtained under the ordinary Type-II PC1 censoring scheme are more accurate than their corresponding values obtained under the PC2 and PC3 progressive censoring schemes. One explanation for this result is that, under PC1, there will be no removed survival units in the inspected intervals except for the last one. However, when employing PC2 and PC3 progressive censoring schemes, the life testing experiment will terminate in a reasonable time. Therefore, the experimenters have to quantify the test plans based on the intended objectives and available conditions of the life testing experiment.

    As a new extension of the standard Rayleigh distribution, concluding results confirmed that PRD can be effectively used in the step stress-accelerated life tests for more analysis of reliability and survival data. Future research may extend this work, taking into account other estimating methods and acceptance sampling plans.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    The authors would like to express their sincere gratitude to the anonymous referees for their insightful recommendations and remarks.

    The authors declare there is no conflict of interest.

    To find the second derivatives of L1(c,p,β) with respect to the parameters c, p, and β, set:

    U=τ2β12c2S2p1

    then

    2Uc2=3τ2β1c4S2p1
    2Ucβ=2βln(τ1)τ2β1c3S2p1
    2Ucp=2pln(S1)τ2β1c3S2p1
    2Uβ2=ln(τ1)τ2β1c2S2p1(2β2ln(τ1)+1)
    2Up2=ln(S1)τ2β1c2S2p1(12p2ln(S1))
    2Upβ=2βpln(τ1)ln(S1)τ2β1c2S2p1

    To find the second derivatives of L2(c,p,β) with respect to the parameters c, p, and β, set

    fj=(τjτj1+j1i=1( SjSi)pβτi)2β2c2S2pj,j=1,,k,
    Pj=efj1efj ,j=2,,k.

    This implies

    2Pjc2=efj1((fj1c)2+2fj1c2)+efj((fjc)2+2fjc2). (A1)

    Similarly, (2Pjp2,2Pjβ2) are obtained by substituting p, β in place of c in Eq (A1).

    2Pjcp=efj1(2fj1cp+fj1cfj1p)+efj(2fjcp+fjcfjp). (A2)

    2Pjcβ,2Pjβp can be also obtained by substituting β in place of p in Eq (A2) and β in place of c in Eq (A2), respectively. So, it is enough to find the second derivatives of fj, j=1,,k with respect to c, p, β as:

    2fjp2=u2βjc2({zjβuj+Z2juj22ln(Sj)uj}ln(Sj)(Zjβuj2ln(Sj))),
    2fjc2=3(τjτj1+j1i=1(SjSi)pβτi)2βc4S2pj,
    2fjβ2=1c2Sj2p({pZjuj2β1β(pzjβ2Zjp(2β1)Zjβ2uj+ln(uj)1β)}puj2ββ2(Zj+ln(uj)Zj)),
    2fjpβ=u2βjc2Sj2p{(Zj((2β1)Zjβuj2+zjβZjuj2ln(Sj)uj))+2ln(Sj)(Zjujln(Sj))}

    Where:

    uj=(τjτj1+j1i=1(SjSi)pβτi), Zj=j1i=1(SjSi)pβln(SjSi)τi, zj=j1i=1(SjSi)pβ(ln(SjSi))2τi.


    [1] N. Balakrishnan, D. Han, Exact inference for a simple step-stress model with competing risks for failure from exponential distribution under Type-II censoring, J. Stat. Plan. Infer., 138 (2008), 4172–4186. https://doi.org/10.1016/j.jspi.2008.03.036 doi: 10.1016/j.jspi.2008.03.036
    [2] N. Balakrishnan, D. Han, Optimal step-stress testing for progressively Type-I censored data from exponential distribution, J. Stat. Plan. Infer., 139 (2009), 1782–1798. https://doi.org/10.1016/j.jspi.2008.05.030 doi: 10.1016/j.jspi.2008.05.030
    [3] F. Haghighi, Optimal design of accelerated life tests for an extension of the exponential distribution, Reliab. Eng. Syst. Safe., 131 (2014), 251–256. https://doi.org/10.1016/j.ress.2014.04.017 doi: 10.1016/j.ress.2014.04.017
    [4] M. W. Lu, R. J. Rudy, Step-stress accelerated test, Int. J. Mater. Prod. Tech., 17 (2002), 425–434. https://doi.org/10.1504/ijmpt.2002.005468 doi: 10.1504/ijmpt.2002.005468
    [5] E. O. McSorley, J. C. Lu, C. S. Li, Performance of parameter-estimates in step-stress accelerated life-tests with various sample-sizes, IEEE Trans. Reliab., 51 (2002), 271–277. https://doi.org/10.1109/tr.2002.802888 doi: 10.1109/tr.2002.802888
    [6] Y. Komori, Properties of the Weibull cumulative exposure model, J. Appl. Stat., 33 (2006), 17–34. https://doi.org/10.1080/02664760500389475 doi: 10.1080/02664760500389475
    [7] W. Chung, D. S. Bai, Optimal designs of simple step-stress accelerated life tests for lognormal lifetime distributions, Int. J. Reliab. Quality Safety Eng., 5 (1998), 315–336. https://doi.org/10.1142/s0218539398000285 doi: 10.1142/s0218539398000285
    [8] M. A. H. Ebrahem, A. Q. Al-Masri, Optimum simple step-stress plan for log-logistic cumulative exposure model, Metron-Int. J. Stat., 65 (2007), 23–34.
    [9] S. O. Bleed, H. M. A. Hasan, Estimating and planning step stress accelerated life test for generalized Logistic distribution under type-I censoring, Int. J. Appl. Math. Stat. Sci., 2 (2013), 1–16.
    [10] S.Saxena, S. Zarrin, M. Kamal, A. Ul-Islam, Optimum step stress accelerated life testing for Rayleigh distribution, Int. J. Stat. Appl., 2 (2012), 120–125. https://doi.org/10.5923/j.statistics.20120206.05 doi: 10.5923/j.statistics.20120206.05
    [11] K. Ahmadi, M. Rezaei, F. Yousefzadeh, , Estimation for the generalized half-normal distribution based on progressive type-II censoring, J. Stat. Comput. Sim., 85 (2015), 1128–1150. https://doi.org/10.1080/00949655.2013.867494 doi: 10.1080/00949655.2013.867494
    [12] S. J. Wu, Y. P. Lin, S. T. Chen, Optimal step-stress test under type I progressive group-censoring with random removals, J. Stat. Plan. Infer., 138 (2008), 817–826. https://doi.org/10.1016/j.jspi.2007.02.004 doi: 10.1016/j.jspi.2007.02.004
    [13] N. Balakrishnan, L. Zhang, Q. Xie, Inference for a simple step-stress model with Type-I censoring and lognormally distributed lifetimes, Commun. Stat.–Theor. Methods, 38 (2009), 1690–1709. https://doi.org/10.1080/03610920902866966 doi: 10.1080/03610920902866966
    [14] C. T. Lin, N. Balakrishnan, S. J. Wu, planning life tests based on progressively type-I grouped censored data from the Weibull distribution, Commun. Stat.–Simul. Comput., 40 (2011), 574–595. https://doi.org/10.1080/03610918.2010.549278 doi: 10.1080/03610918.2010.549278
    [15] K. U. S. Coşkun, Y. Akdogan, S. J. Wu, Planning life tests for burr XII distributed products under progressive group-censoring with cost considerations, Gazi Univer. J. Sci., 25 (2012), 425–434.
    [16] C. Kuş, Y. Akdoğan, S. J. Wu, Optimal progressive group censoring scheme under cost considerations for Pareto distribution, J. Appl. Stat., 40 (2013), 2437–2450. https://doi.org/10.1080/02664763.2013.818107 doi: 10.1080/02664763.2013.818107
    [17] A. S. Hassan, S. M. Assar, A. Shelbaia, Multiple-step stress accelerated life for Weibull Poisson distribution with type I censoring, Int. J. Basic Appl. Sci., 3 (2014), 180. https://doi.org/10.14419/ijbas.v3i3.2533 doi: 10.14419/ijbas.v3i3.2533
    [18] A. A. Ismail, Corrigendum to "Estimating the parameters of Weibull distribution and the acceleration factor from hybrid partially accelerated life test"[Appl. Math. Modell. 36 (2012) 2920–2925], Appl. Math. Model., 39 (2015), 2743. https://doi.org/10.1016/j.apm.2015.07.008 doi: 10.1016/j.apm.2015.07.008
    [19] S. Budhiraja, B. Pradhan, Computing optimum design parameters of a progressive type I interval censored life test from a cost model, Appl. Stoch. Models Bus. Ind., 35 (2017), 494–506. https://doi.org/10.1002/asmb.2251 doi: 10.1002/asmb.2251
    [20] S. Roy, B. Pradhan, Bayesian C-optimal life testing plans under progressive type-I interval censoring scheme, Appl. Math. Model., 70 (2019), 299–314. https://doi.org/10.1016/j.apm.2019.01.023 doi: 10.1016/j.apm.2019.01.023
    [21] J. Wang, Data analysis of step-stress accelerated life test with random group effects under Weibull distribution, Math. Probl. Eng., 2020 (2020), 1–11. https://doi.org/10.1155/2020/4898123 doi: 10.1155/2020/4898123
    [22] R. M. EL-Sagheer, M. A. Khder, Estimation in K-stage step-stress partially accelerated life tests for generalized Pareto distribution with progressive type-I censoring, Appl. Math. Inf. Sci., 15 (2021), 299–305. https://doi.org/10.18576/amis/150307 doi: 10.18576/amis/150307
    [23] M. Kamal, S. A. Siddiqui, A. Rahman, H. Alsuhabi, I. Alkhairy, T. S. Barry, Parameter estimation in step stress partially accelerated life testing under different types of censored data, Comput. Intel. Neurosc., 2022 (2022). https://doi.org/10.1155/2022/3491732
    [24] L. Zhuang, A. Xu, B. Wang, Y. Xue, S. Zhange, Data analysis of progressive-stress accelerated life tests with group effects, Qual. Technol. Quant. Manage., 20 (2023). https://doi.org/10.1080/16843703.2022.2147690
    [25] R. Alotaibi, A. A. Mutairi, E. M. Almetwally, C. Park, H. Rezk, Optimal design for a bivariate step-stress accelerated life test with alpha power exponential distribution based on type-I progressive censored samples, Symmetry, 14 (2022). https://doi.org/10.3390/sym14040830
    [26] X. Bai, Y. Shi, H. K. T. Ng, Statistical inference of Type-I progressively censored step-stress accelerated life test with dependent competing risks, Commun. Stat.-Theor. Methods, 10 (2022), 3077–3103. https://doi.org/10.1080/03610926.2020.1788081 doi: 10.1080/03610926.2020.1788081
    [27] A. M. Almarashi, Inferences of generalized inverted exponential distribution based on partially constant-stress accelerated life testing under progressive Type-II censoring, Alexandria Eng. J., 63 (2023), 223–232. https://doi.org/10.1016/j.aej.2022.07.063 doi: 10.1016/j.aej.2022.07.063
    [28] M. Nassar, A. Elshahhat, statistical analysis of inverse Weibull constant-stress partially accelerated life tests with adaptive progressively type I censored data, Mathematics, 11 (2023). https://doi.org/10.3390/math11020370
    [29] I. Alam, A. Haq, L. K. Sharma, S. Sharma, Ritika, Warranty costs analysis under accelerated life test for power Ishita distribution with type-I censored data, Int. J. Qual. Reliab. Manage., 40 (2023), 1983–1998. https://doi.org/10.1108/IJQRM-08-2022-0251 doi: 10.1108/IJQRM-08-2022-0251
    [30] A. A. Bhat, S. P. Ahmad, A new generalization of Rayleigh distribution: Properties and applications, Pak. J. Statist., 36 (2020), 225–250.
    [31] K. Ateeq, T. B. Qasim, A. R. Alvi, An extension of Rayleigh distribution and applications, Cogent Math. Stat., 6 (2019), 1622191. https://doi.org/10.1080/25742558.2019.1622191 doi: 10.1080/25742558.2019.1622191
    [32] D. Kundu, M. Z. Raqab, Generalized Rayleigh distribution: different methods of estimations, Comput. Stat. Data Anal., 49 (2005), 187–200. https://doi.org/10.1016/j.csda.2004.05.008 doi: 10.1016/j.csda.2004.05.008
    [33] A. H. Tolba, T. A. Abushal, D. A. Ramadan, Statistical inference with joint progressive censoring for two populations using power Rayleigh lifetime distribution, Sci. Rep., 13 (2023), 3832.
    [34] H. S. Migdadi, N. M. Al-Olaimat, M. Mohiuddin, O. Meqdadi, , Statistical inference for the Power Rayleigh distribution based on adaptive progressive Type-II censored data, AIMS Math., 8 (2023), 22553–22576. http://dx.doi.org/10.3934/math.20231149 doi: 10.3934/math.20231149
    [35] W. Nelson, Accelerated life testing-step-stress models and data analyses, IEEE Trans. Reliab., 29 (1980), 103–108. https://doi.org/10.1109/tr.1980.5220742 doi: 10.1109/tr.1980.5220742
    [36] S. J. Wu, Y. P. Lin, Y. J. Chen, Planning step-stress life test with progressively type I group-censored exponential data, Stat. Neerl., 60 (2006), 46–56. https://doi.org/10.1111/j.1467-9574.2006.00309.x doi: 10.1111/j.1467-9574.2006.00309.x
    [37] H. M. Aly, Planning step stress accelerated life test for log logistic distribution under progressive type I group censoring, in Proceedings of the 20th Annual Conference on Statistics and Modeling in Human and Social Science, Cairo University, (2008), 107–125.
    [38] J. F. Lawless, Statistical models and methods for lifetime data, John Wiley & Sons, 2003.
    [39] P. Hall, Theoretical comparison of bootstrap confidence intervals, Ann. Stat., 16 (1988), 927–935. https://doi.org/10.1214/aos/1176350933 doi: 10.1214/aos/1176350933
    [40] B. Efron, R. J. Tibshirani, An introduction to the bootstrap, CRC press, 1994.
    [41] E. Gouno, A. Sen, N. Balakrishnan, Optimal step-stress test under progressive Type-I censoring, IEEE Trans. Reliab., 53 (2004), 388–393. https://doi.org/10.1109/tr.2004.833320 doi: 10.1109/tr.2004.833320
    [42] R. Al-Aqtash, C. Lee, F. Famoye, Gumbel-Weibull distribution: Properties and applications, J. Modern Appl. Stat. Methods, 13 (2014), 201–255. http://dx.doi.org/10.22237/jmasm/1414815000 doi: 10.22237/jmasm/1414815000
    [43] H. Alkasasbeh, F. M. Al Faqih, A. S. Shoul, Computational Simulation of Magneto Convection Flow of Williamson Hybrid Nanofluid with Thermal Radiation Effect, CFD Letters, 15 (2023), 92–105. https://doi.org/10.37934/cfdl.15.4.92105 doi: 10.37934/cfdl.15.4.92105
  • This article has been cited by:

    1. M. Al-Hawari, Mahmoud Alazzam, On Numerical Radius Inequalities for Hilbert Space Operators, 2024, 19, 2224-2856, 267, 10.37394/23203.2024.19.29
  • Reader Comments
  • © 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1330) PDF downloads(74) Cited by(1)

Figures and Tables

Figures(3)  /  Tables(7)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog