Reliability sequential compliance method for a partially observable gear system subject to vibration monitoring

2.1. Gearbox life distribution

We obtained 18 samples of gearbox life history data for mechanical fault diagnosis from the Applied Research Laboratory at Pennsylvania State University. All the gears worked under the same or similar conditions. A goodness-of-fit test was performed for several probability models to determine the best fitness of the gearbox life data. Four most possible kinds of probability distribution functions were tested, the goodness-of-fit values of which are shown in Fig. 2. It can be seen from Fig. 2 that the life distribution of gears does not obey exponential distribution, therefore the SPRT for the exponential distribution in the standards is not suitable for the gear reliability compliance test. It is shown from the test that the Weibull distribution is the best one for the life of gears with a high $p$-value of 0.450. Its shape and scale parameters are shown in Fig. 3. Next, we derive the sequential compliance testing scheme for Weibull distribution according to the theory of SPRT.

Fig. 2Goodness-of-fit test for gear life data

Fig. 3Gear life distribution ft;η,m=m/ηmηm-1e-t/ηm, t≥0, η=158.6, m=1.846

2.2. Operating characteristic curve and sampling risk

The ideal operating characteristic (OC) curve of sampling is shown in Fig. 4 with ${\theta }_0$ as the life expectancy of products. Rules are given that: if the life expectancy of the batch of products $\theta \ge {\theta }_0$, the life of the product is deemed to be qualified, and the acceptance probability $L\left(\theta \right)=\text{1}$; if not, the life of the product is deemed to be unqualified, and the acceptance probability $L\left(\theta \right)=0$.

There are two hypotheses existing in the SPRT: $H_0$: $\theta ={\theta }_0$, $H_1$: $\theta ={\theta }_1$, ${\theta }_0>{\theta }_1$. The sampling test, which determines the whole batch of products through checking the samples, will lead to two types of errors due to the randomness of sampling. A Type I error determines the certified products as uncertified, bringing loss to the producer and rejects $H_0$ when it is true, the probability of which is defined as the producer’s risk $\left(\alpha \right)$. A Type II error determines the uncertified products as certified, bringing loss to the consumer and accepts $H_1$ when it is false, the probability of which is defined as the consumer’s risk $\left(\beta \right)$. Both the producer and the consumer hope to take lower risk, therefore, the upper test mean time between failure (MTBF) $\left({\theta }_0\right)$ and lower test MTBF $\left({\theta }_1\right)$ are proposed. The acceptance probability $L\left(\theta \right)$ changes with $\theta$; Fig. 5 displays the actual OC curve of sampling.

Fig. 4The ideal OC curve of sampling

Fig. 5The actual OC curve of sampling

2.3. SPRT for Weibull distribution

The cumulative distribution function (CDF) $F\left(t\right)$ and probability density function (PDF) $f\left(t\right)$ of the two-parameter Weibull distribution are given by:

where $m$ is shape parameter and $\eta$ is scale parameter.

Assume that the random variable $X$ obeys Weibull distribution, the probability distribution function of which is denoted by $f\left(x,\theta \right)$, where $\theta$ refers to the life parameter. Both producer and consumer take products’ MTBF to verify whether the products meet the reliability requirement or not:

With pre-assigned ${\theta }_0$ and ${\theta }_1$, then for the samples $(x_1,x_2,\dots ,x_n)$, of which the probability density function of population samples are $f\left(x,\theta \right)$, and the joint probability density function of the random variables $(x_1,x_2,\dots ,x_n)$ is given by: $P_{\theta }=f\left(x_1,x_2,\dots ,x_n;\theta \right).$

If $\theta ={\theta }_0$, then the joint probability density function is $P_{{\theta }_0}=f\left(x_1,x_2,\dots ,x_n;{\theta }_0\right)=1-\alpha .$

If $\theta ={\theta }_1$, then the joint probability density function is $P_{{\theta }_1}=f\left(x_1,x_2,\dots ,x_n;{\theta }_1\right)=\beta$, where $\alpha$ is the producer’s risk and $\beta$ is the consumer’s risk.

Then, the probability ratio is $P_{{\theta }_0}/P_{{\theta }_1}$. This is the principle of the sequential probability ratio sampling test plan. According to the sequential sampling test rules, two constants, $A$ and $B$, need to be chosen. Wald [1] put forward the SPRT judge boundaries: $A=(1-\beta )/\alpha$, $B=\beta /(1-\alpha )$.

1) If $P_{{\theta }_0}/P_{{\theta }_1}\le B$, then $\theta ={\theta }_0$, “accept” ($P_{{\theta }_0}$ is lager and “accept” with high probability);

2) If $P_{{\theta }_0}/P_{{\theta }_1}\ge A$, then $\theta ={\theta }_1$, “reject” ($P_{{\theta }_0}$ is larger and “reject” with high probability);

3) If $B<P_{{\theta }_0}/P_{{\theta }_1}<A$, then decisions to “accept” or “reject” cannot be made, so the test should be continued.

Judgment criteria for the sequential validation test under the exponential distribution is given in GJB 899A-2009 and MIL-STD-781D [7, 8]. By transforming the two-parameter Weibull distribution into the exponential distribution, the SPRT scheme for Weibull distribution can be obtained. Substituting $y=t^m$, $\delta ={\eta }^m$ into Eq. (1) yields:

Then $y$ obeys the exponential distribution, thus the Weibull distribution has been converted to the exponential distribution. Correspondingly, if $N$ samples are put into the reliability sequential test and $r$ samples fail until $t$ with $t_{\left(i\right)}$ as the failure time of the $i$th sample, then the joint probability density function of the samples $(t_{\left(1\right)},t_{\left(2\right)},\dots ,t_{\left(r\right)})$ is:

The probability that there are $r$ failed items before cumulative test time $t$ is:

Under the different requirements of MTBF, the probability ratio of $r$ failed items before cumulative test time $t$ is:

where ${\delta }_0={\eta }_0{\left({\theta }_0\right)}^m$, ${\delta }_1={\eta }_1{\left({\theta }_1\right)}^m$.

Different from Hauck and Keats [13], a predetermined reliability $R$ must be given when converting to ${\eta }_0{\left({\theta }_0\right)}^m$ and ${\eta }_1{\left({\theta }_1\right)}^m$ with Eq. (7), while MTBF is the verification index of product, so the value of $R$ will affect the formulation of the test plan.

By substituting ${\theta }_0$ and ${\theta }_1$into the Eq. (3), ${\eta }_0{\left({\theta }_0\right)}^m$ and ${\eta }_1{\left({\theta }_1\right)}^m$ can be obtained:

The judge boundaries of SPRT is given by:

Taking natural logarithms on both sides of inequality Eq. (10), judge conditions of the SPRT plan for Weibull distribution is given by:

Sequential compliance criterion for Weibull distribution is given by:

where:

$T_{r,N\left(t\right)}$ is the cumulative failure time, in the non-replacement case:

In the replacement case:

Taking an example of ${\theta }_0=$135$h$, ${\theta }_1=$ 90$h$, $\alpha =\beta =$ 0.10, $m=$1.846, the sequential schematic diagram for Weibull distribution is shown in Fig. 6. In the sequential test, once the failure occurs, the point $\left(T_{r,N\left(t\right)},r\right)$ should be put into Fig. 6 to make a decision. A point $\left(T_{r,N\left(t\right)},r\right)$ plotted in the “reject” area results in rejecting the hypotheses. A point $\left(T_{r,N\left(t\right)},r\right)$ plotted in the “accept” area results in accepting the hypotheses. Testing continues as long as the point $\left(T_{r,N\left(t\right)},r\right)$ remains in the “test continuing” area.

Fig. 6Schematic of SPRT for Weibull distribution

3.1. Time synchronous averaging (TSA) algorithm

The TSA algorithm is widely applied in fault diagnosis of gears. The TSA signal of vibration monitoring data is given by:

where $K$ is the number of sampling points in a single operating period and $M$ is the rotation period number included in each sampling data file.

In a complete revolution, the TSA signal of a gear of interest meshing signal in the healthy state is given by:

where $n=\mathrm{0,1},\dots ,N$ is the meshing harmonic number; $A_n$ is the amplitude at the $n$th harmonic frequency $f_n$ ($f_n=n\times N_g\times f_g$, where $N_g$ is the number of teeth and $f_g$ is the shaft rotation frequency); $a_n\left(t\right)$ is the amplitude modulation function, ${\beta }_n$ is the initial phase, and ${\theta }_n\left(t\right)$ is the phase modulation function at the $n$th harmonic.

When a localized fault occurs, the impact-induced resonant vibration $z\left(t\right)$ in one revolution of the monitored gear is [23]:

where $d\left(t\right)$ is the envelope function, $f_s$ is the resonance frequency and ${\beta }_s$ is the corresponding initial phase. Hence, the measured signal $y\left(t\right)$ can be written as the following form:

where ${\tilde{a}}_n\left(t\right)$ is the modified amplitude modulation functions at the $n$th harmonic, and ${\tilde{\theta }}_n\left(t\right)$ is the modified phase modulation functions at the $n$th harmonic.

Take for example the #12 gearbox, in which the first failure occurs. File 5, file 18 and file 31 of the data file A02 and A03 at 300 % output torque were selected for TSA signal analysis, which are shown in Fig. 13 and Fig. 14.

Fig. 13A02 data analysis: a) original data of file 5, b) TSA signal of file 5; c) original data of file 18, d) TSA signal of file 18; e) original data of file 31, f) TSA signal of file 31

3.2. Vector autoregression (VAR) model

In this paper, the healthy portion of the TSA signals was used to fit a VAR model. With the established model and estimated model parameters, the residuals of the model can be further calculated. Two-dimensional data can be expressed as $\left\{Z_1^i,Z_2^i,\dots ,Z_{t_i}^i\right\}$, $i=$ 1, 2. Suppose that the healthy portion of the data obeys a stationary VAR process:

where ${\epsilon }_n$ are independent identically distributed (i.i.d) and follow $N_2\left(0,\mathrm{\Sigma }\right)$; ${\mathrm{\Phi }}_r\in {\mathbb{R}}^{2\times 2}$ is the autocorrelation matrices; ${\delta }_0\in {\mathbb{R}}^2$ is the mean value; $\mathrm{\Sigma }\in {\mathbb{R}}^{2\times 2}$ is the covariance; $p\in \mathbb{N}$ is the model order. Let $\delta ={\delta }_0-\sum _{r=1}^p{\mathrm{\Phi }}_r{\delta }_0$, Eq. (19) can be transformed as follows:

The Akaike Information Criterion (AIC) criterion is used in this paper to determine the model order $p$, more details can be found in Reinsel [31].

Fig. 14A03 data analysis: a) original data of file 5, b) TSA signal of file 5; c) original data of file 18, d) TSA signal of file 18; e) original data of file 31, f) TSA signal of file 31

3.3. Calculation of residuals

The estimated VAR model parameter $\widehat{\gamma }=\left(\widehat{p},\widehat{\delta },{\widehat{\mathrm{\Phi }}}_1,{\widehat{\mathrm{\Phi }}}_2,\dots ,{\widehat{\mathrm{\Phi }}}_p\right)$ can be utilized to calculate the residual $Y_n$ of the TSA signals. The residual $Y_n$ can be computed as follows:

Taking #12 gearbox as an example, in which failure first occurred. Fig. 15 exhibits the residuals from A02 and A03 data files at 300 % output torque. We selected the root mean square (RMS) of characteristic parameters to process the residuals, which was commonly used in vibration signal processing, to reflect the states of gears, as shown in Fig. 16.

From Fig. 16, the residual RMS values in files 1 to 16 from A02 and A03 sensors are more stable, and we assume this part of the data is the healthy portion. Those in files 17 to 31 increased significantly, therefore we assume this part of the data is the unhealthy portion. Normality and independence tests were carried out for these two parts of data, of which the results are shown in Table 1.

Fig. 15Residuals for data files 1-31

Fig. 16RMS Residuals for data files 1-31

From Table 1, both the RMS values of the residuals for A02 and A03 passed the normality and independence tests, indicating that RMS residuals are independent and obey the multivariate normal distribution, which is consistent with the observation values of the HMM in Section 4 follow the multivariate normal distribution.

4.1. Hidden Markov model (HMM)

Kim et al. [32] showed that a 3-state Markov was sufficient to describe the degradation process. Assume that the gear system can be described by a 3-state HMM (state space is $S=\left\{\mathrm{0,1},2\right\}$): state 0 (healthy or good state), state 1 (unhealthy or warning state) and state 2 (failure state). State 0 and state 1 are unobservable. Only state 2 is observable. The system state transition is illustrated in Fig. 17, in which the transition probability matrix is $P=\left[P_{ij}\right]\left(i,j\in S\right)$, where $P_{ij}$ denotes the probability that the state transitions from state $i$ to state $j$.

Fig. 17Schematic for system state transition of 3-state HMM

Suppose that the system starts in the healthy state, the instantaneous transition rate matrix is given by:

where ${\lambda }_{01}$, ${\lambda }_{02}$, ${\lambda }_{12}\in \left(0,+\infty \right)$ are unknown parameters.

Let $\xi =\mathrm{i}\mathrm{n}\mathrm{f}\left\{t\ge 0:X_t=2\right\}$ represent the system failure time. Observation $Y=\left(Y_n:n\in \mathbb{N}\right)$ is independent given the system state. $y_{\mathrm{\Delta }},y_{2\mathrm{\Delta }},\dots ,y_{\mathrm{k}\mathrm{\Delta }}\in {\mathbb{R}}^2$ represents the bivariate observations. Thus, $y_{k\mathrm{\Delta }}$ under the condition of state $x$ follows a bivariate normal distribution of $N_2\left({\mu }_x,{\mathrm{\Sigma }}_x\right)$,$x=$ 0, 1, and the probability density function is given by:

where ${\mu }_0$, ${\mu }_1\in {\mathbb{R}}^2$ and ${\mathrm{\Sigma }}_0$, ${\mathrm{\Sigma }}_1\in {\mathbb{R}}^{2\times 2}$ are unknown observation parameters.

As the test proceeds, three gearboxes failed in total while the #10 gearbox still worked without failure. The failure times which they shut down for gearboxes #12, #13 and #5 are 111.5 h, 114.5 h and 131 h respectively. With this failure time data, decisions to accept or reject still cannot be made (see Section 5 for analysis results). In fact, three groups of failure data and one group of suspension data provides sufficient information to built a HMM for residual life prediction. Let $F_1$,…,$F_N$ denote the failure data sets, where $N=\text{3}$ (for #12, #13 and #5 gearboxes). Failure data $F_i$ is expressed by $Y_i=\left(y_1^i,y_2^i,\dots ,y_{T_i}^i\right)$ and the respective failure time is ${\xi }_i=t_i$, where $T_i\mathrm{\Delta }<t_i\le \left(T_i+1\right)\mathrm{\Delta }$. Let $S_M$ denote the collected suspension data sets, where $M=1$ (for #10 gearbox). Similarly, suspension data $S_j$ is expressed by $Y_j=\left(y_1^j,y_2^j,\dots ,y_{T_j}^j\right)$ and the failure time ${\xi }_i>T_i\mathrm{\Delta }$. $O=\left\{F_1,\dots ,F_N,S_1,\dots ,S_M\right\}$ represent the observation data and $L=\left(\mathrm{\Lambda },\mathrm{\Psi }|O\right)$ is the corresponding likelihood function, where $\mathrm{\Lambda }=\left({\lambda }_{01},{\lambda }_{02},{\lambda }_{12}\right)$ and $\mathrm{\Psi }=\left({\mu }_0,{\mu }_1,{\mathrm{\Sigma }}_0,{\mathrm{\Sigma }}_1\right)$ are the unknown parameters to be estimated.

The EM algorithm is used to estimate the unknown parameters. The EM algorithm works as follows:

E-step: Compute the pseudo likelihood function:

where $\widehat{\mathrm{\Lambda }}$ and $\widehat{\mathrm{\Psi }}$ are the initial values; $\stackrel{-}{O}$ represents the complete data set, in which each failure history $F_i$ and suspension history $S_j$ of the observable data set $O$ have been augmented with the unobservable sample path information of the state process.

M-step: Choose ${\mathrm{\Lambda }}^{*}$, ${\mathrm{\Psi }}^{*}$such that:

Updated parameters ${\mathrm{\Lambda }}^{*}$, ${\mathrm{\Psi }}^{*}$ in each step are then used as new initial values to the E-step, leading the iteration in E-step and M-step until the Euclidean norm satisfies $\left|\left({\mathrm{\Lambda }}^{*},{\mathrm{\Psi }}^{*}\right)-\left(\widehat{\mathrm{\Lambda }},\widehat{\mathrm{\Psi }}\right)\right|<\epsilon$. The detailed explicit formulae can be found in Kim and Makis [32]. The expectation is updated in each step, for the maximization of which there exists a unique stationary point. The updated explicit formulae for ${\mathrm{\Lambda }}^{*}=\left({\lambda }_{01}^{*},{\lambda }_{02}^{*},{\lambda }_{12}^{*}\right)$ in each step are:

The updated explicit formulae for ${\mathrm{\Psi }}^{*}=\left({\mu }_0^{*},{\mu }_1^{*},{\mathrm{\Sigma }}_0^{*},{\mathrm{\Sigma }}_1^{*}\right)$ in each step are:

Initial values are assigned to $\widehat{\mathrm{\Lambda }}$ and $\widehat{\mathrm{\Psi }}$ and the EM algorithm is used to solve the parameters to be estimated, the convergence criterion of which is $\left|\left({\mathrm{\Lambda }}^{*},{\mathrm{\Psi }}^{*}\right)-\left(\widehat{\mathrm{\Lambda }},\widehat{\mathrm{\Psi }}\right)\right|<1e-6$. The estimated results are shown in Table 2.

4.2. Remaining useful life prediction

In partially observable Markov decision process (POMDP) theory, ${\prod }_k$ is defined as the posterior probability that the system is in the warning state at time $k\mathrm{\Delta }$ given the observations $y_{\mathrm{\Delta }},y_{2\mathrm{\Delta }},\dots ,y_{k\mathrm{\Delta }}\in {\mathbb{R}}^2$ up to time $k\mathrm{\Delta }$ [33]:

According to Bayes’ theorem, at each sampling epoch, the posterior probability ${\prod }_k$ can be updated by:

where the initial value ${\prod }_0=0$. The transition probability matrix $P_{ij}\left(t\right)=P\left(X_t=j|X_0=i\right)$, in Eq. (34) can be obtained by solving Kolmogorov backward differential equations, which is given by:

Details can be found in Appendix A.

So, the conditional reliability function of the RUL at sampling point $\mathrm{k}\mathrm{\Delta }$ is:

Thus, the PDF of the RUL is:

The MRL is given by:

By the time $t=$ 131 h, 70 data files were collected for #10 gearbox. The original data in A02 and A03 are shown in Fig. 18. In accordance with the data processing method in Section 3, the residuals of RMS values are given in Fig. 19.

Fig. 18Original data of test run #10

Fig. 19RMS residuals for test run #10

With the estimated model parameters and the observation residuals of sampling files of #10 gearbox, the conditional reliability functions can be updated using the Eq. (36) and the observation values obtained in each sampling. The conditional reliability functions corresponding to files 65 to 70 are shown in Fig. 20. The probability density functions from files 65 to 70 are shown in Fig. 21, where “*” represents the actual residual life. The corresponding mean residual lives are shown in Fig. 22. The predicted RUL is 18.28 hours at file 60. It changes a little up to file 65, meaning that the gearbox operates under a relatively good state. The MRL decreases to 14.13 hours after file 68, which indicates that the gearbox runs in the warning state. It can be found that the actual RUL from file 68 is 14.45 hours, and from file 70 is 13.45 hours, which are very close to the predicted RUL values. Thus, the proposed life prediction approach is applicable to predict the RUL for the gearboxes.

Fig. 20Conditional reliability function

Fig. 21Probability density function

Fig. 22Remaining useful life prediction for #10 gearbox

4.3. Comparison with other prediction methods

In this subsection, the RUL prediction based on HMM is compared with other commonly used prediction methods: (i) stochastic filtering model (SFM) and (ii) Wiener process [26].

In SFM, the unobservable states and the variables related to observable states are used to establish a state space model to describe the degradation process of the system. Suppose that the condition monitoring data $y_i$ in the good and warning states follow the two-parameter Weibull distribution, which is denoted by $f\left(y_i;\alpha ,\beta \right)$, where $\alpha$ is the scale parameter and $\beta$ is the shape parameter. Therefore, the relationship between $y_i$ and RUL has the following form:

Further, the stochastic relationship between the current condition monitoring data ${Y_j}^{\text{'}}$ obtained when the gear system starts in the warning state and RUL $r_j^{\text{'}}$ is given by:

where $t_0^{\text{'}}$ is the first time when failure occurs.

The unknown model parameters in SFM can be estimated using maximum likelihood estimation (MLE). Thus, the RUL can be further obtained.

Next, the Wiener process is employed to model the deterioration of the gearboxes. Let $X\left(t\right)$ be the degradation measure at time $t$. In general, the Wiener process can be represented as the following form:

where $B\left(t\right)$ is the standard Brownian motion, $B\left(t\right)~N\left(0,t\right)$, $\lambda >0$ is the drift parameter, $\sigma >0$ is the diffusion coefficient.

Denoted by $w$ is the failure threshold, thus the life of the gear system $T$ is defined by the first time that the degradation measure is above the failure threshold, which is given by:

From Eq. (40), the PDF of the RUL can be computed as follows:

Therefore, the expectation of RUL is given by:

The data from files 60 to 70 are selected for MRL prediction. At each sampling epoch, the life prediction is performed when the new observations are available, and then the predicted values using SFM and the Wiener process are compared with that using HMM. Meanwhile, the relative errors (RE) between actual remaining lives (ARL) and the prediction results using different models are calculated, which is given by:

The prediction results are shown in Table 3. Table 3 evidences that with the increased number of collected data, the RE of the prediction results using HMM are smaller than those using SFM and Wiener process, which are closer to the ARL. The average RE using HMM is 3.90 %, while using SFM is 13.60 % and the Wiener process is 7.47 %. Thus, the life prediction method proposed in this paper yields better results than other methods.

5.1. SPRT for Weibull distribution

When the gearbox shut down due to failure, the failure time was substituted into the sequential decision diagram and a decision was made. If the failure time point plotted in the “accept” or “reject” area, stop the test and make the decision to “accept” or “reject” the hypotheses; if the point plotted in the “test continuing” area, then no decision will be made and the test needs to be continued until the next failure for further analysis. Fig. 23 exhibits the SPRT decision diagram for Weibull distribution using MRL predictions based on HMM and true life value.

Fig. 23The test judgment figure using Weibull SPRT

As illustrated in Fig. 23, the life data of the first three gearboxes with failures were put into the SPRT decision diagram for Weibull distribution, finding that three points all plotted in the “test continuing” area, while the life data point of the fourth gearbox plotted in the “accept” area, leading to the “accept” decision. The residual life predicted from the 70th file of the #10 gearbox is 13.96 h, which is very similar to the ARL of 13.45 h. Using the actual failure time of the #10 gearbox, the decision to “accept” is obtained. Thus, the proposed model can be applied to the prediction of the residual life of the gearbox degeneration system.

5.2. Comparison with other compliance methods

We will now compare the proposed SPRT with other reliability sequential compliance methods. It is known that GJB 899A-2009 [7] and MIL-STD-781D [8] provide the standard SPRT for exponential life distribution. Thus, we first consider the exponential SPRT. The four true failure times of the gearboxes and three prediction values of the fourth gearbox were substituted into the exponential SPRT, the results are presented in Fig. 24. As illustrated in Fig. 24, the life data of the first three gearboxes with failure was put into the SPRT decision diagram for exponential distribution. It was found that the first three points all plotted in the “test continuing” area. While the true value and the prediction values that obtained using HMM, SFM and Wiener process all plotted in the “test continuing” area. Neither a decision to “accept” nor “reject” the hypotheses can be made for all four gearboxes using the SPRT for exponential distribution.

We then consider the SPRT decision diagram for Weibull distribution using true values and prediction values. Fig. 25 shows the first three points all plotted in the “test continuing” area, which is similar to the results obtained using exponential SPRT. The fourth life prediction values obtained using SFM and Wiener process all plotted in the “test continuing” area. However, both the prediction value using HMM and the true value plotted in the “accept” area, indicating that the decision to “accept” the hypotheses can be made.

It can be seen from Fig. 24 and Fig. 25 that adopting different life prediction methods and SPRT methods will lead to quite different decisions. In the SPRT for exponential distribution, using the life prediction values and the true values, one can hardly determine whether to accept or reject the hypotheses before all the samples fail. To make further analysis, new samples must be provided for testing. Furthermore, there is no reason to use the SPRT for exponential distribution because the life of the gearboxes in fact obeys Weibull distribution. Better results can be achieved when adopting the SPRT for Weibull distribution using HMM for RUL prediction. We can accept the hypotheses according to the fourth prediction data which is consistent with the results using the ARL value. While based on the life prediction values that used SFM and the Wiener process, neither decision to accept or reject the hypotheses can be made, and the test must continue for further decision making. The comparison shows the efficacy of the prediction method based on HMM and the proposed SPRT for Weibull distribution.

Fig. 24The test judgment figure using exponential SPRT

Fig. 25The test judgment figure using Weibull SPRT