0 引言
1 产品威布尔分布无故障数据建模
F(t)=1-exp
2 无故障数据威布尔分布模型计算
2.1 故障率先验分布的确立
λ(t)= =
0≤pi≤1-
πl(pl)=
2.2 故障率贝叶斯统计推断
L(pi)=(1-pi
= piπi(pi )dpi=
2.3 威布尔分布模型参数确定
pi=P(T≤ti)=1-exp
E(m,η)=
(t)=exp
3 威布尔分布可靠性试验方案设计
λ=
dm=
E(S(t))≈ +
4 工程案例
表1 无故障数据统计Table 1 Zero-failure data statistics |
| Serial number | ti/h | ni | si |
|---|---|---|---|
| 1 | 95 | 2 | 7 |
| 2 | 105 | 1 | 5 |
| 3 | 120 | 1 | 4 |
| 4 | 135 | 1 | 3 |
| 5 | 142 | 1 | 2 |
| 6 | 153 | 1 | 1 |
表2 各时刻的故障率先验值Table 2 Prior values of failure rates at each moment |
| Parameter | λ1 | λ2 | λ3 | λ4 | λ5 | λ6 |
|---|---|---|---|---|---|---|
| Value | 0.063 3 | 0.069 8 | 0.079 3 | 0.088 8 | 0.093 2 | 0.100 0 |
表3 各时刻故障率的贝叶斯估计Table 3 Bayesian estimation of failure rate at each moment |
| Parameter | ||||||
|---|---|---|---|---|---|---|
| Value | 0.029 0 | 0.032 6 | 0.037 3 | 0.042 2 | 0.045 0 | 0.049 1 |
表4 威布尔分布可靠性鉴定试验方案(鉴别比d=1.5)Table 4 Reliability evaluationtest scheme of Weibull distribution (identification ratio d=1.5) |
| Scheme | Allowable number of failures | Nominal risk/% | Actual risk/% | Test time/h | ||
|---|---|---|---|---|---|---|
| User's | Producer's | User's | Producer's | |||
| 1 | 32 | 10 | 10 | 31.22 | 1.36 | 1 891 |
| 2 | 13 | 20 | 20 | 29.58 | 12.01 | 1 246 |
| 3 | 5 | 30 | 30 | 33.66 | 25.11 | 547 |
表5 指数分布可靠性鉴定试验方案(鉴别比d=1.5)Table 5 Reliability evaluation test scheme of exponential distribution (identification ratio d=1.5) |
| Scheme | Allowable number of failures | Nominal risk/% | Actual risk/% | Test time/h | ||
|---|---|---|---|---|---|---|
| User's | Producer's | User's | Producer's | |||
| 1 | 39 | 10 | 10 | 10.17 | 10 | 3 450 |
| 2 | 17 | 20 | 20 | 19.43 | 20 | 1 535 |
| 3 | 6 | 30 | 30 | 29.95 | 30 | 578 |