《可靠性增长试验及其评价方法.doc》由会员分享,可在线阅读,更多相关《可靠性增长试验及其评价方法.doc(10页珍藏版)》请在金锄头文库上搜索。
1、可靠性增长试验及其评价方法3可靠性增长试验试验数据的处理由于产品在可靠性增长试验过程中,受试产品的技术状态处于不断地变化之中,其故障率也在不断地变动,因此可靠性增长过程中的产品,不能采用指数分布的假设对其故障数据进行处理。除非在产品每次状态发生变化后,单独进行一次该状态下的可靠性试验,来评价变化后的可靠性水平,但这样做,费时费钱,工程上很难支持这种做法。a增长趋势有效性检验为了对产品可靠性增长试验中发生的故障数据进行有效地处理与分析,以便对可靠性增长过程实施科学地监控。根据统计数据处理的一般要求,必须首先对所获得的故障数据,进行增长趋势的有效性检验。其目的是确认产品经过不断地设计和制造工艺等方
2、面的改进后,其可靠性是否已有明显提高(统计意义上)。增长趋势有效性检验的方法,可借用国际标准IEC 60605-6或国标GB 5080.6推荐的关于恒定失效率假设的有效性检验方法,其具体方法如下:设受试产品总数为个,为发生第次故障时所有参试产品的总累积有效试验时间(),为试验中止时所有受试产品的总累积有效试验时间。当第个故障发生时试验立即中止,有;否则在其它时间中止,有。按下式求出检验用统计量值: (1)其中 当时 当时选取检验显著性水平,这里为置信度,常取90%和95%,如出现 则可认为该产品具有显著的可靠性增长趋势,否则不能确认其可靠性有明显的增长趋势。值可由表1查得。在MIL-HDBK-
3、781和GJB 1407中,还推荐了另一种用于确认产品可靠性增长趋势的U检验法。即先求出检验用统计量U的值 (2)式中意义同(1)式规定。 表1增长趋势检验表mf(0.8,m)f(0.9,m)f(0.95,m)11.6092.3032.99622.9943.8904.74434.2795.3226.29645.5156.6817.75356.7217.9939.15467.9069.27410.5179.07610.5311.84810.2311.7713.15911.3812.9914.431012.5214.2115.711113.6515.4116.961214.7816.6018.21
4、1315.9017.7819.441417.0118.9620.671518.1320.1321.891619.2321.3023.091720.3322.4624.291821.4323.5925.501922.5424.7726.692023.6425.9127.893034.4837.2239.53选取检验显著特性水平,如出现 则认为产品可靠性有显著的增长趋势;否则不予确认,值可由表2查得。b增长模型的拟合优度检验当确认产品具有明显的可靠性增长趋势后,为了对其增长过程进行定量评价和实施科学的过程管理,目前常采用可靠性增长模型对其故障数据进行拟合。为了确认所选用的增长模型是否合适和有效,统
5、计学上可采用拟合优度检验的方法进行判断。表2增长趋势检验表1234560.2%1.732.342.642.782.863.091%1.722.212.382.452.472.582%1.702.102.222.252.272.335%1.651.901.941.941.941.9610%1.561.681.661.651.651.6520%1.391.351.311.311.301.28AMSAA模型是目前常用的一种可靠性增长模型,可采用下述Cramer-Von Mises检验方法对其拟合优度进行检查。其检验用统计量为:(3)其中,式中的意义同(1)式。选定检验的显著性水平(通常取0.1)并根
6、据由表3查得临界值。若,则拒绝AMSAAA模型,反之则认为AMSAA模型可作为本次可靠性增长的评价模型。表3Cramer-Von Mises检验临界值表显著性水平0.200.150.100.050.0120.1380.1490.1620.1750.18630.1210.1350.1540.1840.23140.1210.1360.1550.1910.27950.1210.1370.1600.1990.29560.1230.1390.1620.2040.30770.1240.1400.1650.2090.31680.1240.1410.1650.2100.31990.1250.1420.1670
7、.2120.323100.1250.1420.1670.2120.324150.1260.1440.1690.2150.327200.1280.1460.1720.2170.333300.1280.1460.1720.2180.333600.1280.1470.1730.2210.3331000.1290.1470.1730.2210.3334.AMSAA可靠性增长模型AMSAA模型是把产品的可靠性增长过程作为统计学上的一个随机过程来处理的,并认为产品发生故障的累积过程是一个非齐次Poisson过程(NHPP),即从0到时刻t,产品共发生r次故障的概率为: r=0,1,2, (4)表4 AMS
8、AA模型时间截尾区间估计系数表0.80.90.950.982.1319.325.10019.33.07939.33.06299.353.2224.217.1756.491.1459.700.11616.074.2893.182.2344.460.1976.070.1618.8585.3412.709.2823.614.2404.690.2006.4346.3822.429.3213.137.2763.948.2335.2127.4172.242.3532.827.3073.481.2614.4718.4472.106.3822.608.3343.158.2873.9729.4722.004.4
9、062.444.3582.920.3103.61210.4941.922.4282.318.3792.738.3303.34111.5141.855.4472.215.3982.593.3493.12812.5311.801.4652.130.4152.474.3662.95713.5461.755.4812.060.4312.376.3812.81514.5611.714.4951.999.4462.293.3962.69715.5731.680.5091.948.4592.220.4092.59616.5851.649.5211.902.4722.158.4212.50817.5961.6
10、22.5321.862.4832.104.4332.43218.6061.598.5431.826.4942.055.4442.36419.6161.575.5521.793.5042.011.4542.30420.6241.556.5611.765.5131.972.4642.25121.6321.538.5701.738.5221.937.4722.20322.6401.522.5781.714.5311.905.4812.15823.6471.506.5861.692.5391.876.4892.11924.6541.492.5931.672.5461.849.4962.08225.66
11、01.478.6001.653.5531.824.5042.04926.6651.466.6071.636.5601.801.5112.01727.6711.455.6121.620.5661.780.5171.98928.6771.444.6181.605.5731.760.5241.96229.6821.435.6241.590.5781.741.5301.93730.6871.426.6291.577.5841.724.5361.91435.7081.386.6531.520.6091.650.5621.81740.7261.355.6731.477.6301.599.5841.74345.7411.331.6891.443.6471.550.6031.68550.7541.310.7041.414.6621.513.6191.63860.7741.278.7271.370.6881.456.6461.56470.7901.254.7451.337.7081.414.6681.51180.8031.235.7591.311.7251.382.686