Downloaded from https://www.everyspec.com AMCP 700-1U tests. During the simulation, we may wish there's an almost 50-50 chance that the fail- to change our assignment of prior degree-of- ure probability is 0.03. The analysis was belief. correct; so the assumptions were bad. The designer was too optimistic-a very common Suppose a portable power tool is to be circumstance. Row 5 is a revised "before tested for insulation breakdown under super- test" degree-of-belief. It still has a large peak severe conditions (high temperature, high ut condition 5, but the bad conditions have load, high humidity, high vibration), If the higher degrees-of-belief.
Now with an hy- insulation breaks down in field use, Injury pothesized 0-pass, 2-fail test result, the "after or death could result. We generate row I of test"' dogree-of-bellef in row 7 seems more TabSz 3-11 by considering what we want to reasonable. Let's try the same "before test" know. If the failure probability is high, we degree-of-belief with 2-pass, 0-fail results; as don't really care exactly what it is because shown in rows 8-10. Row 10 gives the it's "back to the drawing board" anyway. "after test" degree-of-belief, and it too seems The lower the failure probability, the closer reasonable. In practice, you should try many we want to know it.
In order to keep the more simulations of test results before coming example simple, we are supposing that only to a conclusion about your "before test" 6 possible failure probabilities exist, as shown degree-of-belief. J in row 1. Row 2 shows the degree-of-bellef You may wish to have your problem pro- of the designer before submitting to this grammed for a computer; then you can accelerated test. simulate much more extensively and easily Sooner or later, someone will advise you Now let's simulate.
Suppose the test re- to use an "ignorance" prior, I.e., a prior that suits are 0-pass, 2-fail, Row 4 shows the assumes "complete ignorance of the situa- "after test" degree-of-belief for each condi- tion", Reject that advice. Proceed by put- tion. No one in his right mind still believes ting down what you think you do believe; TABLE 3.11 PORTABLE POWER TOOL INSULATION TEST Condition #1 #2 #3 #4 #5 #8 (1) probbblllty of failing test 0.9 0.5 0.2 0,1 0.03 0101 (2) "before test" d.of-b 0.1% 0.1% 0.1% 017% 95% 4% (3) likellhood of () ieio ofi 0,81 0,25 0.04 0.01 0.000 0,0001 2.fail, O-palss:
(4) "after test" d-of-b 40% 12% 1.9% 3,9% 42% 0.2% (5) "before test" d-of b 1% 1% 1% 1% 95% 1% (6) likelihood of 2faill, 0Opass 0,81 0.26 0.04 0,01 0.0009 0,0001 (7) "after test" d-of-b 68% 21% 3,2% 0.8% 7% 0.0,11% (8) "before test" d.of.b 1% 1.% % 1% 95% 1% (9) likelihood of 0.01 0.26 0.64 0.81 0.94 0,8 0Ofall, 2-pass 0. . (10) "after test" d-of-b 0,01% 0.3% 0.7% 0.9% 970f 1.1% d-of-b "degree-0,!.beliof, I,., Bayeslon probablilty э)Ij~~~a.... .......... ........... ......... ... .. .. . ..............