BITE is not the answer (but what is the question?)

摘要

Complicated avionics systems increasingly rely on built-in-test equipment (BITE) to perform fault detection and aid in fault isolation. Recent experience with the large quantity of avionics in the passenger cabins of commercial airlines indicates that (a) reliability is generally inadequate, and (b) maintenance regularly results in a no faults found ratio (NFF) of 50%. The expectation that adding BITE will improve reliability and reduce NFF has not been realized for these large cabin systems. Reducing NFF requires (1) verifying that a failure has actually occurred prior to taking action, and then (2) correctly replacing the failed unit on the first maintenance action. This in turn depends on the correct isolation of failure by whatever methods are available. We propose a methodology for performing line maintenance that is less dependent on the scope of built-in-test than is traditionarily thought to be necessary, and, when followed scrupulously, results in low NFF rates. The method flows out of the failure modes and effects analysis (FMEA), by identifying and correlating the effects of failure (including annunciated test results and observables such as loss or degradation of function). By working "backward" from the failure effects using the analysis, the most-probable source of failure can be identified. We analyze an extensive cabin system with audio and video distribution, and show that the NFF rate is relatively insensitive to the actual amount of BITE, and very sensitive to following the recommended procedures for isolating a failure and replacing suspected units. The question is not, "How do I find out what is failed?", which is the focus of BITE, but is, "What must be done to restore full functionality?".