Downloaded from https://www.everyspec.com ADS-79D-HDBK CI/HI versus its projected benefit. Any CI/HI that fails to meet these criteria should be eliminated from the development process . e. The ideal case for a CI/HI is that it should exhibit monotonic behavior (increasing or decreasing with increasing fault size) if the value of the CI/HI is to be used to assess fault severity. f. The CI/HI should be insensitive to extraneous factors (those unrelated to the fault origin or operational state of the aircraft) or able to account for those extraneous factors. g.
The CI/HI should be capable of detecting the fault as required by engineering analysis to ensure that the fault is detected at the minimum severity specified. h. Redundancy between CI/HI algorithms is discouraged, but correlation between non- redundant algorithms may be used in specific cases, e.g. symptom or fault cascades. The CI/HI should be computationally efficient. The calculation of CIs/HIs should be able to meet requirements for timeliness and effective action by maintenance and engineering personnel.
For example, computation of CI/HI values should be able to be completed prior to the next flight of the aircraft, in order for maintenance personnel to be able to take the appropriate action to restore system operation to normal. i. CIs/HIs which are derived from proprietary algorithms are authorized as long as: i. Their functional description is provided to, understood by, and accepted by the Government ii. The results of the CI/HI are validated, verified, and documented during the development process. j. HIs that combine multiple CI values can use any of the following methods (not intended to be an exclusive list), subject to validation and verification of effectiveness: i.
Weighted Averages: Using weights that modify the CI values for criticality and severity ii. Bayesian Reasoning iii. Dempster-Schafer Theory: A formalized method for managing uncertainty iv. Fuzzy Logic Inference k. HIs that use CI values to assess system health should have a clear understanding of CI correlation to fault growth. The non linear behavior of many faults and corresponding CI values precludes the ability to base actions on simple ―linear trend analysis.‖ D.4.2 CI and HI Confidence To ensure confidence in fault detection, CIs should be characterized by accuracy, detectability, separability, and identifiability.
A class is representative of a specific failure mode or the base class of normal operation. To meet this confidence requirement, the following guidelines are recommended. D.4.2.1 False positive rate. CI and HI based maintenance actions on the aircraft should have a false alert rate of no more than 10%. A false positive is a warning that results in the unnecessary removal of a component or other unnecessary maintenance actions. 78