Diagnosis and prescription Everyday diagnostic reasoning is fundamentally abductive, and abduction has even shown promise in medicine. Through detecting which anomalies remain unexplained, and assembling more complex hypotheses out of simpler fragments, multiple simultaneous faults can be diagnosed. Through reasoning from cause to effect, to determine the effects of a proposed prescription on the ailing system, we can derive performance metrics for each prescription. Many prescriptions can be evaluated, and the tradeoffs between them (cost of repair, time for repair, depletion of stocks, etc.) examined so that the best prescription, with respect to costs and prognosis, may be chosen. Thus, our approach provides a framework that couples powerful diagnosis with the ability to identify the best prescription.
Example We studied the effect of the composition of the Readiness Spares Package on a sixty-day wartime deployment of six A-10A Thunderbolts, nine F-15A Eagles, four KC-135R Stratotankers, and two MH-53M Pave Low helicopters. The unusual mix of aircraft was chosen to maximize the use of the data we had and to complicate the effect of the potential for parts cannibalization between aircraft types. We used stochastic models to track the failure and repair of 635 different types of parts among these aircraft. Performance of a parts kit was based on its weight, volume, price, number of parts, number of types of parts, and the number of repair days needed for each type of aircraft given that parts kit. The number of repair days was determined by modeling the wear on each part for each flying hour, the parts inventory currently installed on each aircraft, and the duration of the replacement, repair and cannibalization processes that underlie the maintenance procedures. Additionally, repair days and flying hours missed were calculated for unfavourable circumstances in which every part has an increased rate of failure. We used evolutionary algorithms to find the best parts kits for that particular deployment, showing the tradeoff between the size of the parts kit and its ability to keep the aircraft flying even under harsh conditions.