Aircraft configuration management using constraint satisfaction techniques

Aircraft fault central warning systems (CWS) cannot match the Flight Engineer´s ability to monitor trends, predict failures and diagnose root faults. This is principally due to the simple monitoring of parameters which does not allow trend analysis or the extraction of consequential failures from root causes. As the pilots are primarily concerned with flight management they do not have the time to monitor system parameters and so warnings are generally unexpected. They then have to understand the failures, identify the root faults and devise strategies whilst flying the aircraft in a degraded state. The advanced CWS is split into the structural model network (SMN) and network executive (NE). The SMN monitors system parameters, predicts trends and diagnoses faults by using model based reasoning. Fuzzy models of systems and subsystems are being developed and linked in a network representative of the interactions between aircraft systems and subsystems. Analysis of discrepancies between models and real parameters is used to predict trends, diagnose faults and produce warnings. The subsequent warnings produced are used by the network executive to monitor the aircraft state and produce the appropriate remedial procedures. Thus the task of the NE is to manage the aircraft configuration so that safety is optimised. Hill (1993) introduced a solution to this problem, this paper expands and enhances this approach by using constraint satisfaction techniques.