Reducing No Fault Found using statistical processing and an expert system

Abstract

This paper describes a method for capturing avionics test failure results from Automated Test Equipment (ATE) and statistically processing this data to provide decision support for software engineers in reducing No Fault Found (NFF) cases at various testing levels. NFFs have plagued the avionics test and repair environment for years at enormous cost to readiness and logistics support. The costs in terms of depot repair and user exchange dollars that are wasted annually for unresolved cases are graphically illustrated. A diagnostic data model is presented, which automatically captures, archives and statistically processes test parameters and failure results which are then used to determine if an NFF at the next testing level resulted from a test anomaly. The model includes statistical process methods, which produce historical trend patterns for each part and serial numbered unit tested. An Expert System is used to detect statistical pattern changes and stores that information in a knowledge base. A Decision Support System (DSS) provides advisories for engineers and technicians by combining the statistical test pattern with unit performance changes in the knowledge base. Examples of specific F-16 NFF reduction results are provided.