F-15 Radar

               The purpose of this effort was to demonstrate improvements in flight-line diagnostic capability.  Several activities were accomplished under this effort.  First, a model was created from design data and used to correlate both radar built-in test and calibration test results to derive a more accurate diagnostics.  The Diagnostician was used to provide this correlation.  The results were significantly more accurate flight-line diagnostics.  Second, maintenance history data was used to further refine diagnostics in the cases where a diagnostic session resulted in an ambiguity group.  Third, the Hughes Technical Services AIMSS IETM package was integrated in order to provide context sensitive on-line technical information related to repair procedures. The fault diagnosis results in accessing on-line technical information at the appropriate repair procedures within the IETM package. Fourth, Hughes VSAT (Very Small Aperture Terminal) technology was implemented to move up-to-date maintenance information resources to the flight-line maintenance aid when needed.

               A demonstration was prepared under the LabVIEW graphical programming environment.

 Demonstrated Diagnostic Capability

               The demonstrated improved diagnostic capability was implemented using Giordano Automation's Diagnostic Profiler and Diagnostician.  The Diagnostic Profiler was used to create a diagnostic knowledge base of the radar from CAD outputs (Edif files). A hierarchical design approach was used wherein the radar was represented at a high level as a parent model, and the radar subsystem/units were created as child models, hierarchically linked to the parent model.  Major focus was on the Radar Exciter/Receiver unit. 

               Radar built-in test and calibration tests were "mapped" across the Diagnostic Knowledge Base (DKB) using the Diagnostic Profiler.  Relative failure rate values were assigned to radar subsystem and unit components.  In some cases, estimates of actual failure rates were used based upon data availability. A run-time DKB was generated and hosted on the maintenance aid.  A fault simulator is used to simulate single and multiple fault events for demonstration purposes. 

               Resulting diagnostic capability using the Diagnostician to correlate BIT and calibration test results to the diagnostic knowledge base is 68.5% to a single replaceable item, 93.6% to two replaceable items, and 96.6% to three replaceable items.  The largest ambiguity group possible is four (4) replaceable units. 

               When a diagnostic session results in an ambiguity group, the Diagnostician displays the items in the ambiguity group in a list, with the most likely faulty item first in the list, the second most likely item second in the list and so on.  The "most likely" item is based upon a combination of the evidence (test results correlated to the model) and the failure rate.  This is referred to as "Evidence Weight" and can be skewed at run-time to provide more weight to either the empirical test results or to the failure rate weightings.  Initial failure rate data is input during DKB development using the Diagnostic Profiler.  Normally, failure rate data during development is based upon engineering judgment.  A goal in this project was to provide the capability to incorporate actual field failure rate data into the diagnostic reasoning process.

               The approach used is to glean actual field failure rates from maintenance history data bases, such as the Air Force TICARRS maintenance data base.  The TICARRS data base was analyzed with respect to its content to determine if the appropriate data is included to determine actual field failure rate data.  It was determined that TICARRS contains a great deal of relevant data, but does not contain all of the data necessary to accomplish the goal of determining actual failure rates.  This is primarily due to the lack of confirmation of resulting maintenance actions.  What is required is the ability to confirm that a maintenance action taken in the field actually resulted in fixing the problem, and in the cases where more than one item was removed from the aircraft, which of those items was actually faulty and which were fault-free, removed as a result of an ambiguity group situation. This link between O-level results and intermediate or depot level results is essential for resolving Cannot Duplicate and Re-Test OK problems that have plagued military maintenance for years. An assumption was made that this essential data will become available in the future.  A data base file was created, using TICARRS as a basis, which includes this essential data element. The data base file is in FoxPro format, but could be created in any relational data base format.

               The Diagnostician was augmented with the processing capability to extract actual field failure rate experience from the maintenance history data base and assign these actual relative failure rates to the items in an ambiguity group.  When the diagnostic session results in an ambiguity group, the items in the ambiguity group are displayed, again in order of most likely items.  Additionally, each item is displayed with a probability assigned, expressed as a percentage. The percentage represents the probability that the item in the ambiguity group is the item which is faulty.