Maintenance data mining and visualization for fault trend analysis

This paper describes research efforts currently underway to acquire and analyze test data to determine whether trends and other tendencies may exist that may be indicative of future circuit board failures and potential reduced weapon system readiness. We begin by citing that in today´s test environment using test program sets (TPSs) hosted on automatic test equipment (ATE), no provisions are made for capturing or analyzing Unit Under Test (UUT) data, on a large scale. The distributed resources used to perform UUT testing further complicate the situation,since no methodology currently exists that can demonstrate whether trends or events exist in the data that may be indicative of supportability, maintainability, or readiness problems. Our approach. is based upon fulfilling the need to recognize changes in the tolerance of equipment performance. This can be accomplished through the large-scale recording and analysis of test data that can aid in the performance of remote testing and recognition of tolerance changes and other issues that effect diagnostic ability. This would also facilitate taking appropriate corrective action to predict and/or compensate for such behavior before significant mission impact or failure occurs