Advanced Analytics for Recurrent Event Data Analysis

ABSTRACT: This SBIR project will develop and implement a practical system for analysis and evaluation of usage and event data of fielded repairable parts and systems. The system will be capable of timely evaluation of relevant performance measures, thus facilitating the discovery of important trends and causes of substandard performance, and the identification of remedial actions. The effort will focus on supplementing Mean Time Between Failures (MTBF) with more refined and flexible measures of unit performance, including but not limited to Mean Cumulative Function (MCF). A comprehensive data analysis will connect individual histories of parts and systems in a unified framework, enabling early identification of"bad actors"that contribute disproportionally to system failures. The data analysis will also incorporate the impact of maintenance actions (scheduled and unscheduled). The root cause of substandard performance will be attributed in a statistically rigorous manner to specific factors that include the following: subpopulation of parts, subpopulations of systems hosting those parts, maintenance actions, and usage patterns. The developed procedure will allow ranking the contributing factors in terms of their relative impact on the performance. As a result, specific and actionable guidance will be provided aimed at restoring the system performance in a cost effective manner. BENEFIT: This project will develop software package for evaluating usage and event data by using repairs information available in current USAF systems. The developed algorithms will find trends, outliers, or anomalies in the data and suggest remedial actions. The project will demonstrate performance and the scalability of proposed approach, providing a clear path for technology transition into USAF programs and commercial aircraft. Applying the developed analytical software to data with unknown root causes of substandard performance will provide immediate actionable items for the Air Force. The developed software will be applicable in many commercial industrial applications such as commercial aviation, nuclear power systems, power transmission and distribution, and industrial process plants.