The U.S. Air Force strives to maintain high operational availability of assets and has ongoing needs to reduce total ownership costs, increase reliability, and extend the life-cycle of equipment to improve overall readiness. Most machines generate vibrations, and vibration analysis is key to detecting machinery degradation before the equipment fails. Machine faults can be diagnosed by changes in modal parameters, such as natural frequency, damping, stiffness, etc. The statistical features of vibration signals in the time, frequency, and time-frequency domains each have different strengths for detecting fault patterns. Integration and hybridization of feature extraction algorithms can yield synergies that combine strengths and eliminate weaknesses. Advent Innovations proposes to develop Ensemble Feature Extraction (EFEX) Algorithms and Software for Machine Fault Classification. Advanced signal processing and machine learning methods will be developed to enhance the sensitivity, accuracy, efficiency, and specificity of fault classification through vibration auditing. In order to meet this challenge, Advent will team with the University of South Carolina (USC). Advent will evaluate the equipment failure modes and degradation, develop the open system software architecture, and conduct feasibility testing and demonstration. USC will develop the underlying statistical feature extraction algorithms in the time, frequency, and time-frequency domains.
