GreenSight proposes to develop Automated Machine-Learning Prediction of Motor Degradation (AMP MD), a novel new system to detect motor issues, specifically insulation failures of stators, before the issue results in complete motor failure. The system uses the magnitude of current delivered to each phase of the motor to detect issues, a machine learning algorithm to detect anomalous behavior, and a classifier to inform the operator whether the motors are good, degraded, or have encountered a severe fault. The system uses machine learning models to characterize data from motor tests, outputs from a digital twin, and GAN augmented data to build a robust data set for the models to train effectively. The final system will consist of current sensors as well as a data acquisition system and machine learning software project that can seamlessly integrate on future Urban Air Mobility platforms to provide operators with actionable information about safety critical systems lending itself to a clearer path toward flight certification and authorization. GreenSight brings extensive experience in UAS flight control software and automation powered by machine learning to produce high reliability systems. Potential NASA Applications (Limit 1500 characters, approximately 150 words): The proposed AMP MD work fills critical gaps outlined in 2022 SBIR topic area A1.06. AMP MD by providing new methods to detect safety critical faults before cascading failures. This system can integrate seamlessly into an eVTOLs electronic speed controllers and embedded control system. The technique will enable rapid progress towards NASA objectives for Urban Air Mobility (UAM) and Advanced Air Mobility (AAM), to meet the high assurance levels required for aviation certification. Potential Non-NASA Applications (Limit 1500 characters, approximately 150 words): AMP MD fills an industry requirement to improve reliability of safety critical components prior to authorizing the future Urban Air Mobility vehicles for commercial operations. This system can easily be adapted for commercial and military unmanned aircraft systems that require FAA approval prior to more widespread adoption of the available aircraft. Duration: 6
