Projects
Graduate
Fault diagnosis and prognosis of satellite actuators including reaction wheels and control moment gyros
Real-time object detection in industrial settings for automated inspection and quality check procedures
Development of a Fan & AGS Control Strategy for AC Duty Cycles
Undergraduate
Machine Learning applied to Reaction Wheels Fault Diagnosis
Reaction wheels (RW) are one of the most commonly used actuators on-board satellites. A RW is a flywheel mounted on an electric motor that produced momentum by spinning the flywheel at different speeds. This project is concerned with using machine learning (ML) algorithms to detect faults in a RW unit. The data for this project would be obtained using highly-nonlinear model of a RW developed by Bialke. There will be output for healthy conditions and multiple faulty conditions with the goal of using machine learning methods to find the source of fault.
Research Tasks and Activities:
1. Conducting literature review on RW faults and ML algorithms to detect and diagnose them
2. Running simulations based on the highly-nonlinear model of RW developed by Bialke
3. Store output of the simulations for healthy conditions and various faulty conditions of the RW
4. Apply various ML algorithms on the generated data to find the root cause of faults
Design, Fabrication, and Test of a Single Rotor Modular UAV
Modular unmanned aerial vehicles (UAV) are becoming more and more popular due to their size, cost of manufacturing and flexibility in applications. This project is concerned with sizing/design of a single rotor modular UAV suitable to fit payloads including a camera, a sampling unit, or a dispenser compartment while is controllable and stabilizable within the range of operation required. After design, the fabrication of the unit would proceed to have a working prototype. Finally, some initial tests will be conducted to examine the performance of the UAV through various scenarios.
Research Tasks and Activities:
1. Conducting literature review on sizing and design of a single rotor UAV
2. Design/sizing of a single rotor UAV
3. Fabrication of a single rotor UAV using commercial-off-the-shelf (COTS) and 3D printing
4. Test and data analysis on the single rotor UAV
Machine Learning applied to Reaction Wheels Fault Prognosis
Reaction wheels (RW) are one of the most commonly used actuators on-board satellites. A RW is a flywheel mounted on an electric motor that produced momentum by spinning the flywheel at different speeds. This project is concerned with using machine learning (ML) algorithms to determine the remaining useful life of defective RW units. The data for this project would be obtained using highly-nonlinear model of a RW developed by Bialke. There will be output for healthy conditions and multiple faulty conditions with the goal of using machine learning methods to find the source of fault and predict the remaining useful life of the unit.
Research Tasks and Activities:
1. Conducting literature review on RWs and ML algorithms for fault prognosis
2. Running simulations based on the highly-nonlinear model of RW developed by Bialke
3. Store output of the simulations for healthy conditions and various faulty conditions of the RW
4. Apply various ML algorithms on the generated data to predict the remaining useful life