Back to results
US 12,572,781 B2Granted
Virtual-sensing integrated for predictive reliability
AssigneeGeorgia Tech Research Corporation
InventorsDavid Alvord, David I. Pendleton, Aimee N. Williams
Life of this patent
Estimating the term…
Abstract
An exemplary virtual sensing method and system are disclosed for predictive reliability (VIPR) procedure and/or controls that employ artificial intelligence and machine learning (AI/ML), particularly deep neural networks and multi-modal deep learning, with vehicle sensor data to create virtual sensors. The virtual sensors can be used to estimate measurements and operating conditions in a hostile environment in rockets and vehicle systems.
Claims18
A virtual sensing deep neural network comprising:
- a deep neural network trained with a plurality of sensor data and plume image data measured from a plurality of sensors and a camera sensor of a test cell housing an engine,
- wherein the deep neural network is configured to generate virtual plume image data from a second plurality of sensor data of a jet engine or rocket motor, wherein the virtual plume image data is analyzed to determine a presence and localization of shock wave data in the virtual plume image data, and wherein the presence and localization of shock wave data in the virtual plume image data are used to determine a virtual sensed Mach number or an internal pressure of the jet engine or rocket motor, wherein the generated virtual plume image data is employed in the control or monitoring of the jet engine or rocket motor.
A method of training a virtual sensing deep neural network, the method comprising:
- providing a deep neural network trained with a plurality of sensor data and plume image data measured from a plurality of sensors and a camera sensor of a test cell housing an engine,
- generating, using the trained deep neural network, virtual plume image data from a second plurality of sensor data of a jet engine or rocket motor, wherein the generated virtual plume image data is analyzed to determine a presence and localization of shock wave data in the virtual plume image data;
- determining a virtual sensed Mach number or an internal pressure of the jet engine or rocket motor based on the presence and localization of shock wave data in the virtual plume image data;
- and
- outputting, via a report or to a controller, the virtual sensed Mach number or an internal pressure of the jet engine or rocket motor,
- wherein the generated virtual sensed Mach number or an internal pressure is employed in the control or monitoring of the jet engine or rocket motor.
3 more elements
Dependencies are inferred from the claim text.