Thursday, November 14
10:00 AM – 3:30 PM
Room 204
Co-Chair: Dr. Prakash Patnaik, NRC Aerospace Research Centre, Ottawa, Canada
Co-Chair: Prof. Sreeraman Rajan, Systems & Computer Engineering, Carleton University, Ottawa Canada
Agenda
10:00 AM-10:30 AM: Prof. Mohamed Atia, Topic: Navigation in Satellite denied environments: Challenges and Possible Solutions
10:30 AM-11:30 AM: Keynote by Dr. Prakash Patnaik Topic: Sensors for Aerospace Applications
11:30 AM-12:00 PM: Dr. Roy ChihChung Wang, Topic: Independent samples from low-dimensional and compactly supported unnormalized densities
12:00 PM-13:30 PM: Lunch Break (on your own)
1:30 PM-2:00 PM: Mr. David Luong, Topic: Quantum Radars and Aerospace Systems
2:00 PM-3:00 PM: Prof. R. Doraiswami, Topic: Automated Proactive and Predictive Health and Usage Monitoring Systems
Keynote:
Sensors for Aerospace Applications
By Dr. Prakash Patnaik
Principal Research Scientist
Aerospace Defence Science & Technology
National Research Council Canada
Aerospace Research Centre, Ottawa, Ontario, Canada
Abstract: This presentation will address topical themes of sensors for aerospace operational needs. The spectrum of sensor requirements on-board a modem aircraft (both military and civil markets) is extremely broad and complex. The impact of new system architectures and technology developments, places new requirements on sensor performance and reliability. This presentation will review the development of advanced sensors for aerospace applications, e.g., engine and structures, highlighting the need for product quality and integrity whilst recognizing the increasing needs for their applications in extreme operating environments.
AN AUTOMATED PROACTIVE AND PREDICTIVE HEALTH AND USAGE MONITORING SYSTEMS
Rajamani Doraiswami (1) and Sreeraman Rajan (2)
1 Department of Electrical and Computer Engineering, University of New Brunswick, Fredericton
New Brunswick, Canada dorai@unb.ca
2 Systems and Computer Science, Carleton University, Ottawa, Ontario, Canada sreeramanr@sce.carleton.sa
Abstract: An automated Health and Usage Monitoring Systems (HUMS), fault detection and isolation (FDI), condition-based maintenance (CBM), and predictive maintenance using a hierarchical approach is proposed for a wider class of complex, nonlinear aerial vehicles including aircrafts and drones, whose outputs are corrupted by stochastic and deterministic disturbances. Critical components are monitored in real time to avoid developing excessive stresses and vibrations A linear model is derived using the computed torque approach by introducing a new input to replace all the nonlinear terms including the applied, Coriolis, and gravitational forces. The resulting system formed of the aerial vehicle and the disturbances is described using Box-Jenkins (BJ) model. In order to develop proactive and predictive maintenance, an emulator is connected to all accessible inputs and its parameters are varied to mimic likely scenarios that system can encounter. The system and the associated Kalman filter are identified using the emulator-generated data. The vehicle model, termed signal model, and disturbance-free output¸ termed signal, are respectively the identified model and its output. To meet the critical requirement of high probability of correct decision with low false alarm probability, a fusion of model-based and model-free approach employed using the estimated signal and not the corrupted data. Model-free approach includes Limit-checking and Plausibility analysis (LP); artificial neural network (ANN); and fuzzy inference (ANFIS) system. The Kalman filter is the backbone of model-based scheme. A fault is quickly detected via the faster of the two schemes and the health of the subsystems such as the actuator, and sensors is unfolded sequentially. Decisions from model-free and model-based approaches are weighted by a Bayes classifier. The proposed scheme was successfully evaluated.
The FDI of the faulty subsystems such as the sensor, the actuator or others are accurately determined using a Bayes’ Classifier and tagged for maintenance. The proposed scheme is evaluated on a simulated system.
Quantum Radars and Aerospace Systems
David Luong (1), Sreeraman Rajan (1) and Balaji Bhashyam (2)
1. Department of Systems and Computer Engineering, Carleton University, Ottawa, Ontario, Canada
2. Defence Research and Development Canada, Ottawa, Ontario, Canada
Abstract: Recently, much attention has been focused on radars which rely on quantum mechanical phenomena to enhance detection performance. In this talk, a practically realizable quantum radar scheme will be presented, and possible applications to the detection of aircraft are discussed.
Navigation in Satellite-denied Environments, Challenges and Possible Solutions
Mohamed Atia
Department of Systems and Computer Engineering, Carleton University, Ottawa, Ontario, Canada
Abstract: Global Positioning System (GPS), the first operational satellite-based positioning system, is a powerful system that dramatically expanded autonomous systems capabilities to navigate anywhere anytime worldwide. However, GPS, and other satellite-based systems, is vulnerable to interference and jamming. This presentation will discuss how sensor fusion and multi-sensors integrated navigation technologies can enhance the accuracy and expand the navigation and positioning services in GPS-challenging environments. The presentation will discuss new emerging technologies used for multi-sensors navigation such as inertial sensors, laser scanners, vision sensors and radar. The talk will discuss how these sensors can be fused on-board in real-time to support the operation of unmanned aerial vehicles in satellite-denied environments.
Independent samples from low-dimensional and compactly supported unnormalized densities
Roy.C-C. Wang, Sreeraman Rajan, James. R. Green
Department of Systems and Computer Engineering, Carleton University, Ottawa, Ontario, Canada
Abstract: Uncertainty quantification has many applications in risk management and anomaly detection. Sequential Monte Carlo is a method for online uncertainty quantification via a Bayesian inference framework. Essential to this family of simulation-based methods is the ability to draw independent samples from the prior or conditional probability densities.
Our current work in progress is about using a particular transport map, called the Knothe-Rosenblatt rearrangement, to draw independent samples from a given unnormalized density that satisfies certain constraints.