Research Goals
The primary challenge to controller design is the uncertainty in the changing physical system (as a result of faults due to component aging, damage, and failure), operating environment, and user behaviors. A conventional approach is to passively design a robust controller with a fixed structure based on limited a priori knowledge and/or the worst-case scenario, hence trading performance, service life, and operability for robustness. In contrast, intelligent control will actively on-line monitor and predict the changes in system health, environment, and user behaviors, and adapt to such changes accordingly.
Our research strives to bring us closer to this goal of self-situational-awareness and self-adapting intelligent systems. Our current focus is on health management and intelligent control systems, aiming to achieve automatic detection and isolation of faulty components, prediction of component remaining useful life, and safe and graceful performance degradation to avoid catastrophic failure, hence potentially offering maximization of productivity, safety, reliability, availability, affordability, and maintenance cost reduction.
Research Projects
- Lead Acid Battery State-of-Health Monitoring (Sponsored by General Motors R&D Center): The number of electrical devices in modern vehicles has been rapidly increasing in the last two decades, and this trend will accelerate. The electric power management system serves to balance the power demanded and supplied as well as to ensure the vehicle's start-up ability. To achieve these goals, accurate and reliable knowledge of the battery state is essential. In this collaborative project with General Motors R&D Center, we propose to develop and evaluate several different battery state-of-health (SOH) monitoring methods with different sensor options.
- Distributed Fault Diagnosis and Fault-Tolerant Control (Sponsored by Wright State University's Research Challenge Program) . The design and analysis of health management systems for large-scale networked control systems (NCS) poses significant new challenges beyond traditional techniques utilizing a centralized architecture. The systems are controlled by many different and possibly distant local control systems via a communication network. Each local control system only has access to limited information of the overall system, because of constraints of communication bandwidth and computational power. However, in the presence of a catastrophic failure in a subsystem, the effect of the fault will be quickly propagated to other subsystems due to their interconnections. Supported by Wright State University's Research Challenge fund, the objective of this project is to develop real-time fault detection, isolation, and health management strategies to improve fault-tolerance of of the overall system.
- Multi-Agent Health Management System (Sponsored by Army TARDEC). In this project, Wright State University, in collaboration with Intelligent Automation Inc. (IAI) and General Motors R&D Center, proposes to develop a novel Automated Algorithm Generator (A2G) system for distributed health management of fleet-wide ground vehicles. The A2G system will automatically generate and maintain diagnostic/prognostic algorithms for any vehicle/platform based on fleet-wide statistics and trending. The A2G system will be developed under IAI's multi-agent software infrastructure called Cybele.
- A Nonlinear Adaptive Approach to Isolation of Sensor Faults and Component Faults (Sponsored by NASA Glenn Research Center). In this project, Wright State University, in collaboration with Impact Technologies, LLC and Pratt & Whitney, proposes to develop an innovative nonlinear adaptive method for detecting and isolating sensor faults, actuator faults, and component faults for jet engines. The effectiveness of algorithms will be demonstrated using NASA's C-MAPSS engine model and Pratt & Whitney's PW6000 engine model.
- Fault Prognosis of Automotive Electrical Power Generation
and Storage System (Sponsored by General Motors R&D Center).
In this collaborative project with General Motors R&D Center, the objective is to develop a system level approach to fault prognosis of automotive electric power generation and storage system. Specifically, we will develop a parity-relation-based model characterizing the correlations among multiple system signals under normal operating conditions. The system health degradation will be inferred from deviations from such normal correlations among system signals. Furthermore, based on the responses of different parity-relations, the health status of the alternator and the battery can be estimated and predicted, respectively.
- Autonomous Data Fusion for Distributed Sensing with Unmanned Vehicles (Sponsored by Office of Naval Research).
Wright State University, in collaboration with Impact Technologies, LLC, proposes to develop and demonstrate a decentralized autonomous system that fuses the data from heterogeneous, geographically dispersed sensors deployed on unmanned vehicles into an integrated ISR dataset for mission planning on Littoral Combat Ships.
If you are interested in joining the Intelligent Control System lab, please contact Dr. Frank Zhang.
