Prognostic and Health Management (PHM), Intelligent Systems, and Systems and Controls
Dr. Bin Zhangs research falls into these three interconnected areas which are the keys to designing smart systems with self-situational-awareness and self-adapting capabilities to meet continuously growing requirements at different levels: performance (by advanced design and control), intelligence (by adaptation, learning, cooperation, and fusion), and reliability/survivability (by fault detection and isolation, failure prognosis, fault tolerance and reconfiguration). The three areas are integrated by a mix of theory, design, and implementation.
Conventional solutions to these requirements are often designed in a passive and ad-hoc fashion. For instance, the work may design a controller with a fixed structure and parameters based on limited a priori knowledge and the worst-case scenario to guarantee the required performance. In contrast, active approaches are able to monitor the changes in system health, environment and user behavior and adapt to such changes accordingly. These approaches, due to their promising features and solutions, attract more and more attention.
Dr. Zhangs research strives to design active approaches to achieve afore mentioned intelligent smart systems. His current work includes two aspects:
Diagnosis and prognosis using physics-based model and data-driven techniques, as well as computational intelligence techniques including pattern recognition and machine learning;
Fault-tolerance and health management of nonlinear systems using adaptive, learning, optimal, and intelligent analysis and control approaches.
These technologies not only maximize system safety, reliability, availability, and survivability, but also substantially reduce the logistic and maintenance cost. When large-scale complex systems are considered, particular attention needs to be paid to environment-and human-in-the-loop, as well as distributed and cooperative health monitoring. The application domains of interest to demonstrate my research results will be structure health monitoring, unmanned/manned vehicles, power distribution systems, multi-agent systems, and networked systems.
Safety-Critical Cyber-Physical-Human Systems (CPHS)
Dr. Wangs research of CPHS includes both automation and human interaction to ensure high levels of performance in safety-critical contexts. Although many approaches have been developed to deal with either physical or cyber failures, there are no systematic approaches that can handle cyber and physical failures simultaneously. The cyber-physical aspects of this research focus on resilient control with fault-tolerant, reduced complexity software architectures to enhance system safety and reliability. Human operators act as supervisory managers of automated systems providing enhanced levels of safety and efficiency. One of the challenges born of the human factor is how to present the actual status of automation to operators so that they can make appropriate control decisions in response to failures. This necessitates the design of system interfaces that provide operators with high levels of situation awareness of both controlled systems and automation itself. To integrate these aspects, we developed RSimplex architecture that integrates humans and autonomous cyber-physical technologies and compensates for each other’s weaknesses to achieve levels of safety and efficiency, transcending the limitations of either unaided humans or automation acting alone.
Reliable Coordination of Unmanned Vehicle Networks
Dr. Wangs research is concurrently focused on the difficulty of controlling unmanned aerial vehicle (UAV) networks; which, lies in the presence of physical, communication, and computation constraints, as well as different types of uncertainties. The limitation of communication and computation resources may bring in serious cyber effects, such as task jitters, delays, and packet loss. Modeling errors, exogenous disturbances, and potential physical failures raise uncertainties inside the system. All of these factors diminish the system predictability. As a result, the real system behavior may be far away from the ideal model and therefore violate the physical constraints of the system. Thus, the question is how to ensure reliable, predictable, and safe operation of UAV networks with limited resources in unpredictable and unstructured environments. Dr. Wangs project aims at reliable and cost-efficient methods to manage such systems.