UAV Adaptive Fault Tolerant Control and State Estimation
The aim of this project is to design and investigate the numerical implementation of an advanced fault tolerance control system for an Unmanned Air Vehicle (UAV) model using Sliding Mode Control (SMC) and Model Reference Adaptive Control (MRAC). The control system will be designed in the new framework of Generalized Dynamic Inversion (GDI) control, and it will require less knowledge of the UAV model's parameters to maintain closed loop stability and performance under mathematical model uncertainty in addition to failures in control surface actuators. Another aim of the project is the augmentation of the GDI control design with a state estimator such as a High Gain Observer (HGO) or a Luenberger observer to aid in estimating the aircraft states, and to evaluate the performance of the resulting estimator-based closed loop control system.
Design of Fractional Controllers: Implementation and Testing on Physical Processes
Currently, the use of the concept of fractional order differentiation in control theory is to consider an integer model of the system and use a fractional controller. The advantage of such controller resides in the non-integer order aspect of differentiation or integration therein. This additional degree of freedom can significantly improve the control laws used in classical integer controllers. The main objective of this project is to develop new structures of controllers in both transfer function and state space representations. The other objective of the project is the implementation of these fractional control laws experimentally. These tests will experimentally validate all theoretical results and show the superiority of fractional controllers over the classical integer controllers, which then can lead to their adoption to the control of industrial processes.
A Novel Class of Stochastic Gradient Adaptive Algorithms Based on q-Derivative
In this research project, we aim to develop a novel class of stochastic gradient adaptive algorithms which will be based on q-derivative. We will employ the q-derivative based gradient operator to the classical steepest decent optimization. Using this concept, we will derive a generalized least mean algorithm which we called as q- least mean (q-LM) algorithm. In addition to developing the generalized q-LM algorithm, we also propose to investigate the convergence behavior of the derived algorithm. Thus, both the convergence in the mean and the mean-square will be investigated. In this context, we aim to carry out both transient and steady-state analysis of the proposed algorithm. We will carry out extensive simulations to validate our theoretical findings.
Distributed Optimization for Cooperative Estimation and Control in Multi-agent Systems
Distributed cooperative estimation and control, where multiple agents collectively achieve estimation and control objectives in the presence of only local information and local interaction among neighbors, is of great significance in multi-agent systems enabling numerous applications. The main objective of the proposed research is to address the inherent challenges in distributed optimization for cooperative estimation and control.