Prognosis Based Control Reconfiguration for an Aircraft with Faulty Actuator to Enable Performance in a Degraded State

Vishal V Mahulkar, Purdue University

Abstract

A conventional feedback control design for a process plant or vehicle system may result in unsatisfactory performance (even instability), in the event of malfunctions in actuators, sensors, or other components of the system. In order to overcome these limitations, new controllers need to be developed which are capable of tolerating component malfunctions whilst still maintaining desirable and robust performance and stability properties. In the case where the malfunctions are faults, additional fault growth dynamics is introduced which leads to a continuously changing system structure and response. Due to cost and safety requirements, the ability to accommodate faults constitutes a desirable characteristic which can be incorporated in the control design process of a high performance aircraft. The main objective of the research undertaken was to develop a framework for prognosis based control of systems with faults. The specific case of a high performance aircraft with a faulty hydraulic actuator for a control surface was the main topic of investigation. A full six degree of freedom nonlinear model for an F-16 aircraft was developed using Simulink. A full nonlinear model of the hydraulic actuator with an internal leakage fault due to a faulty seal was also developed and combined with the aircraft model. The aircraft model was then simplified through linearization to allow development of a prognosis based control reconfiguration strategy and implementation in real-time. The strategy was based on Model Predictive Control, in which an optimal control problem is solved every iteration. To allow real-time implementation, the optimal control problem was broken into an offline component and an online component. Offline component was used to generate a static map, which was employed for prognosis. The online component used this static map to solve mixed integer programming problem to optimize response speed and trajectory, satisfying constraints on degradation. The effectiveness of the reconfiguration strategy was demonstrated for simple missions in a longitudinal plane. At the execution level, a divided difference filtering algorithm was implemented to identify parametric faults in the control actuator. The fault information was then used in an adaptive framework to develop a fault tolerant controller for the actuator. A hydraulic actuator test bench was designed to validate the developed models and control strategies through hardware-in-the-loop simulations. Experimental results demonstrated the effectiveness of the fault identification algorithm and the performance improvement obtained through implementation of the adaptive control strategy. The effectiveness of the reconfiguration strategy was also demonstrated experimentally by implementing an unknown wear function and comparing the final degradation in absence and presence of the reconfiguration algorithm.

Degree

Ph.D.

Advisors

Adams, Purdue University.

Subject Area

Mechanical engineering

COinS