Control reconfiguration in submersible vehicles using artificial neural networks

Abstract

Complex systems require highly sophisticated controllers to ensure that demanding performance specifications can be achieved under adverse operating conditions. In addition, these already elaborate controllers can be modified by augmentation to provide a control reconfiguration capability in the presence of actuator failure. This complicates the design and can lead to instability. An alternative approach to the reconfiguration problem is addressed in this paper. Two strategies (Linear Model Following, LMF, and Error Vector Suppression, EVS) are designed and tested for different failures of control surfaces. The LMF strategy is based on stored control laws tailored to each anticipated fault condition. The EVS strategy involves a control algorithm based on artificial neural networks which could significantly provide major simplifications to the conventional methodologies. Results obtained using this strategy for a deeply submerged submarine are compared with those obtained using the LMF strategy.