Homeostatic-inspired controller algorithm for a hybrid-driven autonomous underwater glider

• Main Author: Isa, Khalid
• Format: Thesis
• Language: English
• Published: 2015
• Subjects: TK Electrical engineering. Electronics Nuclear engineering; TK7800-8360 Electronics
• Online Access: http://eprints.uthm.edu.my/1733/1/24p%20KHALID%20ISA.pdf; http://eprints.uthm.edu.my/1733/

The autonomous hybrid-driven underwater glider presented in this thesis combines the concept of a buoyancy-driven underwater glider and a conventional autonomous underwater vehicle. It is classified as a new kind of autonomous underwater glider. The existing buoyancy-driven gliders have proven to be powerful tools in oceanographic applications. This is because they are inexpensive, high-endurance and energy-efficient. However, they still have weaknesses in terms of speed and manoeuvrability due to the under-actuated system; relatively slow; limited propulsion forces; and limited external control surfaces. Furthermore, it is difficult to control the glider because of the high nonlinearity and complexity of the glider dynamics, coupled with the underwater environments and disturbances. Thus, the main objective of this research is to design and develop a controller algorithm that is able to make the glider adaptive despite facing these constraints. A robust and reliable homeostatic-inspired controller system has been designed for this purpose. The controller is able to adapt efficiently to the dynamically changing conditions and is able to compensate the disturbance from water currents. The controller algorithm has been designed based on the human innate control mechanism by integrating three artificial systems: the artificial neural network (ANN), artificial endocrine system (AES), and artificial immune system (AIS). According to simulation results of control methods benchmarking, the homeostatic controller was able to achieve the desired pitch angle at the fastest settling time, which was 12.5 seconds faster than the model predictive control (MPC); 9 seconds faster than the linear-quadratic regulator (LQR); 6.5 seconds faster than the neural network (NN) controller; and 3.75 seconds faster than the neuroendocrine controller. In addition, the homeostatic controller was able to optimise the ballast mass and distance of the sliding mass in order to achieve the desired pitch angle by shortening the sliding mass distance up to 53.7% and reducing the ballast mass up to 17.7% when compared with the LQR and MPC. Overall, the homeostatic controller has achieved the best performance compared with the LQR, MPC, NN and neuroendocrine controllers. Furthermore, the validation analyses between the simulation and experimental results have shown that the homeostatic control system produces very satisfactory performance, with the homeostatic controller able to achieve the desired angle.