Search papers, labs, and topics across Lattice.
This paper addresses the need for accurate and adaptable energy consumption models for underwater gliders (UGs) by challenging the uniform oil bladder mass distribution assumption in existing models. They derive a gray-box model (GBM) that separately analyzes the kinetic energy of the piston and cavity to quantify the relationship between buoyancy, mass displacement, and pitch angle. The GBM is trained using kriging modeling and updated online via recursive least squares, demonstrating improved estimation accuracy and adaptability compared to conventional dynamics-based models through simulations and sea trials.
By separately modeling piston and cavity kinetics, this gray-box model delivers more accurate, real-time energy consumption estimates for underwater gliders, crucial for extending mission life.
The energy consumption model of the underwater glider (UG), which depends on its gliding parameters, serves as a crucial foundation for motion planning and energy management. Developing a model that aligns with real-world application scenarios and offers high estimation accuracy is of great importance. In this article, we challenge the assumption of uniform oil bladder mass distribution, building upon the baseline energy consumption model. We separately analyze the kinetic energy of the piston and the cavity to quantify the coupled relationship between net buoyancy, movable mass displacement, and pitch angle. Using the kriging modeling method, we then train a single-profile gray-box model (GBM) for UG energy consumption. The GBM is not only structurally simple but also capable of online updating based on real-time gliding data using the recursive least squares algorithm to compensate for the effects of unmodeled factors. Data from hardware-in-the-loop simulation, high-fidelity dynamic model, and sea trials are used to validate the model’s estimation accuracy and online updating capability. The cosimulation results show that the GBM has good estimation accuracy and effective online updating, which is a significant advantage over the energy consumption model derived from conventional dynamics.
