Leonardo Cecchin

Syscop and Bosch Corporate Research

Thursday, February 13, 2025, 14:00 - 16:00

SR 02-016/18, Geb. 101

Abstract

Off-highway vehicles contribute substantially to air pollution, accounting for 15-20% of NOX and 25-40% of PM2.5 emissions among mobile pollution sources in the US and EU. The low efficiency of the hydraulic systems - often below 30% - calls for strategies to mitigate energy wastage. This thesis addresses the need for energy-efficient control approaches in hydraulic machinery, with a focus on Model Predictive Controller (MPC) strategies.

The study is focused on the modeling, design, and implementation of optimal control across various applications. MPC is central to this work, chosen for its capability to handle complex dynamics and constraints without requiring extensive case-by-case tuning. The work also explores data-driven modeling methods like Locally Weighted Regression (LWR) for function approximation and online learning, to manage low-volume, high-variety hydraulic components where physics-based models are impractical. The controllers are validated experimentally to evaluate the feasibility and effectiveness on real platforms.

Key contributions include the development and validation of several controllers: a nonlinear MPC for a Variable Speed Variable Displacement (VSVD) pump unit, an MPC for a single-cylinder system with Independent Metering Valves (IMV), and a distributed MPC framework for multi-cylinder IMV systems. Additionally, an MPC controller for grading assistance based on Local Model Networks (LMNs) was developed, along with an optimal trajectory planning approach for semi-autonomous excavators, achieving significant energy savings and productivity increase over current state-of-the-art methods.

The results underscore the potential of MPC to enhance energy efficiency, productivity, and accuracy of hydraulic systems. Although the computational burden is a challenge, this thesis shows how an efficient problem formulation and an appropriate solver choice enable real-time implementation. The results are applicable to a wide range of off-highway and industrial machinery, reducing environmental impact and improving productivity of hydraulic systems.

Commitee

Prof. Dr. Joschka Boedecker (Vorsitz)

Prof. Dr. Stephan Rupitsch (Beisitz)

Prof. Dr. Moritz Diehl (Prüfer)

Prof. Dr. Lorenzo Fagiano (Prüfer)