The goal of this project is to find structured controllers applicable to general transportation problems. The controllers should be structured in a way that facilitates an efficient implementation. Making them applicable for both large-scale systems, or for transportation systems with limited communication bandwidth.
The transportation systems will be modeled as a graph, where nodes correspond to a storage location for the quantity under transportation and the links to the
possible transportation paths. For example, for a logistic system, the nodes correspond to the different warehouses and stores, while the links correspond to the different truck routes used to transport the goods.
We will study the behavior around a nominal solution. This nominal solution could be found on a more detailed model, taking effects such as saturation, quantization, and congestion into account.
Most control designs are either centralized, requiring full knowledge of the system in every part of the system, or decentralized, using none of the information available in other nodes.
The controllers designed in this project will be a middle ground between the centralized and decentralized structures. The goal is that this will result in an optimal trade-off between controller performance and implementation complexity. This will result in using as much information as possible, while still allowing the controller to be implemented for systems with many nodes, or systems with limited communication bandwidth.
In 2021 the derived controller was applied to a detailed model for water irrigation networks where it was shown to have better performance than PI controllers.