Abstract
The thesis addresses two different - but related - problems: the average consensus problem over unreliable networks, and the cooperative adaptive cruise control in presence of packet losses.
The average consensus problem is more challenging to solve when the network is changing over time due to edges between nodes being broken due to communication failures between nodes. While the convergence of the network can usually still be achieved when some links are unavailable for transmission, the nodes might converge to a nal state that is not the average of the initial states of the network. The thesis presents a novel compensation method (the Average Preserving compensation method), based on changing the weights of the weighted adjacency matrix that describes the network, and a basic transmission algorithm to implement such synchronous compensation mechanism while
using asynchronous transmissions between nodes. Moreover, we improved the basic Average Preserving compensation method, including a gain to increase the speed of convergence of the network, providing a set of conditions to compute the gain and guarantee convergence (the so-called safe gain) as well as an empirical method to compute the gain, still safeguarding the convergence of the network while obtaining a faster convergence speed.
The Cooperative Adaptive Cruise Control problem under packet losses is tackled by first introducing the novel concepts of input string stability, input mean string stability and input stochastic string stability, that allow to obtain conditions to decrease the propagation of the error in the downstream direction of the platoon when the communication between cars is susceptible of packet losses. Using the stochastic and mean string stability concepts, we first developed a simpler controller to obtain a satisfactory (and not string stable) behaviour for the platoon, and then designed an H1 controller and an Unknown Input Observer to obtain a stochastic string stable behaviour for a platoon of
vehicles experiencing disruption in the communication.
Lastly, our results were confirmed by simulations, which can be supported by experiments as future work.
The average consensus problem is more challenging to solve when the network is changing over time due to edges between nodes being broken due to communication failures between nodes. While the convergence of the network can usually still be achieved when some links are unavailable for transmission, the nodes might converge to a nal state that is not the average of the initial states of the network. The thesis presents a novel compensation method (the Average Preserving compensation method), based on changing the weights of the weighted adjacency matrix that describes the network, and a basic transmission algorithm to implement such synchronous compensation mechanism while
using asynchronous transmissions between nodes. Moreover, we improved the basic Average Preserving compensation method, including a gain to increase the speed of convergence of the network, providing a set of conditions to compute the gain and guarantee convergence (the so-called safe gain) as well as an empirical method to compute the gain, still safeguarding the convergence of the network while obtaining a faster convergence speed.
The Cooperative Adaptive Cruise Control problem under packet losses is tackled by first introducing the novel concepts of input string stability, input mean string stability and input stochastic string stability, that allow to obtain conditions to decrease the propagation of the error in the downstream direction of the platoon when the communication between cars is susceptible of packet losses. Using the stochastic and mean string stability concepts, we first developed a simpler controller to obtain a satisfactory (and not string stable) behaviour for the platoon, and then designed an H1 controller and an Unknown Input Observer to obtain a stochastic string stable behaviour for a platoon of
vehicles experiencing disruption in the communication.
Lastly, our results were confirmed by simulations, which can be supported by experiments as future work.
Original language | English |
---|---|
Qualification | Doctor of Philosophy |
Awarding Institution |
|
Supervisors/Advisors |
|
Award date | 17 Jun 2022 |
Place of Publication | Enschede |
Publisher | |
Print ISBNs | 978-90-365-5380-3 |
DOIs | |
Publication status | Published - 17 Jun 2022 |