Model Predictive Control (MPC) is a well established control mehtod, known for its high control performances and its capability of directly handling constraints. One of the main problems of MPC is its poor robustness to disturbances acting on the system and model uncertainties, thus requiring Robust Model Predictive Control formulations. A large quantity of approaches have been proposed in literature, which result in high computational complexity and high conservativeness of the solution. A possible approach to overcome these problems, instead, is based on Sliding Mode Control (SMC) and Integral Sliding Mode Control (ISMC), which is known as Sliding Mode Predictive Control. In this work, Sliding Mode Predictive Control approach is treated by exploiting the combination of Model Predictive Control and Integral Sliding Mode Control in four different control laws: two in which ISMC is adapted from continuous time formulations and two direct synthesis formulations. The controllers are implemented and tested on a Double Mass-Spring-Dumper system affected by different classes of disturbances, along with Offset-Free MPC for having an as fair as possible comparison. The control metrics and the qualitative results show that all ISMC formulations can guarantee robustness by rejecting uncertainties, while Offset-Free can only deal with step-like disturbances. Moreover, direct synthesis formulations result to have better performances than the ones adapted from continuous time.
Model Predictive Control (MPC) è una tecnica di controllo estremamente diffusa, nota per garantire alte prestazioni e per la sua capacità di gestire i vincoli in maniera diretta. Uno dei problemi principali di MPC è la sua scarsa robustezza rispetto a disturbi che agiscono sul sistema e alle incertezze sul modello, richiedendo pertanto formulazioni robuste di Model Predictive Control. In letteratura sono stati proposti numerosi approcci, che risultano in un'elevata complessità computazionale e in una forte conservatività della soluzione. Un possibile approccio per risolvere queste problematiche, invece, si basa su Sliding Mode Control (SMC) e su Integral Sliding Mode Control (ISMC), ed è noto come Sliding Mode Predictive Control. In questo elaborato Sliding Mode Predictive Control è approfondito, utilizzando la combinazione di Model Predictive Control e Integral Sliding Mode Control con quattro diverse leggi di controllo: due in cui ISMC è adattato da formulazioni a tempo continuo e due formulazioni derivanti da sintesi diretta. I controllori vengono implementati e testati su un sistema Doppia Massa-Molla-Smorzatore su cui agiscono diverse classi di disturbi, insieme ad Offsete-Free MPC, per ottenere dei paragoni quanto più imparziali è possible. Le metrice di controllo e i risultati qualitativi ottenuti mostrano che tutte le formulazioni ISMC sono in grado di garantire la robustezza reittando le incertezze, mentre Offset-Free è in grado di gestire solo i disturbi a gradino o equivalenti. Inoltre, le formulazioni a sintesi diretta risultano avere migliori prestazioni rispetto a quelle adattate dal tempo continuo.
Enanching Robustness of Model Predictive Control through Complementary Techniques
ROMANO, MICHELANGELO
2023/2024
Abstract
Model Predictive Control (MPC) is a well established control mehtod, known for its high control performances and its capability of directly handling constraints. One of the main problems of MPC is its poor robustness to disturbances acting on the system and model uncertainties, thus requiring Robust Model Predictive Control formulations. A large quantity of approaches have been proposed in literature, which result in high computational complexity and high conservativeness of the solution. A possible approach to overcome these problems, instead, is based on Sliding Mode Control (SMC) and Integral Sliding Mode Control (ISMC), which is known as Sliding Mode Predictive Control. In this work, Sliding Mode Predictive Control approach is treated by exploiting the combination of Model Predictive Control and Integral Sliding Mode Control in four different control laws: two in which ISMC is adapted from continuous time formulations and two direct synthesis formulations. The controllers are implemented and tested on a Double Mass-Spring-Dumper system affected by different classes of disturbances, along with Offset-Free MPC for having an as fair as possible comparison. The control metrics and the qualitative results show that all ISMC formulations can guarantee robustness by rejecting uncertainties, while Offset-Free can only deal with step-like disturbances. Moreover, direct synthesis formulations result to have better performances than the ones adapted from continuous time.File | Dimensione | Formato | |
---|---|---|---|
Romano_Michelangelo.pdf
accesso riservato
Dimensione
4.62 MB
Formato
Adobe PDF
|
4.62 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/69269