Harmonic Control techniques aimed at reducing tonal disturbances have been extensively studied in the last few decades, with particular attention to a representation of the system as a linear model constructed in the frequency domain, the T-matrix model. The precise knowledge of its elements is necessary for a proper functioning of the overall control system, and classical employed controllers resorting to the linear quadratic theory have not be framed to deal with model and parametric uncertainties or nonlinearities, possible causes for degraded performance or instability of the closed-loop system. Adaptive control variants, coupled with a suitable offline or online identification method, have been employed in literature to handle this problem, but very little effort has been devoted to the analysis of the trade-off between robustness and adaptation in their deployment. In this dissertation, a discrete-time H_inf approach and a systematic methodology to the design of a robust Harmonic Control algorithm is proposed, for both SISO and MIMO system representations. The proposed control solution allows to account for model and parametric uncertainty in the control design problem, and provides a further benefit when dealing with the tuning problem, in particular when a multivariable plant is considered and different performance requirements are associated to the disturbance attenuation on each considered output. Indeed, specifications in terms of steady attenuation levels and desired transient performance can be directly incorporated in the robust problem statement. Moreover, the control design approach has been modified to deal with the explicit accounting for also predictable changes of the system, such as the ones induced by actuator characteristics and nonlinearities. Their role and effects on closed-loop performance and stability have been considered by resorting to analysis and synthesis frameworks, as the Describing Function and Linear Parameter Varying approaches represent. Both are introduced as modifications of the original T-matrix model. While the first is used to analyze the cascade connection of a static nonlinear function with the plant matrix, the second allows to recast the problem both from a modeling and control design perspectives, combining the advantages of the robust control solution with the accounting of predictable plant changes handled within a gain-scheduling framework. The Thesis ends with a validation of the proposed methodologies on a classical Harmonic Control application like the helicopter's rotor vibration problem. As known, it can be formulated in terms of compensating a periodic disturbance at rotor frequency acting at the output of an uncertain (possibly time-periodic) linear system. Three case studies are proposed, including the general rotor-induced vibration problem, its variant based on semi-active lag dampers and the structural noise/vibration problem.
Negli ultimi decenni notevoli progressi sono stati raggiunti nell’ambito delle tecniche di controllo armonico volte a ridurre disturbi tonali e multitonali, con particolare attenzione a una rappresentazione del sistema come un modello lineare costruito nel dominio della frequenza, il modello T-matrix. La conoscenza esatta dei suoi elementi è fondamentale e necessaria per un corretto funzionamento del sistema di controllo generale, e tecniche classiche quali la teoria quadratica lineare si sono rivelate finora non adatte a trattare in maniera adeguata incertezze di modello o parametriche, piuttosto che non linearità che possano essere causa di una riduzione notevole di prestazioni fino all'instabilità del sistema in anello chiuso. Varianti adattative sono state impiegate in letteratura per gestire questo tipo di problema, ma in generale non si è dato il giusto peso all'analisi del compromesso tra robustezza e adattamento. In questo lavoro ci si è concentrati sullo sviluppo di un approccio H_inf a tempo discreto e una metodo sistematico per la progettazione di algoritmi di controllo armonico con proprietà di robustezza, valutandone le performance a fronte di rappresentazioni sia SISO che MIMO del sistema. La soluzione di controllo proposta consente di inserire a monte, in fase di design, le incertezze parametriche e di modello, e fornisce un ulteriore vantaggio in termini di tuning, in particolare quando si tratta un plant fortemente multivariabile in cui siano richiesti requisiti di prestazioni diverse sulle molteplici uscite considerate. Infatti, le specifiche in termini di livelli di attenuazione e durata del transitorio possono essere direttamente incorporati nella formulazione del problema robusto. Inoltre, l'approccio al progetto del controllore è stato modificato per trattare in maniera esplicita le incertezze che siano in qualche modo prevedibili, come quelli indotte dalle caratteristiche degli attuatori e dalla presenza di non linearità. Il loro ruolo e gli effetti sulla stabilità e le prestazioni in anello chiuso e sono stati considerati ricorrendo a metodi di analisi e sintesi quali la funzione descrittiva e rappresentazioni LPV del sistema sotto controllo. Entrambi sono introdotti come modifiche al modello T-matrix originale. Mentre il primo viene utilizzato per analizzare il collegamento in cascata tra l’elemento di non linearità e il plant, il secondo permette di riformulare il problema di controllo a monte, unendo i vantaggi di una soluzione robusta affiancata a un framework gain-scheduling. La tesi si conclude con alcuni esempi atti a validare le metodologie proposte, concentrandosi in particolare sul problema delle vibrazioni indotte dal rotore di un elicottero.
Robust harmonic control for disturbance rejection: methods and applications
MURA, ROBERTO
Abstract
Harmonic Control techniques aimed at reducing tonal disturbances have been extensively studied in the last few decades, with particular attention to a representation of the system as a linear model constructed in the frequency domain, the T-matrix model. The precise knowledge of its elements is necessary for a proper functioning of the overall control system, and classical employed controllers resorting to the linear quadratic theory have not be framed to deal with model and parametric uncertainties or nonlinearities, possible causes for degraded performance or instability of the closed-loop system. Adaptive control variants, coupled with a suitable offline or online identification method, have been employed in literature to handle this problem, but very little effort has been devoted to the analysis of the trade-off between robustness and adaptation in their deployment. In this dissertation, a discrete-time H_inf approach and a systematic methodology to the design of a robust Harmonic Control algorithm is proposed, for both SISO and MIMO system representations. The proposed control solution allows to account for model and parametric uncertainty in the control design problem, and provides a further benefit when dealing with the tuning problem, in particular when a multivariable plant is considered and different performance requirements are associated to the disturbance attenuation on each considered output. Indeed, specifications in terms of steady attenuation levels and desired transient performance can be directly incorporated in the robust problem statement. Moreover, the control design approach has been modified to deal with the explicit accounting for also predictable changes of the system, such as the ones induced by actuator characteristics and nonlinearities. Their role and effects on closed-loop performance and stability have been considered by resorting to analysis and synthesis frameworks, as the Describing Function and Linear Parameter Varying approaches represent. Both are introduced as modifications of the original T-matrix model. While the first is used to analyze the cascade connection of a static nonlinear function with the plant matrix, the second allows to recast the problem both from a modeling and control design perspectives, combining the advantages of the robust control solution with the accounting of predictable plant changes handled within a gain-scheduling framework. The Thesis ends with a validation of the proposed methodologies on a classical Harmonic Control application like the helicopter's rotor vibration problem. As known, it can be formulated in terms of compensating a periodic disturbance at rotor frequency acting at the output of an uncertain (possibly time-periodic) linear system. Three case studies are proposed, including the general rotor-induced vibration problem, its variant based on semi-active lag dampers and the structural noise/vibration problem.File | Dimensione | Formato | |
---|---|---|---|
Mura_PhD_Thesis.pdf
Open Access dal 30/12/2016
Descrizione: Tesi completa
Dimensione
7.18 MB
Formato
Adobe PDF
|
7.18 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/117514