In a rapidly changing context, dealing and managing with natural resources has become an increasing problem over the years. The world is facing a huge hydroclimatic and socioeconomic change and the population is growing: it has become even more essential than it was in the past to use the existing infrastructures in the most efficient way as possible and to adapt them to the new conditions in order to avoid them to fail in producing the level of benefit that provided the economic justification for their development. Many large storage projects worldwide have had their operations designed in prior decades, assuming normal hydroclimatic conditions and considering a restricted number of operating objectives, with operating rules often conditioned on very simple information systems; many of them are still operated typically basing on rudimentary statistical analysis, simple considerations and the operator’s experience. Over the years, an increasing number of accurate tools and policies have been provided by Environmental Engineering in order to deal with the problem of managing operations of water reservoirs. Even though these approaches has been proved to be effective, they suffer of some issues mostly related with the model identification of the natural system and the computational complexity. This work aim is to compare what has been already obtained with usual design policies, proper of Environmental Engineering, in the planning and management of water systems resources, against a data-driven control for a multipurpose water reservoir coupled with a predictive controller through the analysis of the Vietnamese Hoa Binh water reservoir and to determine if the results show a good approximation of what is currently being obtained. The first part aims to define an internal controller with the Virtual Reference Feedback Tuning (VRFT), which defines a closed-loop model capable of following an optimal trajectory of the water storage in the dam. The second part aims to set up an external, outer-loop, Linear Quadratic Regulator (LQR) to enhance the performance of the inner-loop. The goodness of the approach will be evaluated with respect to hydropower production index and flooding protection risk index. The whole work has been carried out entirely in a Matlab environment.
In un contesto in rapida evoluzione, gestire le risorse naturali è diventato un problema crescente nel corso degli anni. Il mondo sta affrontando un enorme cambiamento idroclimatico e socioeconomico e la popolazione sta crescendo: rispetto al passato, è diventato ancora più importante utilizzare le infrastrutture esistenti nel modo più efficiente e adattarle alle nuove condizioni per garantire che producano il livello di benefici che ne ha costituito la giustificazione economica. Molti serbatioi, nel mondo, funzionano ancora sulla base di progetti pensati in decadi precedenti, assumendo normali condizioni idroclimatiche e considerando un ristretto numero di obiettivi, con regole spesso condizionate da sistemi di informazione basilari; molti di questi sono ancora gestiti, generalmente, sulla base di rudimentali analisi statistiche e sull’esperienza dell’operatore. Nel corso degli anni, l’Ingegneria Ambientale ha messo a punto un certo numero di strumenti accurati per affrontare il problema della gestione operativa dei serbatoi idrici. Anche se questi approcci si sono dimostrati efficaci, sono affetti da alcuni problemi legati all’identificazione del modello del sistema naturale e alla complessità computazionale. L’obiettivo di questo lavoro è comparare ciò che è stato ottenuto tramite i tradizionali metodi, propri dell’Ingegneria Ambientale, riguardo la gestione e il controllo dei sistemi idrici, rispetto a un controllo data-driven per un sistema multi-obiettivo in coppia con un predictive controller mediante l’analisi del serbatoio idrico vietnamita di Hoa Binh e determinare se i risultati costituiscono una buona approssimazione di quanto attualmente ottenuto. La prima parte mira a definire un controllore interno mediante la Virtual Reference Feedback Tuning (VRFT), la quale definisce un modello ad anello chiuso in grado di seguire una traiettoria ottimale della riserva d’acqua nella diga. La seconda è atta a regolare un Linear Quadratic Regulator (LQR), un anello esterno, per migliorare le prestazioni dell’anello interno. La bontà dell’approccio sarà valutata rispetto all’indice di produzione di energia idroelettrica e all’indice di rischio di protezione dalle inondazioni. L’intero lavoro è stato sviluppato in ambiente Matlab.
Exploring data-driven predictive control for optimal reservoir operation
VAGHI, FILIPPO ANDREA MARIA
2018/2019
Abstract
In a rapidly changing context, dealing and managing with natural resources has become an increasing problem over the years. The world is facing a huge hydroclimatic and socioeconomic change and the population is growing: it has become even more essential than it was in the past to use the existing infrastructures in the most efficient way as possible and to adapt them to the new conditions in order to avoid them to fail in producing the level of benefit that provided the economic justification for their development. Many large storage projects worldwide have had their operations designed in prior decades, assuming normal hydroclimatic conditions and considering a restricted number of operating objectives, with operating rules often conditioned on very simple information systems; many of them are still operated typically basing on rudimentary statistical analysis, simple considerations and the operator’s experience. Over the years, an increasing number of accurate tools and policies have been provided by Environmental Engineering in order to deal with the problem of managing operations of water reservoirs. Even though these approaches has been proved to be effective, they suffer of some issues mostly related with the model identification of the natural system and the computational complexity. This work aim is to compare what has been already obtained with usual design policies, proper of Environmental Engineering, in the planning and management of water systems resources, against a data-driven control for a multipurpose water reservoir coupled with a predictive controller through the analysis of the Vietnamese Hoa Binh water reservoir and to determine if the results show a good approximation of what is currently being obtained. The first part aims to define an internal controller with the Virtual Reference Feedback Tuning (VRFT), which defines a closed-loop model capable of following an optimal trajectory of the water storage in the dam. The second part aims to set up an external, outer-loop, Linear Quadratic Regulator (LQR) to enhance the performance of the inner-loop. The goodness of the approach will be evaluated with respect to hydropower production index and flooding protection risk index. The whole work has been carried out entirely in a Matlab environment.File | Dimensione | Formato | |
---|---|---|---|
2019_07_Vaghi.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Thesis text
Dimensione
3.67 MB
Formato
Adobe PDF
|
3.67 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/10589/148580