This master thesis presents the integration of various cutting edge methods which have allowed an in depth analysis and an accurate modelling of the switching losses of a Two Level Three Phase VSI and a subsequent study of the tuning of three different control techniques in order to achieve desiderable performance of the system in terms of efficiency (h) and THD. In the first part, it is derived a set of equations from the integration of the product of the linear approximation of the typical switching characteristics. This model, which only needs a bunch of parameters and works using pre and post switching voltages and currents, has lead to use in simulation ideal components (IGBTs used as switches and fast recovery diodes set in anti-parallel configuration) and reasonable time step sizes. Furthermore this estimation has provided the possibility of a precise evaluation of the efficiency of the system useful to validate in simulation the controllers implemented. The second part deals with the description of the three different control schemes: hysteresis current controller (HCC), Two phase modulation method based on space vector theory (SVM) and Model Predictive Feedback Modulation (MPFM). For each of them, it has been highlighted strengths and weaknesses and a particular clue was also put on the main differences between them. The HCC has been used as an example of a simple but limited solution unable, in the topology presented, to consider different objective from the tracking performance. The SVM method derives from a careful pre-analysis of the theory regarding the limitation of the switching losses in terms of reduction of switching times (Two phase modulation method which identifies optimal switching patterns) and switching amplitude (inserting a term representing the switching losses in the cost function). The MPFM method joins the Model Predictive Control (MPC) approach with a relatively simple cost function (which includes a term that weights the switching losses) to the Feedback Quantizer Modulation (FBQM) which leads to reduce the prediction horizon N while maintaining satisfactory performances. The dissertation ends with the presentation of the simulation results (obtained using MATLAB and simulink) used to set up the aforementioned empirical tuning of the regulators and a comparative analysis which also considers the self time of execution evaluated through the Profiler. What can be concluded is that, given the application considered and under the assumptions made, the scheme which is the most promising as for what regards h and the THD is MPFM. However, at the same time, it is the more difficult to tune and the one which is characterized by the higher computational burden.
Questa tesi magistrale presenta l’integrazione di diverse tecniche all’avanguardia che hanno permesso un’analisi dettagliata e un’accurata modellazione delle perdite di commutazione di un inverter trifase a due livelli a tensione impressa e il successivo studio della taratura di tre diversi controllori con l’obiettivo di ottenere soddisfacenti prestazioni del sistema sia in termini di efficienza che di THD. La prima parte riguarda la derivazione di un sistema di equazioni dall’integrazione del prodotto delle approssimazioni lineari delle tipiche caratteristiche di commutazione. Questo modello, che richiede solo la conoscenza di gruppo di parametri e utilizza i valori di correnti e tensioni prima e dopo la commutazione, ha consentito di usare in simulazione componenti ideali (IGBT come interruttori e diodi dal recupero veloce messi in configurazione antiparallela) e dimensioni ragionevoli del periodo di campionamento. Inoltre, questa stima ha dato la possibilità di una precisa valutazione dell’efficienza del sistema che è servita a validare in simulazione i controllori implementati. La seconda parte tratta la descrizione dei tre diversi schemi di controllo: controllore di corrente ad isteresi (CCI), tecnica di modulazione ‘a due fasi’ basata sulla teoria dello ‘Space Vector’ (SVM), tecnica ‘Model Predictive Feedback Modulation’ (MPFM). Per ciascuno di questi sono stati evidenziati i punti di forza e di debolezza e sono state anche mostrate le maggiori differenze tra essi. Il CCI, nella tipologia considerata, è stato usato come esempio di una soluzione semplice ma limitata nel considerare obiettivi diversi dall’inseguimento del riferimento. Il metodo SVM deriva da un’attenta pre-analisi della teoria volta alla limitazione delle perdite di commutazione grazie alla riduzione del numero di transizioni (tecnica di modulazione ‘a due fasi’ che identifica delle successioni di vettori ottimali) e del relativo modulo (inserendo un termine che rappresenta le perdite di transizione nella funzione di costo). Il metodo MPFM unisce l’approccio MPC con una funzione di costo relativamente semplice (che include un termine che pesa le perdite di transizione) alla tecnica ‘Feedback Quantizer Modulation’ (FBQM) che permette di ridurre l’orizzonte di predizione N mantenendo delle prestazioni soddisfacenti. La trattazione termina con la presentazione dei risultati di simulazione (ottenuti usando MATLAB e simulink) usati per impostare la taratura empirica dei regolatori sopra citata e un’analisi comparativa che considera anche il ‘self time’ di esecuzione (stimato tramite il ‘Profiler’). La conclusione che si può trarre è che, data l’applicazione considerata e le assunzioni fatte, MPFM è lo schema più promettente in termini di efficienza e THD. Tuttavia, allo stesso tempo, quest’ultimo è il più difficile da tarare e quello caratterizzato dal più alto peso computazionale.
Switching losses of a two-level three-phase VSI : modelling, analysis and control
BESANA, DAVIDE
2017/2018
Abstract
This master thesis presents the integration of various cutting edge methods which have allowed an in depth analysis and an accurate modelling of the switching losses of a Two Level Three Phase VSI and a subsequent study of the tuning of three different control techniques in order to achieve desiderable performance of the system in terms of efficiency (h) and THD. In the first part, it is derived a set of equations from the integration of the product of the linear approximation of the typical switching characteristics. This model, which only needs a bunch of parameters and works using pre and post switching voltages and currents, has lead to use in simulation ideal components (IGBTs used as switches and fast recovery diodes set in anti-parallel configuration) and reasonable time step sizes. Furthermore this estimation has provided the possibility of a precise evaluation of the efficiency of the system useful to validate in simulation the controllers implemented. The second part deals with the description of the three different control schemes: hysteresis current controller (HCC), Two phase modulation method based on space vector theory (SVM) and Model Predictive Feedback Modulation (MPFM). For each of them, it has been highlighted strengths and weaknesses and a particular clue was also put on the main differences between them. The HCC has been used as an example of a simple but limited solution unable, in the topology presented, to consider different objective from the tracking performance. The SVM method derives from a careful pre-analysis of the theory regarding the limitation of the switching losses in terms of reduction of switching times (Two phase modulation method which identifies optimal switching patterns) and switching amplitude (inserting a term representing the switching losses in the cost function). The MPFM method joins the Model Predictive Control (MPC) approach with a relatively simple cost function (which includes a term that weights the switching losses) to the Feedback Quantizer Modulation (FBQM) which leads to reduce the prediction horizon N while maintaining satisfactory performances. The dissertation ends with the presentation of the simulation results (obtained using MATLAB and simulink) used to set up the aforementioned empirical tuning of the regulators and a comparative analysis which also considers the self time of execution evaluated through the Profiler. What can be concluded is that, given the application considered and under the assumptions made, the scheme which is the most promising as for what regards h and the THD is MPFM. However, at the same time, it is the more difficult to tune and the one which is characterized by the higher computational burden.File | Dimensione | Formato | |
---|---|---|---|
2018_12_Besana.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Testo della tesi
Dimensione
24.77 MB
Formato
Adobe PDF
|
24.77 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/10589/144891