Thesis FINITE CONTROL SET MODEL PREDICTIVE CONTROL FOR RENEWABLE ENERGY CONVERSION SYSTEMS
Loading...
Date
2018
Authors
AGUIRRE MARTÍNEZ, MATÍAS ALEJANDRO
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Control predictivo (MPC por sus siglas en inglés), y en particular MPC de Estadosde Control Finitos (FCS-MPC), ha demostrado características interesantes que la hacenadecuada para múltiples aplicaciones en el ámbito de conversión de potencia. Entre estascaracterísticas están la alta velocidad de control, fácil implementación, facilidad paraincluir objetivos de control no triviales, y la selección automática de las señales de actuación, entre otros. Estas características hacen de FCS-MPC una estrategia de controlatractiva para sistemas de generación de energías renovables, donde la alta velocidad decontrol sobre múltiple objetivos permite una integración mas flexible con la red.Una de las principales desventajas de FCS-MPC hasta ahora es la alta variabilidad dela frecuencia de conmutación en los semiconductores. Esta variabilidad lleva a múltiplesproblemas, tales como altas perdidas por conmutación, estimulación de resonancias indeseadasdebido al amplio espectro de voltaje, y distribución desigual entre los semiconductoresde un dispositivo de conversión. Estos problemas son especialmente relevantesen sistemas de media o alta tensión, donde una baja eficiencia se traduce en perdidas importantes,resultando en altas temperaturas y una rápida degradación de los componenteselectrónicos. En este trabajo dos estrategias son analizadas para abordar estos problemas.La primera propuesta consiste en resolver el problema de frecuencia de conmutaciónvariable. Esto se logra mediante un cambio en los objetivos de control, no abordando elproblema desde la frecuencia, si no a través del periodo de conmutación, siendo bautizadacomo “Period Control”. Para esto, una nueva medición es diseñada, lo que permitea FCS-MPC monitorear y controlar el periodo de conmutación, ademas de cambios en laestructura de control para permitir que esta medición sea efectiva. Algunos elementos dediseño para el factor de peso y la referencia de periodo son estudiado con el fin de simplificarsu implementación. Simulaciones y resultados experimentales son presentados paravalidar la propuesta, además de comparaciones con alternativas previamente propuestaspor la literatura.La segunda propuesta apunta a disponer de esta frecuencia variables con el objeto demejorar el desempeño de control en términos de las perdidas de potencia. Esto se logramediante el uso de una penalización de la conmutación con un peso variable, fortaleciendo este peso para altas corrientes, y relajándolo para corrientes bajas. El desempeño es evaluadomediante una comparación directa entre las perdidas causadas por la conmutación y ladistorción armónica provocada por la reducción de las conmutaciones. Múltiples estrategiasson evaluadas mediante simulación para este peso variables, y son comparadas entreellas y contra estrategias clásicas de reducción de conmutaciones.
Model predictive control (MPC), and in particular Finite Control Set MPC (FCSMPC),has shown many interesting characteristics that makes it suitable for multiple applicationof power conversion. Among these favorable traits are its high control speed response,easy implementation, the ability to include multiple non trivial control objectivesinto the strategy and the automatic selection of the output signals, among others. Thischaracteristics makes FCS-MPC a suitable control strategy for renewable energy conversionsystems, where a fast control response on multiple objectives permits a more flexibleintegration to the mains.One of the main drawbacks of FCS-MPC strategy so far is the high variability ofthe switching frequency of the semiconductors. This variability leads to many problems,such as switching losses due to excessive amount of commutations, undesired resonancestimulation due to the wide voltage spectrum and uneven power distribution among thesemiconductor devices. These problems are specially relevant in medium and high powergeneration systems, where low efficiency translates in major power losses, leading to hightemperatures and fast degradation of the electronic components. Two main approaches arestudied in this work to address these issues.The first approach consists on solving the issue of variable switching frequency. This isdone through a change in the control objectives by not addressing the switching frequency,but the switching period, thus being named “Period Control”. A new measurement is designedto allow FCS-MPC to monitor and control the switching period, as well as somechanges in the control structure to allow this measurement to be effective. Some designelements for the weight factor and the period reference are studied to simplify its implementation.Simulation and experimental results are presented to validate the proposedapproach, as well as a comparison against some alternatives available in the literature.The second approach aims to embrace the frequency variability in order to improvethe control performance in terms of switching power losses. This is achieved through theuse of a switching penalization with variable weight factor, strengthening the this factorfor high currents, and relaxing it for low currents. The performance is evaluated through adirect comparison between the switching power losses and the harmonic distortion induced by the reduced switching frequency. Multiple functions are evaluated, through simulation,for this variable weight, and are compared against each other and against classic switchingloss reduction strategies.
Model predictive control (MPC), and in particular Finite Control Set MPC (FCSMPC),has shown many interesting characteristics that makes it suitable for multiple applicationof power conversion. Among these favorable traits are its high control speed response,easy implementation, the ability to include multiple non trivial control objectivesinto the strategy and the automatic selection of the output signals, among others. Thischaracteristics makes FCS-MPC a suitable control strategy for renewable energy conversionsystems, where a fast control response on multiple objectives permits a more flexibleintegration to the mains.One of the main drawbacks of FCS-MPC strategy so far is the high variability ofthe switching frequency of the semiconductors. This variability leads to many problems,such as switching losses due to excessive amount of commutations, undesired resonancestimulation due to the wide voltage spectrum and uneven power distribution among thesemiconductor devices. These problems are specially relevant in medium and high powergeneration systems, where low efficiency translates in major power losses, leading to hightemperatures and fast degradation of the electronic components. Two main approaches arestudied in this work to address these issues.The first approach consists on solving the issue of variable switching frequency. This isdone through a change in the control objectives by not addressing the switching frequency,but the switching period, thus being named “Period Control”. A new measurement is designedto allow FCS-MPC to monitor and control the switching period, as well as somechanges in the control structure to allow this measurement to be effective. Some designelements for the weight factor and the period reference are studied to simplify its implementation.Simulation and experimental results are presented to validate the proposedapproach, as well as a comparison against some alternatives available in the literature.The second approach aims to embrace the frequency variability in order to improvethe control performance in terms of switching power losses. This is achieved through theuse of a switching penalization with variable weight factor, strengthening the this factorfor high currents, and relaxing it for low currents. The performance is evaluated through adirect comparison between the switching power losses and the harmonic distortion induced by the reduced switching frequency. Multiple functions are evaluated, through simulation,for this variable weight, and are compared against each other and against classic switchingloss reduction strategies.
Description
Catalogado desde la version PDF de la tesis.
Keywords
CONTROL DE FRECUENCIA , CONTROL PREDICTIVO , ENERGIAS RENOVABLES