Predictive Controller for PMSM Drive
Authors
Quillo, Sonny Sune Mads C T ; Banos Garcia, Judit
Term
4. term
Education
Publication year
2013
Submitted on
2013-06-04
Pages
109
Abstract
Prædiktiv regulering er en familie af metoder, der vælger den næste styrehandling ved at bruge en model til at forudse, hvordan et system vil opføre sig. Med de seneste fremskridt i mikroprocessorer og digitale signalprocessorer undersøger denne afhandling prædiktiv strømstyring (PCC) for permanentmagnet-synkronmaskiner (PMSM), en udbredt og energieffektiv motortype. To PCC-tilgange—deadbeat-styring (som sigter mod at nå målet på ét måletrin) og modelprædiktiv styring (MPC, som optimerer en enkel model over en kort tidshorisont)—blev valgt for deres enkelhed og fleksibilitet. Begge blev i Matlab/Simulink sammenlignet med feltorienteret styring (FOC), den klassiske industristandard. Praktiske ikke-idealiteter som dødtid i inverteren og aktueringsforsinkelse blev modelleret og kompenseret i simulationerne. På tværs af testene gav deadbeat-styring den hurtigste dynamiske respons og lav strømrippel ved en samplingsfrekvens, der kan realiseres på den givne mikrocontroller. For at validere resultaterne blev deadbeat og FOC implementeret i laboratoriet på en dSPACE DS1103-platform. Forsøgene bekræftede, at deadbeat er et intuitivt alternativ til FOC ved styring af PMSM’er og giver forbedret momentrespons og overordnede dynamiske egenskaber.
Predictive control is a family of methods that chooses the next control action by using a model to anticipate how a system will behave. Leveraging recent advances in microprocessors and digital signal processors, this thesis studies predictive current control (PCC) for permanent magnet synchronous machines (PMSMs), a common high-efficiency motor type. Two PCC approaches—deadbeat control (which aims to reach the target in a single sampling step) and model predictive control (MPC, which optimizes a simple model over a short horizon)—were selected for their simplicity and flexibility. Both were compared in Matlab/Simulink against field-oriented control (FOC), the classical and widely used benchmark. Practical non-idealities such as inverter dead-time and actuation delay were modeled and compensated in the simulations. Across tests, deadbeat control delivered the fastest dynamic response and low current ripple at a sampling rate feasible on the available microcontroller. To validate the findings, deadbeat and FOC were implemented in the laboratory on a dSPACE DS1103 platform. Experiments confirmed that deadbeat is an intuitive alternative to FOC for PMSM control, offering improved torque response and overall dynamic performance.
[This abstract was generated with the help of AI]
Documents