Predictive Deadbeat Control For PMSM Drive

Moos, Jesper

4. term

Energy Engineering, Master

2014

2014-06-03

104

Denne afhandling undersøger, hvordan man styrer strømmen i en Permanent Magnet Synkronmaskine (PMSM), en type elmotor med permanente magneter. Den designer, analyserer og implementerer en forudsigende deadbeat-strømregulator, som beregner det næste styresignal for hurtigt at bringe strømmen til sin reference. Som sammenligningsgrundlag udvikles også en klassisk Field Oriented Control (FOC) med en proportional-integral (PI) strømregulator. Afhandlingen gennemgår de grundlæggende principper for drivsystemet og opstiller matematiske modeller af inverteren og PMSM’en. Der udvikles to simuleringsmodeller: én med PI-strømregulatoren og én med deadbeat-regulatoren. Fordi inverterens dødtid og DSP-forsinkelse påvirker deadbeat-regulatorens ydelse, udvikles og indarbejdes kompensering for disse effekter. På baggrund af simulationsresultaterne implementeres begge regulatorer i et laboratorie-setup (dSPACE 1103) og afprøves. Regulatorernes følsomhed over for parametre testes både i simulationer og i forsøg. Derudover viser afhandlingen, hvordan motorens induktans kan estimeres ved hjælp af inverterens nulspændingsvektor, og hvordan kernetab kan reduceres ved at svække magnetfeltets fluxdensitet (felt­svækkelse). Resultaterne viser, at den forudsigende deadbeat-regulator er enkel og intuitiv og leverer fremragende dynamisk ydelse på niveau med den klassiske FOC-PI. Den har dog små stationære fejl, sandsynligvis på grund af afhængighed af modelparametre og inverterens ikke-lineariteter, som skal kompenseres korrekt ved forskellige arbejdsbetingelser.

This thesis examines how to control current in a Permanent Magnet Synchronous Machine (PMSM), a type of electric motor that uses permanent magnets. It designs, analyzes, and implements a predictive deadbeat current controller, which calculates the next control action to drive the current quickly to its reference. For comparison, a classical Field-Oriented Control (FOC) scheme with a proportional–integral (PI) current controller is also developed. The thesis presents the basic principles of the drive system and builds mathematical models of the inverter and the PMSM. Two simulation models are created: one with the PI current controller and one with the deadbeat controller. Because inverter dead-time and DSP delay influence the deadbeat controller’s performance, compensation for these effects is designed and integrated. Based on simulation results, both controllers are implemented on a laboratory setup (dSPACE 1103) and tested. Their sensitivity to parameter variations is evaluated in both simulations and experiments. In addition, the thesis shows how the machine’s inductance can be estimated using the inverter’s zero-voltage vector, and discusses how core losses can be reduced by weakening the magnetic flux density (field weakening). The results indicate that the predictive deadbeat controller is simple and intuitive and delivers excellent dynamic performance comparable to the classical FOC-PI. However, it exhibits small steady-state errors, likely due to dependence on model parameters and inverter nonlinearities, which must be properly compensated at different operating points.

[This abstract was generated with the help of AI]

Deadbeat ; PMSM ; Predictive ; PI ; Controller ; dSPACE ; FOC ; Control ; Current

Download
View record in AAU Student Projects