Adaptive Filter for Harmonic Suppression for Sensorless PMSM Drives
Authors
Clausen, Kasper Juul ; Christiansen, Emil
Term
4. term
Education
Publication year
2023
Abstract
Dette speciale adresserer et praktisk problem i sensorløs styring af permanente magnet synkronmotorer (PMSM): inverterens ikke-lineariteter og rumlige fluxharmoniske forvrænger den estimerede back-EMF med (6±1)-ordens harmonikker, hvilket giver 6.-ordens rippler i den estimerede rotorposition og -hastighed og dermed kan forringe reguleringspræstationen. For at undertrykke disse harmonikker præsenteres to metoder, som integreres i en sensorløs feltorienteret styringsstruktur med en back-EMF-baseret estimator bestående af en lukket kredsløb fluxobservatør (CLFO) og en fase-låst sløjfe (PLL): et komplekskoefficient synkronfrekvensfilter (CCSFF) og et bilineært rekursivt mindste kvadraters (BRLS) adaptivt filter. Parametreringen af filtrene analyseres, og metoderne verificeres eksperimentelt på en laboratorieopstilling. Resultaterne viser, at CCSFF opnår tilstrækkelig 6.-ordens harmonikundertrykkelse i forskellige stationære tilstande, hvor suppressionsevnen forbedres ved højere hastigheder; båndbredden er afgørende og indebærer en afvejning, idet lavere båndbredde øger filtreringen men forringer den dynamiske ydeevne med større faseskift ved hastigheds- og lastændringer, hvilket kan afhjælpes med en adaptiv CCSFF-PLL-båndbredde uden at svække filtreringen. BRLS-filteret giver endnu bedre harmonikundertrykkelse over et bredt hastighedsområde ved at spore og kompensere de specificerede harmoniske i den estimerede back-EMF, med nul faseskift i stationær tilstand og uden at påvirke systemets dynamiske ydeevne. Forbedret positions- og hastighedsestimation reducerer desuden (6±1)-ordens forvrængning i fase-strømmene ved lave til middel hastigheder, men effekten er begrænset af drifthastigheden og strømsløjfens kapacitet, så yderligere metoder kan være nødvendige for yderligere at forbedre strømforvrængningen.
This thesis addresses a practical challenge in sensorless control of permanent magnet synchronous motors (PMSMs): nonlinearities in the voltage source inverter and flux spatial harmonics distort the estimated back-EMF with (6±1)th-order components, producing 6th-order ripples in the estimated rotor position and speed and potentially degrading control performance. To suppress these harmonics, two methods are proposed and integrated into a sensorless field-oriented control structure with a back-EMF-based estimator comprising a closed-loop flux observer (CLFO) and a phase-locked loop (PLL): a complex coefficient synchronous frequency filter (CCSFF) and a bilinear recursive least squares (BRLS) adaptive filter. Filter parameterization is analyzed, and both approaches are experimentally verified on a laboratory setup. Results show that the CCSFF achieves sufficient 6th-order harmonic suppression under various steady-state conditions, with suppression improving as speed increases; its bandwidth governs a trade-off where lower bandwidth enhances filtering but worsens dynamic performance via added phase shift during speed and load changes, which can be mitigated by an adaptive CCSFF-PLL bandwidth without sacrificing filtering capability. The BRLS adaptive filter delivers even stronger suppression across a wide speed range by tracking and compensating specified harmonic components in the estimated back-EMF, with zero steady-state phase shift and no impact on dynamic performance. The improved position and speed estimation also attenuates (6±1)th-order distortion in the phase currents at low to mid speeds, though the benefit is limited by operating speed and current-loop capabilities, indicating that additional methods may be needed for further current distortion reduction.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents