Reducing Wind Turbine Fatigue under Uncertainty : A Zone-Controlled SMPC Approach

Ahmed, Faiyaz Alvi

4. semester

Electronic Systems, Master

2024

2024-05-30

92

Vindmøller arbejder under skiftende og turbulente vindforhold, som skaber usikkerhed i driften. Denne afhandling undersøger tre reguleringsmetoder—Proportional-Integral (PI), deterministisk Model Predictive Control (MPC) og zonekontrolleret stokastisk Model Predictive Control (ZC-SMPC)—for at styre ydeevnen under sådanne forhold. Målet er at designe og evaluere en ZC-SMPC-løsning, der mindsker træthed (slid fra gentagne belastninger) og bevarer driftsstabilitet. Først opbygges en digital tvilling af en vindmølle i OpenFAST-simuleringsmiljøet, hvorefter den deterministiske MPC implementeres og indstilles. ZC-SMPC tilføjer zonekontrol: i stedet for at justere pitch (bladstilling) hele tiden fastlægges en kontrolzone, hvor afvigelser i rotorhastighed får en straf i reguleringen. Det fremmer færre unødige pitch-bevægelser og reducerer bladdræthed uden at forringe andre ydeevnemål. Metoderne sammenlignes med Pareto-frontanalyse for at afveje modstridende mål, især mellem tårn- og bladtræthed. Resultaterne viser, at ZC-SMPC med en straf på rotorhastighed inden for kontrolzonen opnår den bedste balance mellem lastreduktion og stabilitet og giver den laveste blad-DEL (Damage Equivalent Load). PI-regulatoren er enkel, men mindre effektiv på grund af høj variabilitet og overskridelser af driftsgrænser. Den deterministiske MPC er mere tilpasningsdygtig end PI, men når ikke ZC-SMPC’s samlede ydeevne. Samlet set demonstrerer ZC-SMPC et betydeligt potentiale for at optimere reguleringen af vindmøller ved at håndtere afvejninger mellem forskellige træthedsbelastninger og fastholde stabil rotorhastighed.

Wind turbines operate in changing, turbulent winds that introduce uncertainty. This thesis tests three control methods—Proportional-Integral (PI), deterministic Model Predictive Control (MPC), and Zone-Controlled Stochastic Model Predictive Control (ZC-SMPC)—to manage performance under such conditions. The aim is to design and evaluate a ZC-SMPC tailored to wind turbines to reduce fatigue (wear from repeated loads) while maintaining operational stability. A digital twin of a turbine is built in the OpenFAST simulation environment, followed by implementation and tuning of deterministic MPC. ZC-SMPC adds zone control: rather than constantly adjusting blade pitch, it defines a control zone where rotor speed deviations are penalized, encouraging fewer unnecessary pitch changes to reduce blade fatigue without compromising other metrics. The controllers are compared using Pareto frontier analysis to reveal trade-offs between objectives, especially tower versus blade fatigue. Results show that ZC-SMPC, with a penalty on rotor speed within the control zone, achieves the best balance between load reduction and stability and yields the lowest blade Damage Equivalent Load (DEL). The PI controller is simple but less effective due to high variability and constraint violations. Deterministic MPC adapts better than PI but does not match ZC-SMPC’s overall performance. Overall, ZC-SMPC demonstrates strong potential to optimize wind turbine control by managing fatigue trade-offs and keeping rotor speed stable.

[This abstract was generated with the help of AI]

stochastic model predictive control ; SMPC ; MPC ; Fatigue ; PI ; wind turbine ; OpenFast ; Simulink ; Matlab ; unscented kalman filter ; ukf ; Pareto front

Download
View record in AAU Student Projects