زهرا حيدري هفشجاني

The use of advanced controllers can significantly enhance the performance of wind turbines. Improving turbine performance involves extending the operational lifespan of electrical and mechanical components, as well as increasing the turbine's output power. In variable-speed wind turbines, maximum power extraction can be achieved by adapting to turbine speed variations in response to changes in wind speed. Furthermore, implementing advanced control methods in variable-speed wind turbines can enhance power output and minimize energy losses. Another notable advantage of these turbines is their high flexibility in managing variations in wind speed and direction. The primary objective of wind energy systems is to extract the maximum possible power from the wind within the shortest time frame. The Quantum Deep Reinforcement Learning (QDRL) algorithm, which leverages the advantages of quantum computing and reinforcement learning, offers an efficient online controller by reducing the runtime of iterative optimizations. This approach combines the ability to update reinforcement learning algorithms online with the global optimization capabilities of quantum processes, ensuring stable and optimal performance even under varying conditions. In this thesis, the QDRL algorithm is employed to calculate controller parameters online. The proposed controller is applied to a two-mass model of a wind turbine, and its simulation results are compared with those obtained using PID and disturbance-based optimal controllers in the MATLAB Simulink environment. The performance and efficiency of the proposed method are thoroughly analyzed and eva‎luated.

يادگيري تقويتي عميق كوانتومي , توربين بادي , رديابي نقطه بيشينه توان

Quantum deep reinforcement learning , wind turbine , Maximum power point tracking

Zahra Heidari hafshejani

Soheil Ganjefar