Darshan T. Email & Phone Number

Garching, BY, DE

Garching, Bavaria, Germany

Institut Für Strömungsmechanik, TU Braunschweig

• ○ Deep Reinforcement Learning
• Developing Python Library to provide easy-to-use environments of flow control-based reinforcement learning tasks and algorithms.
• Implementing PPO algorithm for “Fluidic pinball” flow environment to reduce the aerodynamic forces on the overall structure, where the action space is multidimensional.
• Optimizing the number of sensors and the position of sensors for state observations using machine learning techniques to achieve flow control.

Brunswick, Lower Saxony, Germany

• ○ Robust Optimization of Wind Turbine Blade Profiles under Several Sources of Uncertainty
• Optimized airfoil geometry for wind turbine blade, resulting in improved lift and reduced drag under uncertainty in wind speed and wind direction measured at the FINO-2 station.
• Applied efficient bi-level surrogate approach for robust optimization with outer level corresponding to surrogate-based optimization and inner level corresponding to surrogate-based uncertainty quantification.
• Implemented Kriging meta-models for both levels of the surrogate-based approach.
• Evaluated the results for deterministic and robust optimization under uncertainties.

Brunswick, Lower Saxony, Germany

• ○ Binary Image Segmentation of Infrared Images
• Performed binary image segmentation of infrared images of flow separation taken by Airbus flight tests.
• Trained and implemented U-Net architecture (NN) with data augmentation.
• Trained and implemented Mobile-Net architecture (NN) with encoder and decoder.

Institut Für Strömungsmechanik, TU Braunschweig

• ○ Active Flow Control by Deep Reinforcement Learning
• Developed an agent for active flow control by deep reinforcement learning to reduce the aerodynamic forces.
• Implemented PPO algorithm for closed-loop control for fluid flow around a cylinder.
• Developed a Gaussian process regression-based model for open-loop control.
• Novel sensor placement to meet real-life application goals such as flow control over an aircraft wing.
• Analyzed the performance of the neural networks over open-loop flow control.

Institut Für Verkehr Und Stadtbauwesen, TU Braunschweig

• ○ GUI for Tracking and Distinguishing Pedestrians and Cyclists from Traffic
• Improved ergonomics of GUI by implementing the functionalities of a conventional video player.
• Implemented functionality for manual-labeling of data from the graphics to be used for retraining.

Gujarat, India

• ○ Modification in Variable Valve System in IC engine
• Design and Analysis of the new design of variable valve system in IC engine to improve the engine efficiency for multiple speed ranges.
• The modification is related to the cam and camshaft of the engine to change the valve opening and closing timings at an optimum degree.

Master Of Science - Msc, Computational Sciences In Engineering

Bachelor Of Engineering - Be, Mechanical Engineering