Optimisation of open-loop control of convective heat transfer with genetic algorithms
In session: TUE 6.2 - Machine Learning III
Please login to view abstract download link
An artificial intelligence approach based on linear genetic algorithm control is utilised to enhance the convective heat transfer in a turbulent boundary layer on a flat plate. The actuator consists of six fully-modulated slot jets in crossflow, which are aligned with the freestream. The cost function includes wall convective heat transfer rate and energy consumption of the actuation. Performance evaluation is conducted using infrared thermography and the interpretation of the control effects is carried out with particle image velocimetry measurements. The algorithm converges to the same frequency and duty cycle for all actuators, with the phase difference between multiple jet actuation driving flow asymmetry. The study highlights the potential of machine learning control and advanced measurement techniques in experimental investigations.