Hide/Show Apps

Development of a hybrid global optimization algorithm and its application to helicopter rotor structural optimization

Bilen, Muhammed Emre
Helicopters are notorious for their high vibration levels and the rotor system are the main contributors to the problem. The rotor vibrations can be minimized by optimizing the rotor structure, which require time-consuming high-fidelity solution for vibration predictions. To solve this problem, an effective and efficient global search algorithm called Explorer-Settler Optimization algorithm is developed by combining the advantageous aspects of Particle Swarm Optimization and Nelder-Mead Optimization algorithms. It is shown that the developed algorithm performs superior when compared to other popular search algorithms in terms of search space exploration with minimum number of objective function calls. Moreover, to reduce the rotor optimization times, surrogate-based models are implemented for rotor blade structural property predictions. Data transformation techniques are evaluated and applied to minimize prediction errors. For the reduction of rotor vibrations, two main approaches are implemented namely, frequency separation approach (FSA) and direct vibration reduction approach (DVRA). Along with vibration reduction, both approaches aim to minimize rotor blade mass while satisfying various dynamic and static constraints. However, FSA attempts to achieve minimum vibrations by separating the natural frequencies of the rotor blades from excitation frequencies to avoid resonances; whereas, DVRA directly targets vibration amplitudes. A four-bladed rotor is optimized using both approaches and the performances of them are compared against each other. The results indicate that DVRA performs better in vibration reduction while FSA provides lighter rotor blades.