Optimizing design parameters with genetic algorithms, sensitivity analysis with data-driven models
The basic task is to enhance the currently available techniques for initial design process parameter optimization specified by customers (in terms of size, maximal allowed power output etc.), in order to obtain an optimal cost-performance ratio of the electrical drives.
One partial goal is a statistical, data-driven oriented study about the impact of certain engine setup parameters onto important target measures (sensitivity analysis) such as power loss or degree of efficiency. Another goal is to improve the currently implemented genetic algorithms for design parameter optimization with respect to computation speed and performance (fitness) and to extrapolate to optimal cost-performance ratio in the design parameter space based on nearby solutions found by the genetic algorithm.
Topics: genetic, evolutionary and memetic algorithms, data-driven system identification and modeling, sensitivity analysis, interpolation