Abstract: In this paper, the key techniques of automatic compressor blade optimization are investigated. A parallel genetic algorithm is adopted as the numerical optimization method for its global and parallel optimization ability. The blade profile parameterization is based on geometry modification to generate controllable and reasonable blade profiles in the process of optimization. To realize the optimization in the whole work range, an objective function is set which also taken into consideration the performance at non-design positions. A software is developed, which could design 2D blade profiles of plane and arbitrary rotating surfaces. The software is used to design high subsonic and supersonic blade profiles of plane and arbitrary rotating surfaces. The constant entropy Mach number distribution of the designed subsonic blade profiles is in good accord with controlled diffusion regularity; the flows are compressed nearly in constant entropy in the potential regions of designed supersonic cascades. Therefore, the designed blade profiles are of low total pressure loss coefficient and large low total pressure loss coefficient range.

Key words: compressor, fan, blade, numerical optimization, genetic algorithm