Numerical optimization of a highly loaded compressor in semi-closed cycles using neural networks and genetic algorithms

摘要

This paper outlines the aerodynamic optimization for stator vane settings of multi-stage compressors in a conceptual semi-closed cycle with the combination of an artificial neural network (ANN) and genetic algorithm (GA). The investigation is conducted on a newly developed 5-stage highly loaded axial flow compressor. A 3-layer perceptron neural network is employed as the surrogate model replacing an in-house one-dimensional blade-stacking computation code, and the influences of changes in physical properties of the working medium with varying ratios of exhaust CO2 recirculation are considered in the computation. The stagger angles of the four stator vanes serve as the input data of the ANN, and the compressor aerodynamic performances are the outputs of the network. The well-trained ANN is then incorporated into the optimization framework which is based on an improved real-coded GA. Some advanced strategies including the elitism operator, blend crossover, non-uniform mutation, and self-adaption parameters are introduced into the GA to promote the searching efficiency and solution globality. A series of numerical optimization is carried out at various CO2 contents under part-speed conditions to achieve the maximum adiabatic efficiency with restrictions on the pressure ratio. The results show that the optimized stator vane settings can improve the adiabatic efficiency by about 1% for most cases, and a considerable reduction of the flow losses near the endwall regions is observed for the reference operating points. Regardless of the assumption of quasi-one-dimensional flow, the effectiveness of the optimization framework in dealing with the stage-mismatching has been demonstrated. This research has allowed to reveal that from the compressor optimization point of view, a semi-closed cycle is feasible using existing technology and that compressor modifications are needed according to situational requirements. (c) 2015 Society of Chemical Industry and John Wiley & Sons, Ltd