High-Dimensional Multiobjective Optimization of an Aeroengine Combustor Based on Cubic Polynomial

This paper innovatively designs a surrogate model for the performance of aeroengine high-temperature rising combustor based on cubic polynomials. According to the comparison of the one-dimensional calculation method, the three-dimensional computational fluid dynamics (CFD) numerical simulation, and the experiment, the feasibility of the one-dimensional (1D) calculation method and the reliability of the sample data calculated by this method are proved. The predictions of the metamodel were compared with those calculated by using the 1D program, and the root-mean square errors of the two target parameters were 0.000077% and 0.001725%. Based on this, sets of test data were used to verify the accuracy of using the surrogate model. The test results were compared with those of an artificial neural network (ANN) model and the 1D program. The root-mean square error (RMSE) of the cubic polynomial model and the ANN model in terms of combustion efficiency and total pressure loss were 0.0150% and 0.0107%, and 0.1628% and 0.3032%, respectively. A global sensitivity analysis identified the oil-to-gas ratio and the total inlet pressure as the most important factors affecting the combustion efficiency and total pressure loss. Particle swarm optimization (PSO) was used to obtain the optimal nondominant Pareto solution set. It provides physical insights for the optimization design of aeroengine combustor.