To reduce flow loss and simplify the structure of the inlet particle separator(IPS)

hybrid optimization algorithm was introduced into the IPS optimization design.Firstly

four order spline curve was adopted as parameterized description of the IPS model and an automated simulation process was established.Optimal Latin hypercube design(OLHD) method was used to obtain the sample points

and an elliptical radial basis function(EBF) surrogate model was established to substitute the simulation code.Then non-dominated sorting genetic algorithm (NSGA-Ⅱ) was used for multi-objective optimization based on surrogate model

and finally non-linear programming by quadratic Lagrangian (NLPQL) algorithm was applied to local optimization.As a result

a new configuration was obtained

with the scavenge efficiency of C-spec distribution improved to 100% and 86.7% for AC coarse distribution fine sand and total pressure reduced to less than 0.6% with smaller main outlet pressure distortion.