Multi-Working Point Performance Optimization of Variable Cycle Engine Based on Ensemble Radial Basis Function Neural Network

This paper proposed a multi-working point performance optimization algorithm based on the radial basis function neural network (RBFNN) for a variable cycle engine. The ensemble RBFNN is used to replace the engine performance simulation model during the optimization process to improve the optimization efficiency. Also, an enhanced multi-objective differential evolution algorithm is proposed to improve the Pareto solutions quality and convergence rate. Finally, a sample for three working points performance optimization of a double-bypass variable cycle engine is presented. This work optimizes a series of variable cycle engine design parameters and adjusting schedules, which consider high takeoff thrust and low cruise fuel consumption.