Abstract: In order to achieve decoupling control of aero-engine over whole envelop, genetic algorithm was used to optimize the proportion parameters of single neural adaptive decoupling controller based on various points within envelop, whilst several groups of parameters optimized by the algorithm were used as sample data to train radial basis function(RBF)network, helping to map height and mach number on proportion parameters after tuning.The proportion parameters at every operating point can be calculated by RBF neural network.The simulation results show that the control system has a fine performance of dynamic and decoupling capability over whole envelop, and the proposed method is simple, feasible and effective.