基于RBF网络的航空发动机单神经元解耦控制

针对航空发动机多变量控制系统中各回路之间存在的耦合现象,提出了一种基于RBF网络辨识的航空发动机多变量单神经元网络解耦控制方法。对发动机的多个控制回路,采用多个RBF网络实时辨识各个回路发动机的数学模型,并将系统的灵敏度信息实时反馈给各回路的控制器,保证了单神经元网络控制器对各回路的准确控制,最终实现对发动机多回路的解耦控制。通过在飞行包线内的仿真,结果表明,该方法不依赖被控对象的精确模型,有效地实现了对发动机的解耦控制,而且具有良好的动静态性能,将其应用于航空发动机多变量解耦控制是行之有效的。

Single neuron decoupling control based on RBF network for aeroengine

To solve the coupling problem in aeroengine multivariable control system,a method of single neuron decoupling control based on RBF network identification model was put forward.For multiple engine control loops,one loop engine model was real-timely identified by one RBF network.The sensitivity information was real-time feed back to the loop controller so that the loop single neuron controller can exactly control the engine.Finally,the decoupling control of aeroengine was realized.With simulation of some turbofan engine in the full flight envelope,the results show that the proposed method is independent on the aeroengine precise model and it can effectively reduce the coupling influence between loops.The controlled plant has good dynamic and static performances and the method applied to aeroengine is effective.