Aeroengine PID Multi-variable Decoupling Control System Based on Dynamic NNI

Contrast to conventional PID multi-variable decoupling control, this paper presented a new PID decoupling method based on dynamic neural network identifier (NNI) for certain turbofan engine multi-variable rotation speed control system. Each dynamic neural network was used to identify proportional coefficient k_P, differential coefficient k_d and integral coefficient k_i, of its relevant PID decoupling controller on-line. Trans-dimensional learning as a software platform is added to the loop to improve the learning efficiency. When system unmodelled dynamics and random noise disturbance are taken into account, simulation results demonstrate the proposed decoupling strategy has strong robustness for the uncertainty and nonlinearity of aero-engine model. And it provides better disturbance rejection and adaptive capacity of the control loop than those achieved by a conventional PID decoupling controller.