航空发动机模型参考模糊自学习控制

Fuzzy Model Reference Learning Control of Aero-Engine

模型参考模糊自学习控制方法利用一个参考模型作为被控对象的输出性能要求 ,并且根据参考模型的输出和实际对象的输出间的误差 ,经过一个逆模糊模型学习产生模糊控制系统的控制规则。本文基于该方法离线构造了某型发动机的模糊控制器 ,并在发动机的其它状态及飞行包线的其它点进行在线学习校正 ,以得到期望的性能。仿真结果表明用该方法设计的模糊控制系统获得了满意的控制效果。

Fuzzy model reference learning control (FMRLC) uses a reference model and a learning mechanism to generate fuzzy control rules, and tunes them on-line to improve the performance of the closed control system. This paper uses the FMRLC to generate control rules of an aero-engine at its design point off-line. At the beginning of the simulation, the fuzzy control rules are zeros, which means that the fuzzy controller doesnot work. The transfer function of a second order system is used to act as the reference model. When there exists deviation between the real model output and the reference model output, the learning mechanism will generate and tune the fuzzy control rules to eliminate the deviation, thus resulting in good performance of the designed fuzzy controller. The control rules are modified under different operating conditions of the aero-engine and other points of the flight envelope on-line to achieve the desired performance. Digital simulation results verify the effectiveness of the control scheme.