Abstract: According to the characteristics of large time delay, nonlinearity and time-varying in micro gas turbines, based on the theory of system identification and adaptive generalized predictive control algorithm, a new method of micro turbine speed control is presented. Frist, through the acquisition of the micro gas turbine rotor system input (fuel) - output (rotor speed) data, the CARIMA model of the micro gas turbine rotor speed system is identified by using the FFRLS algorithm. Then, an adaptive generalized predictive controller of PSO is designed based on the identified CARIMA model. Finally, simulations are carried out in MATLAB. Simulation results show that, when the load is abrupt, the fuel quantity is fast, the speed of the rotating speed is small, the tracking effect is perfect, the robustness is strong, and the control performance is also good.

Key words: Off-line identification, microturbine, particle swarm optimization algorithm, adaptive generalized predictive control

摘要： 针对微型燃气轮机运行时的大时滞、非线性和时变性等特点，基于系统辨识理论和自适应广义预测控制算法设计了一种微型燃气轮机的转速控制器。首先，通过现场采集微型燃气轮机转速控制系统输入(燃料量)-输出(控制转速)的数据，利用变遗忘因子递推最小二乘算法(FFRLS)进行离线辨识，辨识出微型燃气轮机转速控制系统在特定负荷工况下的受控自回归积分滑动平均模型(CARIMA)。然后，基于所辨识的CARIMA模型，研究并设计了基于粒子群算法(PSO)的自适应广义预测控制器。最后，在MATLAB仿真软件中进行了仿真。仿真结果表明，当系统负荷发生突变时，燃料量流量响应迅速，控制转速超调量较小，过渡过程时间较短，跟踪效果好，鲁棒性较强，控制性能好。

关键词: 离线辨识, 微型燃气轮机, 粒子群算法, 自适应广义预测控制