为了实现当直升机旋翼负载变化时,尽量保持功率涡轮转速恒定,并提高系统动态品质,研究了一种针对涡轴发动机的自适应非线性预测控制(ANMPC)算法.基于涡轴发动机稳态数据和动态特性,采用递归最小二乘法(RLS)进行模型参数辨识,建立了具有在线自适应能力的涡轴发动机数值-ARX(auto re-gressive with external input)并联预测模型.在此基础上,通过多步输出预测和反馈校正,利用序列二次型优化(SQP)算法,进行在线滚动优化,从而获得了涡轴发动机ANMPC控制器.仿真结果表明:当旋翼负载变化时,相比于传统的串联PID(比例-积分-控制)控制器,ANMPC控制器能够使得功率涡轮转速收敛更快,超调量/下垂量更小. In order to keep the power turbine rotation speed as constant as possible and improve the dynamic quality of the system when the helicopter rotor load changes, an adaptive nonlinear predictive control (ANMPC) algorithm for turboshaft engine is studied. Based on the steady-state data of turboshaft engine and Dynamic characteristics, recursive least squares (RLS) model parameter identification was used to establish a parallel predictive model of ARX (auto re-gressive with external input) with on-line adaptive capability. On this basis, Step output prediction and feedback correction, the online rolling optimization is realized by the sequential quadratic optimization (SQP) algorithm, and the turbine turbine ANMPC controller is obtained.The simulation results show that when the rotor load changes, compared with the traditional series PID (Proportional-Integral-Control) controller, the ANMPC controller enables faster convergence of the power turbine speed with smaller overshoot / droop.