论文部分内容阅读
针对具有随机有界双侧时延的航空发动机分布式控制系统,提出了一种基于多步预测和关联向量机(RVM)回归误差补偿的控制方案.首先建立航空发动机分布式控制系统(DCS)的神经网络非线性自回归滑动平均(NARMA)模型,利用当前的系统输出和控制量对N步之后的系统输出进行预测;其次用改进的RVM回归多步预测算法估计NARMA模型的的预测误差,并对预测结果进行误差补偿;最后利用补偿之后的预测值和设定值对控制参数进行滚动优化,设计系统的神经网络逆控制器实现系统的自适应控制.仿真结果证明该控制策略能够避免随机有界双侧时延对控制系统的影响,实现对设定值的稳定跟踪,且控制器具有较好的实时性和鲁棒性.低压转子转速阶跃响应的稳态绝对误差小于0.04%,响应时间小于0.3s.
A distributed control system based on multi-step prediction and regression vector regression (RVM) is proposed for aero-engine distributed control system with random and bounded bilateral delays. Firstly, aero-engine distributed control system (DCS) (NARMA) model of neural network is proposed to predict the system output after N steps by using the current system output and control variables. Secondly, the improved RVM regression multi-step prediction algorithm is used to estimate the NARMA model’s prediction error, Finally, the control parameters are scrolled and optimized by using the predicted value and the set value after compensation, and the neural network inverse controller of the system is designed to realize the adaptive control of the system.The simulation results show that the control strategy can avoid random The influence of bounded bilateral delay on the control system and the steady tracking of the setpoint are achieved, and the controller has good real-time performance and robustness.The steady-state absolute error of the step response of low-voltage rotor speed is less than 0.04% Response time is less than 0.3s.