王宁,潘慕绚,黄金泉.基于在线滚动序列核极限学习机的涡轴发动机非线性模型预测控制[J].航空发动机,2018,44(5):
基于在线滚动序列核极限学习机的涡轴发动机非线性模型预测控制
Nonlinear Model Predictive Control for Turbo-Shaft Engine Based on the Online Sliding Sequence KernelExtreme Learning Machine
  
DOI:
中文关键词:  控制系统  核极限学习机  在线滚动序列  非线性模型预测控制  涡轴发动机
英文关键词:dcontrol system  KELM  online sliding sequence  NMPC  turbo-shaft engine
基金项目:南京航空航天大学研究生创新基地(实验室)开放基金(No.kfjj20160211)资助
作者单位E-mail
王宁,潘慕绚,黄金泉 南京航空航天大学能源与动力学院南京210016 2425599148@qq.com 
摘要点击次数: 1060
全文下载次数: 0
中文摘要:
      要:针对涡轴发动机控制系统设计,提出了1 种基于在线滚动序列核极限学习机的非线性模型预测控制方法。综合考虑直 升机旋翼扭矩、燃气涡轮转速、动力涡轮转速、涡轮级间温度和压气机喘振裕度等信息,设计具有较好实时性、精度和泛化能力的多 输出在线滚动序列核极限学习机作为预测模型,引入预测模型输出与发动机输出的误差进行反馈校正,利用序列二次规化算法在 线求解包含限制约束的预测控制问题。在某型直升机/ 涡轴发动机综合平台的仿真环境中进行了直升机大幅度机动飞行仿真验 证,结果表明:该模型预测控制器相比于传统串级控制具有更好的控制品质,可显著降低动力涡轮转速超调/ 下垂量
英文摘要:
      A nonlinear model predictive control (NMPC)method based on the online sliding sequence kernel extreme learning machine was proposed for the design of turbo-shaft engine control system.Considering the information of helicopter rotor torque,gas turbine speed,power turbine speed,turbine stage temperature and compressor surge margin,a multi output on -line sliding sequence kernelextreme learning machine with good real-time,precision and generalization ability was designed as a prediction model. The deviation between predictive model output and engine output was introduced for feedback correction, and the sequential quadratic planning algorithm was used to solve the predictive control problem with restricted constraints online. In the simulation environment of a helicopter / turboshaft engine integrated platform,the simulation of helicopter large maneuver flight was carried out. The results show that the model predictive controlhas better than the traditionalcascade control,and can significantly reduce the overshoot /droopof the powerturbine speed.
查看全文  查看/发表评论  下载PDF阅读器