航空发动机的智能神经网络自适应控制研究  

New intelligent neural network adaptive control scheme research for aeroengine

作　　者：赵俊[1] 陈建军[1] 王灵刚 

机构地区：[1]西安电子科技大学机电工程学院,西安710071 [2]中航一集团电光设备研究所,洛阳471009

出　　处：《航空动力学报》2008年第10期1913-1920,共8页Journal of Aerospace Power

基　　金：国家863项目(2006AA04Z402);陕西省自然科学基金(2005A09)

摘　　要：针对结构复杂、模型不确定、强非线性的航空发动机对象,提出一种综合模糊推理、神经网络自适应和PID简单控制各自优点的控制方案.在改进模糊PID控制器的基础上,进行了新型智能型神经网络控制器的设计,并提出离线混沌蚁群优化与在线误差反传调整相结合的优化方法.应用具有良好泛化能力的最小二乘支持向量机进行系统辨识,对某型航空发动机进行了设计点处的线性和非线性模型控制仿真.结果表明:控制系统具有满意的动、静态性能和较好的鲁棒性,验证了该方案的可行性和有效性.In view of complicated, undetermined and strongly nonlinear aeroengine ob ject, a novel control scheme integrating the merits of fuzzy inference, neural network adapta tion and simple proportional-integral-derivative（PID） method was presented. Based on the improved fuzzy PID controller, a novel intelligent neural network controller was designed. The parameters of the controller were optimized by the mixed learning method integrating of fline chaotic ant colony optimization and online error back propagation （BP） algorithm. By applying least square support vector machine for system identification, with its fine nonlinear mapping ability, Simulation results at the design point show that the control system has good robustness and better performance than conventional control methods. The feasibility and validity of the control scheme are proved.

关 键 词：航空发动机 智能神经网络 模糊PID控制 混沌蚁群优化 误差反传算法 最小二乘支持向量机 

分 类 号：V233.7[航空宇航科学与技术—航空宇航推进理论与工程]