具有网络调度的航空发动机分布式控制系统协同容错控制

Collaborative Fault Tolerant Control for Aero-EngineDistributed Control System with Network Scheduling

关键词:  航空发动机  分布式控制系统  网络调度  网络参数，协同容错控制

Key words:  Aero-engine  Distributed control system  Network scheduling  Network parameter  Collaborative fault tolerant control

摘要:

为解决航空发动机分布式控制系统中网络参数与控制性能之间的冲突问题，提出了一种基于线性矩阵不等式理论的网络参数与保成本容错控制器协同设计方法。建立MEF-TOD调度协议作用下状态变量和控制变量的传输特性；将采样周期和数据包容量作为未知量引入航空发动机分布式控制系统的建模过程，得到调度协议约束下的网络参数与控制系统参数联合模型；给出联合闭环系统渐进稳定且存在成本函数上界的充分条件，并给出了网络参数与控制器增益的具体求解步骤。仿真结果显示，控制器与网络参数的协同设计方法能够求解出最优采样周期和数据包容量，据此得到容错控制器能够使航空发动机分布式控制系统在发生主动丢包故障的情况下，联合闭环系统渐进稳定，且低压转子转速超调量降低了80%，参数摆动降低了66.7%，保证控制器动态性能最优。

Abstract:

A co-design approach of network parameters and guaranteed cost fault tolerant controller based on the theory of linear matrix inequalities was put forward to solve the contradiction between network parameters and control performance. First of all， the transmission characteristics of state and control variables under the MET-TOD protocol were given. Secondly， Sampling period and packet size were introduced into the modeling process of aero-engine distributed control system as the unknown quantity， and the joint model of network parameters and controlling parameters was established under the scheduling agreement. Finally， the sufficient condition was given， through which the joint closed-loop system was asymptotically stable and had the upper bound of cost function. Furthermore， a heuristic optimization method was proposed to get the network and controller gain. The simulation results show that the optimal sampling period and packet size can be obtained by using the co-design method concerning controller and network parameters. Thus the conclusion is drawn that the fault tolerant controller can provide the asymptotical stability of joint closed-loop system for the aero-engine distributed control system and guarantee the optimal performance of the controller at the same time. There were 80% reduction in the overshoot of low-pressure?rotor?speed and 66.7% reduction in parameter fluctuation.