基于GA-SVM梯形区域的故障诊断与可靠配置

Fault diagnosis and reliable configuration of trapezoidal region based on GA-SVM

针对一类线性定常飞控系统,基于梯形区域下,研究了执行器连续增益故障可靠控制的问题。同时,为解决支持向量机选取参量困难而导致故障信息难以识别等问题,提出实值编码遗传算法来实现支持向量机模型参数的自动寻优。与传统的支持向量机相比,该方法具有收敛速度快、能耗低、耗时短、迭代次数少等优点。对于未知系统极点信息难以获取的问题,给出了极点观测器的算法,实现了对极点数据的实时观测。仿真结果表明,依据不同通道发生故障时极点所处区域不同的特点而设计的可靠控制器,不但鲁棒性能好,而且准确率也较高。该方法与粒子群算法(PSO-SVM)、网格搜寻法(gridsearch)相比具有较好的泛化能力及较高的识别度。

To a class of linear time invariant flight control systems,in terms of trapezoidal region,the problem of actuator continuous gain fault and reliable control is studied.At the same time,in order to solve the problem that it is difficult to select parameters of support vector machine,which leads to the difficulty of fault information identification,a real coded genetic algorithm is proposed to realize the automatic optimization of model parameters of support vector machine.Compared with the traditional support vector machine,this method has the advantages of fast convergence speed,low energy consumption,short time consuming and less iterations.For the problem that it is difficult to obtain the pole information of the unknown system in this paper,the algorithm theory of the pole observer is given to realize the real-time observation of the pole data.The simulation results show that the reliable controller designed according to the characteristics of different regions where poles are located when different channels fail has good robustness and high accuracy.Compared with PSO-SVM and grid search,this method has better generalization ability and higher recognition degree.