基于云粒子群小波网络的发动机起动模型辨识  

A Dynamic Identification of Aero-engine Starting Model Based on Cloud Adaptive Particle Swarm Optimization Wavelet Neural Network

作　　者：黄帅[1] 李本威[1] 

机构地区：[1]海军航空工程学院,山东烟台264001

出　　处：《计算机仿真》2011年第6期71-74,共4页Computer Simulation

摘　　要：起动过程是航空发动机顺利进入正常工作的前提,为了克服过程中存在的部件低转速特性获取困难,保证飞行安全,针对机理建模难度大和传统神经网络辨识建模易陷入局部最优、过学习等问题。为解决上述问题提出了一种云粒子群小波网络的起动模型辨识的新方法,方法采用的辨识网络兼具了云粒子群的全局快速寻优能力和小波网络良好的非线性逼近能力,并通过某发动机起动数据样本的训练,建立了起动模型,仿真结果表明方法的辨识精度高、鲁棒性好、泛化能力强,验证了起动过程合理性,为进一步研究起动性能提供了一种可靠的新途径。The starting is the first step in aero-engine running process,but it is difficult to get the low speed characters and to model in mechanism method.And the identification with traditional neural networks often settles in the local minimum error and has low convergence accuracy.In order to solve these problems,the Cloud Adaptive Particle Swarm Optimization（CAPSO） Wavelet Neural Network（WNN） is constructed in this paper which has the advantages of CAPSO and WNN.Then a starting model is constructed based on the CAPSO WNN.The examination result validates the starting charactering and shows that the network has very high accuracy,good stabilization and strong popularizing ability,which supplies a new way to study the starting process higher.

关 键 词：粒子群算法 小波网络 起动模型 辨识 

分 类 号：V23[航空宇航科学与技术—航空宇航推进理论与工程]