摘要
提出一种基于递归小波网络建立直升机动力学模型的方法,给出了模型结构并利用粒子群优化算法对小波网络权参数进行调整。这种网络模型利用内部状态反馈来描述系统的非线性动力学特性,不必事先确定系统模型的类别和阶次,直接建立操纵量与加速度、角速度等变量之间的对应关系,从而避免了建立复杂的气动系数、气动导数模型的过程。与传统的机理模型相比,该模型具有结构简单、并行解算程度高、运算速度快的特点。在某型直升机飞行模拟器中取得了良好的应用效果。
A new modeling method for helicopter dynamic simulation model using recurrent wavelet neural networks was proposed. The whole model was given and its unknown parameters were tuned using a Particle Swarm Optimization method.This kind of network model is used the feedback of internal status to describe the system characteristics of nonlinear dynamics instead of that the system model＇s sort and order must be determinated at first. It is directly established the corresponding relation between the variants of control, acceleration and angle rate so as to avoid the complex course that the models of aerodynamic coefficient and derivative are established.Compared with the traditional mechanism model, the model obtained with the method owns the following advantages, namely simple structure, highly parallel solving and fast speed. Project application to a helicopter simulator demonstrates the effectiveness of the algorithm.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2007年第4期716-719,共4页
Journal of System Simulation
关键词
直升机
动力学模型
递归小波网络
粒子群优化
helicopter
dynamic model
recurrent wavelet neural networks
particle swarm optimization
