基于神经网络自适应滤波的直升机悬停高度稳定研究  

Effective Adaptive Filtering for a Helicopter′s Hovering Altitude Using Neural Networks

作　　者：邢小军[1] 闫建国[1] 袁冬莉[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072

出　　处：《西北工业大学学报》2011年第4期627-631,共5页Journal of Northwestern Polytechnical University

基　　金：国家自然科学基金(60974146);陕西省自然科学基础研究计划(2011GQ8005);西北工业大学基础研究基金(NPU-FFR-JC20100216)资助

摘　　要：针对直升机执行海上搜救任务时定点悬停高度容易受到海浪干扰的问题,提出了一种应用遗传BP神经网络对直升机无线电高度信号进行自适应滤波的方法。首先在建立海浪有理谱简化模型的基础上,应用遗传算法对BP神经网络的初始权值和阈值进行全局粗精度预学习,再应用遗传优化后的BP神经网络对实际海浪干扰信号进行建模,并通过噪声对消原理滤除无线电高度计中混杂的海浪干扰噪声。仿真结果表明,所设计的自适应滤波器能够较好地抑制海浪噪声对直升机高度控制的影响,而且对海浪干扰统计特性的变化具有较强的鲁棒性和适应性。With the purpose of overcoming the problem that a helicopter＇ s hovering altitude control is easily affected by sea wave disturbances when carrying out maritime search and rescue mission, we propose an adaptive filtering approach for the helicopter＇ s altitude signal using GA-based BP neural networks （ GA stands for genetic algorithms and BP stands for backward propagation）. Section 1 of the full paper briefs the model of sea wave disturbances. The core of section 2 consists of： （ 1 ） Fig. 1 is the schematic showing how sea wave disturbances existing in the helicopter＇ s radio altimeter are filtered according to the noise cancellation principle ; （2） we give a six-step procedure for performing adaptive filtering, The simulation results, presented in Figs. 2 through 4, and their analysis show preliminarily that the adaptive filter can indeed suppress the sea wave disturbances effectively with good robustness and adaptability.

关 键 词：神经网络 自适应滤波 算法 直升机 海浪噪声 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]