摘要
针对水下机器人操纵性优化设计中水动力系数预报问题,在水下机器人水动力预报中引入艇体肥瘦指数概念,确定了水下机器人艇体几何描述的五参数模型。提出采用小波神经网络方法预报水下机器人水动力,确定了神经网络的结构,利用均匀试验设计方法,设计了神经网络的学习样本。研究结果表明,只要确定适当的输入参数,选择适当的学习样本和网络结构,利用小波神经网络方法对水下机器人水动力进行预报可以达到较好的精度。
Autonomous underwater vehicles（AUVs） will be widely used for marine search and rescue in the near future. To solve the problem of hydrodynamic coefficient prediction in dirigibility optimizing of AUV design, the fat-thin index of AUV hull is applied in the hydrodynamic prediction and a five-parameter model is established for the geometric description of AUV. This paper presents a method to predict the hydrodynamic coefficients by using wavelet neural networks, confirms the structure of the networks and designs a series of hull models as the swatches for network study by using the method of uniform design. The research results indicate that for the wavelet neural network method, suitable input parameters, study swatches and net structure result in accurate hydrodynamic prediction for AUVs.
出处
《海洋技术》
2015年第1期50-54,共5页
Ocean Technology
关键词
水下机器人
操纵
水动力
小波神经网络
autonomous underwater vehicles（AUVs）
maneuver
hydrodynamic coefficients
wavelet neural networks
