摘要
船舶运动模型是船舶操纵模拟器的基础,以船舶受力分析为基础的水动力模型是当前应用最广、最有效的方法之一,但实际应用中由于干扰因素多、受力非线性、水动力导数难以确定等问题导致了该方法精度与实时性难以兼容。为解决多种外界作用下船舶运动的准确性与实时性,文章提出一种基于径向基神经网络(RBFN)的运动拟合方法,构造三层神经网络模型,其中隐含层和输出层神经元的传递函数分别取高斯函数和线性函数,通过对积累的历史数据进行学习,利用径向基函数网络在逼近能力、分类能力和学习速度等方面优势,实时推算船舶运动趋势。仿真结果表明,方法实时性能好,具有较高的精度,展现了广阔的应用前景。
The ship motion model is the critical part of the ship maneuvering simulation system. The method that stress analysis based on hydrodynamic model is one of the most widely used and effective solution. But in the practical application, there are so many interference factors, nonlinear stress, and the hydrodynamic derivative problems, that makes it difficult to keep the balance accuracy and real-time performance. In order to solve the accuracy of ship motion under the action of a varie- ty of external and real time, a kind of radial basis function neural network（RBFN） is presented based on the moving fitting method. Firstly, a RBFN with the structure of the three layer neural network model is built, the transfer function of the hid- den layer and the output layer neurons are set to the Gauss function and linear function respectively, through the study on ac- cumulation of historical data to takes the advantage of approximation ability, classification ability and learning speed of RBFN to calculate the vessel＇s movement. The result of simulation shows that the method is good at real-time performance, robust- ness, precision, and it has broad application prospects.
出处
《计算机与数字工程》
2015年第9期1588-1591,共4页
Computer & Digital Engineering
关键词
径向基神经网络
船舶操纵模拟系统
响应模型
radical basis function networks, ship maneuvering simulation system, motion model
