基于RBF-ARX模型的绞吸挖泥船横移过程模型设计  

Swing Process Model Design of a Cutter Suction Dredger Based on RBF-ARX Model

作　　者：蒋爽[1,2] 倪福生[1,2] 梁婷[3] 戴伟[1,2] JIANC Shuang;NI Fusheng;LIANC Ting;DAI Wei(Engineering Research Center of Dredging Technology,Hohai University,Changzhou Jiangsu 213022,China;College of Mechanical and Electrical Engineering,Hohai University,Changzhou Jiangsu 213022,China;Changzhou Branch,China Mobile Communications Group Jiangsu Co.,Ltd.,Changzhou Jiangsu 213022,China)

机构地区：[1]河海大学疏浚技术教育部工程研究中心,江苏常州213022 [2]河海大学机电工程学院,江苏常州213022 [3]中国移动通信集团江苏有限公司常州分公司,江苏常州213022

出　　处：《机床与液压》2019年第8期22-28,共7页Machine Tool & Hydraulics

基　　金：江苏省科技成果转化专项资金项目(BA2016143)

摘　　要：由于绞吸挖泥船横移过程的影响因素众多,具有明显的非线性和时变性特征,难以精确建模来描述其动态特性。因此在对挖泥船产量形成过程详细分析的基础上,针对传统神经网络收敛速度慢、隐含层较多、训练阶段无法自适应调整网络结构的缺点,采用RBF-ARX模型结构进行绞吸挖泥船产量和横移速度的建模。通过结构化非线性参数优化方法SNPOM,离线识别出RBF-ARX模型的线性和非线性参数,并利用仿真对所建模型和真实数据的误差进行对比。仿真结果表明:所建立的绞吸式挖泥船横移过程模型能够比较准确描述系统在全局范围内的动态特性,与挖泥船的实际产量输出拟合较好。There are many factors affecting the swing process of a cutter suction dredger. They change with time and nonlinear. So it is difficult to describe the change law accurately. The swing process model was established based on radical basis function-auto regressive exogenous(RBF-ARX)model after detailed analysis of the formation process of dredger production.This method could overcome the shortcomings of traditional neural network, such as slow convergence rate, more hidden layer. Also it could overcome the shortcoming that the network structure could not be adaptively adjusted during the training phase.After modeling, the linear and nonlinear parameters of the RBF-ARX model were identified off-line by using structured nonlinear parameter optimization method(SNPOM).Then the errors of themodel output and the real data were comparedby simulation. The result show that the swing process model can accurately describe the dynamic characteristics of the system in the global range, and the model output is well fitted with the real data of a cutter suction dredger.

关 键 词：绞吸式挖泥船 横移过程模型 RBF-ARX SNPOM 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]