摘要
针对直翼桨推进器螺距点驱动机构存在的强非线性特征及传统解析建模困难的问题,提出一种基于神经网络的螺距点运动学解算方法。基于平面几何原理推导螺距点驱动机构的数值运动学解,通过建立多连杆几何关系方程组,分别求得纵向和横向作动器转角与螺距点坐标间的映射关系。利用神经网络对样本解集进行拟合,构建结构简洁、计算高效的运动学模型。仿真结果显示:神经网络模型能高度逼近理论解,建立的仿真模型与拟合方法具有较高的精度和可行性。实物实验表明,纵向作动器跟踪误差绝大多数为[-0.02,0.04]%,横向作动器误差大部分在[-0.07,0.01]%内,仅个别点达3%;螺距点位置大部分跟踪误差分别为X向[0.20,0.81]mm、Y向[-0.10,0.14]mm,仅个别点大于3.5 mm;PLC计算耗时均小于10 ms,满足实时控制要求。研究结果为直翼桨推进器的控制系统设计提供了建模参考,对提升船舶智能化操控水平具有重要意义。
To address the strong nonlinear characteristics of the pitch point driving mechanism of the voith-schneider propeller and the difficulty of traditional analytical modeling,a neural-network-based kinematic solution method for the pitch point was proposed.Numerical kinematic solutions of the pitch point driving mechanism were derived based on planar geometric principles.By establishing a set of geometric equations for the multi-link mechanism,the mapping relationships between the longitudinal and transverse actuator rotation angles and the pitch point coordinates were obtained.A neural network was then used to fit the solution dataset,resulting in a structurally concise and computationally efficient kinematic model.Simulation results show that the neural network model closely approximates the theoretical solutions,demonstrating high accuracy and feasibility of the modeling and fitting method.Experimental tests indicate that the longitudinal actuator tracking error is mainly within[-0.02,0.04]%,while the transverse actuator error is mostly within[-0.07,0.01]%,with only a few points reaching 3%.The mostly pitch point position tracking errors fall within[0.20,0.81]mm in the X-direction and[-0.10,0.14]mm in the Y-direction,with only a few points larger than 3.5 mm.The PLC computation time is less than 10 ms in all cases,meeting real-time control requirements.This study provides modeling reference for the control system design of VSPs and contributes to enhancing the intelligent maneuvering capabilities of marine vessels.
作者
王刚毅
陈梓豪
常龙
刘佳佳
刘宏宇
陈波
陈正
WANG Gangyi;CHEN Zihao;CHANG Long;LIU Jiajia;LIU Hongyu;CHEN Bo;CHEN Zheng(Shanghai Marine Equipment Research Institute,Shanghai 200031,China;Ocean College,Zhejiang University,Zhoushan Zhejiang 316000,China)
出处
《机床与液压》
北大核心
2025年第23期101-107,共7页
Machine Tool & Hydraulics
关键词
直翼桨
神经网络
螺距点驱动机构
运动学建模
voith-schneider propeller
neural-network
pitch point driving mechanism
kinematic modeling
