基于机器学习的船舶多支撑轴系智能装调方法

Intelligent Assembly and Adjustment Method of Ship Multi Support Shafting Based on Machine Learning

针对船舶多支撑轴系装调效率低、校中调整方案难确定问题,开展基于机器学习的中间轴承精确变位值计算方法研究。通过轴系装调问题分析明确机器学习的任务,选取GA-BP神经网络作为机器学习的模型,调查轴系设计计算、制造检验、实船装调过程中产生的可用于机器学习的数据类型、数量和质量,并以由轴系设计模型计算得到的仿真数据样本集、结合仿真数据与实船装调实测数据的复合样本集,来分别训练轴系装调机器学习模型,提出基于轴系装调机器学习模型的轴承变位值计算方法。以一艘汽车滚装船的轴系校中为例进行验证,结果表明:机器学习模型通过对已有数据的学习能较精确地预测出各个中间轴承的变位值,从而快速形成轴系调整方案,减少调整次数,有助于提升船舶多支撑轴系的安装质量和效率。

This article presents a calculation method of displacement based on machine learning to solve the problem of alignment of ship multi-support shafting.Technological process and difficulty in multi-support shafting alignment are analyzed.Task of machine learning,data type,quantity and quality used in machine learning is clarified.A machine learning model of shafting alignment is constructed,and calculation method of bearing displacement is designed.Taking the shafting alignment of a RoRo vessel as an example,and results show that the machine learning model can accurately predict the displacement of each intermediate bearing,so as to reduce the adjustment time,and improve installation efficiency and quality of multi-support shafting.