Computer Integrated Manufacturing System ›› 2024, Vol. 30 ›› Issue (3): 968-981.DOI: 10.13196/j.cims.2022.0642

Previous Articles     Next Articles

Ship electrical layout method using piror rules and deep neural networks

HUANG Yixue1,QIN Ke1,LUO Wei2+,WU Sheng3,HAO Jia4,XIA Lin2   

  1. 1.Twelve Department,China Ship Research and Design Center
    2.Technology Innovation Center,China Ship Research and Design Center
    3.Ten Department,China Ship Research and Design Center
    4.School of Mechanical Engineering,Beijing Institute of Technology
  • Online:2024-03-31 Published:2024-04-02
  • Supported by:
    Project supported by the National Defense  Basic Research Program,China(No.JCKY2022206B002).

先验规则和深度学习融合驱动的舰船电气图纸布局方法

黄一学1,秦克1,罗威2+,吴盛3,郝佳4,夏琳2   

  1. 1.中国舰船研究设计中心一二室
    2.中国舰船研究设计中心科技创新中心
    3.中国舰船研究设计中心一○室
    4.北京理工大学机械与车辆学院
  • 基金资助:
    国防基础科研计划资助项目(JCKY2022206B002)。

Abstract: Aiming at the problems of low automation degree of electrical drawing design,high labor consumption and error-proneness in the current ship overall design process,an automatic layout of electrical drawings that integrated prior rules and deep learning,overall layout and wiring was proposed,which could be applied to engineering practice.According to the topological characteristics of the current ship electrical schematic layout,the typical prior rules were extracted based on the topology structure of the tree diagram;the electrical connection relationship was used as the input to automatically generate the preliminary electrical schematic diagram;the singular value decomposition method was used to extract connectivity features,and preliminary electrical schematic parameters were optimized using a deep neural network.Seven typical drawings of a certain type of ship were selected to carry out application verification.The results showed that:①the proposed method could realize the task of automatic electrical design layout based on ensuring the correct connection relationship between drawings;②in the wiring of large samples 99.1% of the drawings could be routed within 10 seconds.This technology could be applied to all electrical wiring tasks where the connection relationship was a tree topology or could be converted into a tree topology to achieve correct,fast and reasonable automatic wiring layout of electrical drawings,which effectively improved the degree of automation in electrical layout and wiring work and the overall design capability and efficiency of the ship.

Key words: prior rules, neural network, singular value decomposition, graph decomposition

摘要: 针对目前舰船总体设计过程中电气图纸设计自动化程度低、人力消耗大、易出错等问题,提出一种融合先验规则和深度学习、统筹布局和布线、可应用于工程实践的电气图纸自动化布局方法。首先,根据当前舰船电气原理图布局的拓扑特点,基于树状图拓扑结构提炼出典型的先验规则;其次,以电气连接关系为输入,自动化生成初步电气原理图;最后,采用奇异值分解方法提取连接关系特征,并使用深度神经网络对初步电气原理图参数进行优化。选取某型舰船的7张典型图纸开展应用验证,结果表明:①所提方法可在保证图纸连接关系正确的基础上,实现自动化电气设计布局布线任务,②在大样本的布线试验中,99.1%的图纸都能在10秒内完成布线。所提方法可以应用在所有连接关系为树状拓扑或可以转化为树状拓扑的电气布线任务中,实现电气图纸正确、快速、合理的自动化布线布局,有效提高电气布局布线工作中的自动化程度,进而提高舰船的总体设计能力和效率。

关键词: 先验规则, 神经网络, 奇异值分解, 图分解

CLC Number: