中国安全科学学报 ›› 2023, Vol. 33 ›› Issue (6): 114-121.doi: 10.16265/j.cnki.issn1003-3033.2023.06.1033

• 安全工程技术 • 上一篇    下一篇

基于RS-RBFNN的邮轮建造物资物流集配风险预警

谢露强1(), 徐靖2, 王海燕1,3   

  1. 1 武汉理工大学 交通与物流工程学院,湖北 武汉 430063
    2 上海外高桥造船有限公司,上海 200137
    3 国家水运安全工程技术研究中心 武汉理工大学,湖北 武汉 430063
  • 收稿日期:2023-01-12 修回日期:2023-04-14 出版日期:2023-08-07
  • 作者简介:

    谢露强 (1997—),男,江西抚州人,硕士研究生,研究方向为安全与风险管理。E-mail:

  • 基金资助:
    工业和信息化部研发专项项目(MC-202009-Z03)

Risk early warning model of cruise ship construction material logistics collection and distribution based on RS-RBFNN

XIE Luqiang1(), XU Jing2, WANG Haiyan1,3   

  1. 1 School of Transportation and Logistics Engineering, Wuhan University of Technology, Wuhan Hubei 430063, China
    2 Shanghai Waigaoqiao Shipbuilding Co., Ltd., Shanghai 200137, China
    3 National Engineering Research Center for Water Transport Safety, Wuhan University of Technology, Wuhan Hubei 430063, China
  • Received:2023-01-12 Revised:2023-04-14 Published:2023-08-07

摘要:

为解决邮轮建造物资物流集配层级复杂、耦合因素众多引起的产需脱节问题,提出一种粗糙集(RS)融合径向基神经网络(RBFNN)的集成风险预警模型。首先,基于物资物流集配风险因素分析,构建风险预警指标体系,利用网络分析法(ANP)建立指标间相互依赖和反馈关系评价模型,并据此确定指标重要度;其次,采用功效系数法确定历史数据警情,并以此作为神经网络输出端;最后,利用RS作为RBFNN的前置处理系统,对某邮轮建造过程机电物资物流集配进行风险预警建模,并与RBFNN、反向传播神经网络(BPNN)和RS-BPNN进行性能比较。结果表明:RS-RBFNN模型能有效简化神经网络结构,提高测效率和准确性,克服BP网络训练时间长、稳定性较差且容易陷入局部极小的弊病。

关键词: 粗糙集(RS), 径向基神经网络(RBFNN), 邮轮建造物资, 物流集配, 风险预警, 反向传播神经网络(BPNN)

Abstract:

An ensemble risk early warning model based on RS and RBFNN fusion was proposed. After constructing the risk early warning index system, the ANP was used to analyze the coupling relationship between indicators and determine the weight values. The efficacy coefficient method was used to determine the historical data alarm, which was used as the output of the neural network. The RS-RBFNN model was used to conduct a risk early warning research on the logistics of mechanical and electrical materials during the construction of a cruise ship, and the performance was compared with RBFNN, BPNN and RS-BPNN. The research shows that the RS-RBFNN model effectively simplifies the neural network structure, improves efficiency and accuracy, and overcomes the shortcomings of BPNN of training time long, poor stability and easy to fall into local minima.

Key words: rough set (RS), radial basis function neural network (RBFNN), cruise ship construction material, logistics collection and distribution, risk early warning, back propagation neural network (BPNN)