欢迎访问《兵工学报》官方网站,今天是 分享到:

兵工学报 ›› 2022, Vol. 43 ›› Issue (9): 2379-2387.doi: 10.12382/bgxb.2021.0867

• 论文 • 上一篇    下一篇

基于PSO-LSSVM弹药装配质量预测方法

裘镓荣1,2,3, 曾鹏飞1,2,3, 邵伟平1,2,3, 赵丽俊4, 郝永平2,3   

  1. (1.沈阳理工大学 机械工程学院, 辽宁 沈阳 110159; 2.沈阳理工大学 辽宁省先进制造技术与装备重点实验室, 辽宁 沈阳 110159;3.沈阳理工大学 CAD/CAM技术研究与开发中心, 辽宁 沈阳 110159; 4.北方华安工业集团有限公司, 黑龙江 齐齐哈尔 161046)
  • 上线日期:2022-07-05
  • 通讯作者: 曾鹏飞(1978—),男,副教授,硕士生导师 E-mail:pfzeng@163.com
  • 作者简介:裘镓荣(1998—), 女, 硕士研究生。 E-mail: 939730495@qq.com
  • 基金资助:
    国防技术基础科研项目(JSZL2020208A001); 辽宁省应用基础研究计划项目(2022年)

PSO-LSSVM-Based Ammunition Assembly Quality Prediction Method

QIU Jiarong1,2,3, ZENG Pengfei1,2,3, SHAO Weiping1,2,3, ZHAO Lijun4, HAO Yongping2,3   

  1. (1.School of Mechanical Engineering, Shenyang Ligong University, Shenyang 110159, Liaoning, China; 2.Liaoning Province Key Laboratory of Advanced Manufacturing Technology and Equipment, Shenyang Ligong University, Shenyang 110159, Liaoning, China; 3.R&D Center of CAD/CAM Technology, Shenyang Ligong University, Shenyang 110159, Liaoning, China; 4.North Hua'an Industry Group Co., Ltd., Qiqihar 161046, Heilongjiang, China)
  • Online:2022-07-05

摘要: 针对弹药装配工艺复杂、装配工序质量影响因素多、装配效率低的实际问题,提出基于粒子群优化(PSO)算法—最小二乘支持向量机(LSSVM)的弹药装配质量预测方法。通过灰熵关联分析方法,提取影响弹药装配质量的关键质量特性,并将其作为预测模型的输入向量,降低预测模型复杂度和运算工作量。将PSO-LSSVM作为建模工具,利用PSO算法优化LSSVM参数,建立基于PSO-LSSVM弹药装配质量预测模型;以预测某型号弹药对接装配工序中跳动量为例,与LSSVM 预测模型和BP神经网络预测模型进行对比分析。实验结果表明,提出基于PSO-LSSVM弹药装配质量预测方法具有可行性和有效性,能够很好地实现弹药装配质量的预测。

关键词: 弹药装配, 质量预测, 灰熵关联分析, 粒子群优化, 最小二乘支持向量机

Abstract: Based on particle swarm optimization (PSO)-least squares support vector machines (LSSVM), an ammunition assembly quality prediction method is proposed to address the problems of complex ammunition assembly process, existence of various influencing factors for the assembly process quality, and low assembly efficiency. Through gray entropy correlation analysis, key quality characteristics affecting the ammunition assembly quality are extracted and used as the input vectors of the prediction model to reduce the complexity and calculation workload. Using PSO-LSSVM as the modeling tool, the parameters of LSSVM are optimized by the PSO algorithm, and a prediction model is developed for ammunition assembly quality. Taking the prediction of runout during the docking assembly of a certain type of ammunition as an example, the model is compared with LSSVM model and Back-Propagation (BP) neural network prediction model. The experimental results demonstrate that the proposed PSO-LSSVM-based prediction method for ammunition assembly quality is feasible and effective, which can accurately predict the quality of ammunition assembly.

Key words: ammunitionassembly, qualityprediction, grayentropycorrelationanalysis, particleswarmoptimization, leastsquaressupportvectormachine

中图分类号: