摘要
应急救援车辆的三角履带轮需要在各种地形和路况条件下具备高机动性和可靠性。针对三角履带支撑架在多工况疲劳优化中的计算时程过长问题,以组合代理模型来加速优化求解,为了保证代理模型的精度,提出新的代理模型评价指标。首先,对可更换三角履带轮的支撑架在四个工况下的疲劳寿命进行分析;然后,以克里金模型和响应面模型为基础对支撑架的质量及四个工况的疲劳寿命进行组合代理模型的构建,其中,为了在寻优过程中更易逼近组合代理模型的全局最优点,使得预测值和观测值具有一致的凹凸性,提出了局部趋势误差作为构建组合代理模型的评价标准;最后,对支撑架的数学模型进行了尺寸优化。结果表明,组合代理模型不但保证了计算精度,而且具备一定的鲁棒性,减重后的三角履带轮在保证质量的前提下寿命提高了43.59%。
Triangular track wheels of emergency rescue vehicles need high mobility and reliability under various terrain conditions.Aiming at overlong time of multi-working fatigue optimization calculation,ensemble of surrogate model is used to accelerate the optimization solution.In order to ensure convergence accuracy,a new evaluation index of surrogate model is proposed.First of all,fatigue life of support frame of replaceable triangular track wheel under four working conditions is analyzed.Then,based on Kriging model and response surface model,ensemble of surrogate model is constructed for mass and fatigue life.In order to approach global optimal point more easily,ensure same concavity and convexity between predicted value and observed value,local trend deviation is proposed as evaluation criterion of constructing ensemble model.Finally,size optimization of support frame is constructed.Results show ensemble of surrogate model not only ensures calculation accuracy,but also has robustness.Fatigue life of triangular track wheel after lightweight design is increased by 43.59%.
作者
李永欣
常涛
杨立明
李凯伦
孙银旭
吴凤和
LI YongXin;CHANG Tao;YANG LiMing;LI KaiLun;SUN YinXu;WU FengHe(School of Mechanical Engineering,Yanshan University,Heavy-Duty Intelligent Manufactuing Equipment Innovation Center of Hebei Province,Qinhuangdao 066004,China;Space Star Technology Co.,Ltd.,Beijing 100095,China;Hangcha Group Co.,Ltd.,Hangzhou 311305,China)
出处
《机械强度》
CAS
CSCD
北大核心
2021年第5期1088-1094,共7页
Journal of Mechanical Strength
基金
国家自然科学基金项目(51890881)
河北省教育厅高等学校科技计划青年基金项目(QN2018228)资助。
关键词
三角履带轮
多工况尺寸优化
组合代理模型
疲劳寿命
局部趋势误差
Triangular track wheel
Multi-working size optimization
Ensemble of surrogate model
Fatigue life
Local trend deviation