基于PSO-BP神经网络模型的运输车旋转支撑座轻量化研究  

Light quantization of rotary support seat of transporter based on PSO-BP neural network model

作　　者：刘兆迎 王正方[1,2] 孙传祝 韩鹏飞[1] 张川[1] LIU Zhao-ying;WANG Zheng-fang;SUN Chuan-zhu;HAN Peng-fei;ZHANG Chuan(School of Mechanical Engineering,Shandong University of technology,Zibo 255000,China;School of Electrical ＆ Electronic Engineering,Zibo Vocational College,Zibo 255000,China)

机构地区：[1]山东理工大学机械工程学院,山东淄博255000 [2]淄博职业学院电子电气工程学院,山东淄博255000

出　　处：《机电工程》2018年第6期577-581,共5页Journal of Mechanical & Electrical Engineering

基　　金：山东省重大创新工程资助项目(2017CXGC0219)

摘　　要：针对多功能运输车旋转支撑座结构偏重、生产成本高,且传统设计论证周期长、效率低等问题,提出了一种基于有限元理论与神经网络模型相结合的结构轻量化设计方法。通过Solidworks和Ansys建立了旋转支撑座有限元模型,并对支撑座进行静力进行了仿真分析;根据拉丁超立方抽样及有限元法获取了设计参数样本空间,构建了基于粒子群优化的BP神经网络模型,代替有限元模型确定了设计变量与性能参数之间的非线性关系,最后代入数学模型并在Matlab中计算求得了最优解,并将所求结果代入Ansys中模拟论证。研究结果表明:在满足应力、应变约束条件下,优化后支撑座的质量减少5.4%;该方案论证周期短、设计制造成本低。Aiming at the problem of the heavy structure of the rotary support seat of multifunctional transportation vehicle, the high manufac- turing cost and a long cycle and low efficiency of the traditional design demonstration, a lightweight design method based on finite element theory and neural network model was proposed. Through Solidworks and Ansys, the finite element model of the rotating support was estab- lished and the static simulation analysis of the support was earned out. According to the Latin Hypercube Sampling and finite element meth- od, the sample space of the design parameters was obtained, and a BP neural network model based on particle swam1 optimization （PSO） was constructed. The nonlinear relationship between the design variables and the performance parameters was detemfined instead of the finite element model. Finally, the results were obtained by Ansys simulation demonstration in the calculation of the optimal solution. The results shows that the optimized quality is reduced by 5.4% under the stress and strain constraints. The program has a short period and saves the de- sign and manufacturing costs.

关 键 词：旋转支撑座 有限元 粒子群算法 BP神经网络 MATLAB 

分 类 号：TH122[机械工程—机械设计及理论] S229.1[农业科学—农业机械化工程]