基于反向传播神经网络的注塑模具用零件报价模型

doi:10.16183/j.cnki.jsjtu.2017.10.009

摘要/Abstract

摘要： 针对诸多通用报价方法在注塑模具用零件价格评估过程中存在的效率与精度等问题，提出了基于反向传播（Back Propagation, BP）神经网络的多专家报价模型.利用遗传算法优化BP网络权值和阈值抵御随机参数引起的局部最优解，同时对报价误差的高斯分布规律进行分析以确定理想的拓扑结构.通过多专家模型给出的多个报价，提出了价格评估算法筛选合理报价，避免单一模型报价的随机性.经过验证，该方法有较高的准确性和效率，报价平均误差约为6.90%.此报价模型降低了传统报价的难度，并提高了其稳定性.

关键词: 注塑模具用零件；价格评估；多专家模型；神经网络

Abstract: General part quotations for injection mould often hold shortcomings such as low efficiency and poor accuracy. Aiming at this issue, a multiexpert quotation framework is proposed based on back propagation (BP) neural network (NN) in this paper. Weights and thresholds of the network are optimized by using generic algorithm which can avoid to fall into a local optimum solution. Meanwhile, an ideal topological structure of NN is decided by studying the Gaussian distribution of errors. A rotational quotation is thereby achieved after synthetical consideration of the evaluating results by synthesizing multiple experts. It is proved that the method greatly eliminates the randomness of a single quotation. Experiments exhibit a preferable performance, in which the average error of quotations is about 6.90%. This scheme reduces the difficulty of traditional quotation and enhances its robustness.

Key words: part of injection mould; price estimation; multiexpert model; neural network

中图分类号:

TP 391

柳伟，杨超. 基于反向传播神经网络的注塑模具用零件报价模型[J]. 上海交通大学学报, 2017, 51(10): 1207-1213.

LIU Wei,YANG Chao. A Study on Injection Mould Part Quotation Model
Based on Back Propagation Neural Network[J]. Journal of Shanghai Jiao Tong University, 2017, 51(10): 1207-1213.

参考文献

［1］刘航. 模具价格估算［M］. 北京: 机械工业出版社, 2015.
［2］NGEL G C, BELN R M, JOS L L, et al. A review of conventional and knowledge based systems for machining price quotation［J］. Journal of Intelligent Manufacturing, 2011, 22(6): 823841.
［3］罗志清, 王润孝, 骞爱荣. 模具产品制造成本与生产周期估计模型研究［J］. 计算机集成制造系统, 2005, 11(12): 16591662.
LUO Zhiqing, WANG Runxiao, QIAN Airong. Research on estimation model of die manufacturing cost & production cycle［J］. Computer Integrated Manufacturing Systems, 2005, 11(12): 16591662.
［4］刘吉祥, 柳玉起, 章志兵, 等. 基于零件特征的汽车覆盖件模具精细报价方法研究［J］. 模具工业, 2016, 42(2): 16.
LIU Jixiang, LIU Yuqi, ZHANG Zhibing, et al. Research on fine die quotation methods based on the features of automobile panel［J］. Die & Mould Industry, 2016, 42(2): 16.
［5］李亨, 王成勇, 肖福成. 基于规则的小型冲压模具报价系统［J］. 合肥工业大学学报(自然科学版), 2009, 32(1): 3639.
LI Heng, WANG Chengyong, XIAO Fucheng. Rulebased cost estimation system of smallsized stamping dies［J］. Journal of Hefei University of Technology (Science), 2009, 31(1): 3639.
［6］LAN Hongbo, DING Yucheng, HONG Jun, et al. Webbased quotation system for stereolithography parts［J］. Computers in Industry, 59(8): 777785.
［7］BOUAZIZ Z, YOUNES J B, ZGHAL A. Methodology of machining cost evaluation for die and mold manufacturing［J］. Journal of Materials Processing Technology, 2004, 152(2):237245.
［8］WU J D, LIU J C. A forecasting system for car fuel consumption using a radial basis function neural network［J］. Expert Systems with Applications, 2012, 39(2): 18831888.
［9］CHE Z H. Psobased backpropagation artificial neural network for product and mold cost estimation of plastic injection molding［J］. Computers & Industrial Engineering, 2010, 58(4): 625637.
［10］DENG S, YEH T H. Using least squares support vector machines for the airframe structures manufacturing cost estimation［J］. International Journal of Production Economics, 2011, 131(2): 701708.
［11］JRGEN B. Neural networks for cost estimation: Simulations and pilot application［J］. International Journal of Production Research, 2000, 38(6): 12311254.
［12］LIU W, HE Y J. Representation and retrieval of 3D CAD models in parts library［J］. The International Journal of Advanced Manufacturing Technology, 2008, 36(910): 950958.
［13］LI Z, ZHOU X H, LIU W. A geometric reasoning approach to hierarchical representation for Brep model retrieval［J］. ComputerAided Design, 2015, 62(5): 190202.
［14］SIMON O H. Neural networks and learning machines ［M］. New Jersey: Prentice Hall, 2008.
［15］单汨源, 於永和. 大规模定制产品多级神经网络成本估算方法研究［J］. 中国机械工程, 2004, 15(11): 10041007.
SHAN Miyuan, YU Yonghe. A study on multilayer neural networks cost evaluation of mass customization product［J］. China Mechanical Engineering, 2004, 15(11): 10041007.
［16］蔺威, 朱玉明, 刘继红. 基于前馈神经网络的汽车覆盖件模具报价系统［J］. 计算机集成制造系统, 2009, 15(11): 22802287.
LIN Wei, ZHU Yuming, LIU Jihong. Auto panel die quotaion system based on back propagation neural network［J］. Computer Integrated Manufacturing Systems, 2009, 15(11): 22802287.
［17］贾超. 基于神经网络的多模型自适应控制方法研究［D］. 北京: 北京科技大学自动化学院, 2017.

[1]	赵子任1, 杜世昌1, 黄德林1, 任斐2, 梁鑫光2. 多工序制造系统暂态阶段产品质量#br# 马尔科夫建模与瓶颈分析[J]. 上海交通大学学报, 2017, 51(10): 1166-1173.
[2]	周鹏辉, 马红占, 陈东萍, 陈梦月, 褚学宁. 基于模糊随机故障模式与影响分析的#br# 产品再设计模块识别[J]. 上海交通大学学报, 2017, 51(10): 1189-1195.
[3]	李昌玺1, 2, 周焰1, 林菡3, 李灵芝1, 郭戈1. 基于MIMOFNN模型的弹道导弹目标#br# 时空序贯融合识别方法[J]. 上海交通大学学报, 2017, 51(9): 1138-.
[4]	冯明月, 何明浩, 韩俊, 郁春来. 基于协方差拟合旋转不变子空间信号参数#br# 估计算法的高分辨到达角估计[J]. 上海交通大学学报, 2017, 51(9): 1145-.
[5]	杨平1，盛杰1，王禹程2，李柱永1，金之俭1，洪智勇1. YBa2Cu3O7δ超导带材非均匀性对失超传播特性的影响[J]. 上海交通大学学报（自然版）, 2017, 51(9): 1090-1096.
[6]	王星, 周一鹏, 田元荣, 陈游, 周东青, 贺继渊. 基于改进遗传算法和SinChirplet原子的调频#br# 雷达信号稀疏分解[J]. 上海交通大学学报, 2017, 51(9): 1124-1130.
[7]	张良俊1, 2, 李晓慈1, 吴静怡1, 蔡爱峰1. 大型空间展开机构微重力环境模拟#br# 悬吊装置热结构耦合分析[J]. 上海交通大学学报, 2017, 51(8): 954-961.
[8]	夏海亮1, 2, 刘亚坤1, 2, 刘全桢3, 刘宝全3, 傅正财1, 2. 长持续时间雷电流分量作用下电极形状#br# 对金属烧蚀特性的影响[J]. 上海交通大学学报, 2017, 51(8): 903-908.
[9]	谷家扬, 谢玉林, 陶延武, 黄祥宏, 吴介. 新型浮式钻井生产储油平台#br# 涡激运动数值模拟及试验研究 [J]. 上海交通大学学报, 2017, 51(7): 878-885.
[10]	林达, 朱益佳, 魏小栋, 王志宇, 张武高. 喷油参数对聚甲氧基二甲醚/柴油发动机燃烧及其#br# 颗粒物排放的影响[J]. 上海交通大学学报, 2017, 51(7): 787-795.
[11]	孟庆阳1, 阎威武1, 胡勇1, 程建林1, 陈世和2, 张曦2. 基于子空间方法的超超临界机组#br# 过热蒸汽系统模型辨识[J]. 上海交通大学学报, 2017, 51(6): 672-678.
[12]	蒋华军a, 蔡艳a, b, 李超豪a, 李芳a, b, 华学明a, b. 基于改进Sobel算法的焊缝X射线图像#br# 气孔识别方法[J]. 上海交通大学学报, 2017, 51(6): 665-671.
[13]	董冠华,殷勤,殷国富,向召伟. 机床结合部耦合动刚度的辨识与建模[J]. 上海交通大学学报（自然版）, 2015, 49(09): 1263-1434.
[14]	谢启江，余海东. 硬岩掘进机刀盘载荷与撑靴接触界面刚度的耦合关系[J]. 上海交通大学学报（自然版）, 2015, 49(09): 1269-1275.
[15]	仲健林1，马大为1，任杰1，李士军2，王旭3. 基于平面应变假设的橡胶圆筒静态受压分析[J]. 上海交通大学学报（自然版）, 2015, 49(09): 1276-1280.