【作者】 卢康；

【导师】 欧长劲；

【作者基本信息】 浙江工业大学 ， 机械制造及自动化， 2007， 硕士

【摘要】 气辅成型是近几年发展起来的一种新型塑料加工技术，具有可提高产品表面质量、增加制品强度、减少翘曲、降低成本等优点。由于成型过程增加了气体因素，使其工艺较传统注塑更为复杂。如何筛选出对成型过程有重要影响的工艺参数并进行优化配置以获得满意的制品质量是目前该领域的一大难题。根据目前国内外气辅成型工艺优化的研究现状，本文提出基于神经网络结合遗传算法(NN-GA)的智能多目标优化方法，用于工艺参数最优组合选取的求解，为成型过程工艺参数的优化提供新的解题思路。本文主要研究工作如下：1、综合考察及评价多个工艺参数对成型制品质量的影响，归纳总结气辅制品常见缺陷产生的机理并探讨相应的解决方法。2、利用基于试验统计分析结合数值模拟的方法进行工艺参数初步优化。选取成型过程中影响较大的几个工艺参数为优化对象，选定气辅制品的几个质量指标作为考察目标，通过安排正交试验得出各个工艺参数对质量指标的影响程度，并分析求得一组较优工艺参数组合。3、建立气辅成型工艺参数与制品质量指标关系的RBF神经网络模型。以正交试验得到的较优参数组合值为依据，按均匀设计思想选取学习样本数据训练网络。构建的RBF网络模型用于遗传算法优化过程中目标函数值的快速求解。4、选择NSGA-Ⅱ多目标遗传算法策略，结合建立好的RBF神经网络模型，应用遗传算法进行多目标迭代优化获得工艺参数的Pareto最优解集并实现其组合值的最优化设定，最后通过数值模拟表明了本文基于NN-GA智能多目标优化方法用于气辅成型工艺参数优化的可行性及有效性。更多还原

【Abstract】 Gas-assisted injection molding (GAIM) is a new plastic processtechnology developed these years which has the advantages of improvingproducts surface quality, enhancing intensity, reducing warp and cost.Influenced by the additional factor of gas in molding process, thetechnology of GAIM is more complicated than traditional injectionmolding. At present, it is still a problem that how to screen out the keytechnological parameters and do some optimization to make the producthas good performance. By the research actuality of GAIM technologicalparameter in the world, this thesis presented an intelligent multi-objectiveoptimization method which based on neural network and genetic algorithm(NN-GA) to optimize and select the best technological parametercombination, then provided a new solution of technological parameteroptimization for GAIM.The main content of this thesis is listed as follows:1. Study and appraise the influences of several technological parameters to product quality performance synthetically, generalize thecommon flaw generation mechanism of GAIM product and discuss therelevant solutions.2. Make the preliminary optimization of technological parametersbased on experimentation design and analysis combined with CAEsimulation methods. Select the key technological parameters of moldingprocess to be the optimization objects and define the several qualityperformances of GAIM product to be the judge targets, by arrangingorthogonal experimental design method, find out the influence degree oftechnological parameters to quality performance and solve out a set ofbetter technological parameter combination.3. Build the relationship model of GAIM technological parametersand product quality performance by using RBF neural network. Accordingto the better technological parameters combination value which got byorthogonal experimental method, generate the training sample with thethinking of uniform design then finish the training of neural network. TheRBF network model can be used to make fast calculation of targetsfunction values during optimization process by genetic algorithm.4. Choose the NSGA-Ⅱmulti-objective genetic algorithm to be theoptimization algorithm strategy of this thesis, according to the RBF neuralnetwork model has been built, Make multi-objective optimization to get thePareto optimal set by using genetic algorithm, then realize the optimization of technological parameters finally. At last, by the comparison of CAEsimulation, the NN-GA-based intelligent multi-objective optimizationmethod of this thesis showed to be feasible and effective.更多还原