【作者】 苏同义；

【导师】 姜开宇；

【作者基本信息】 大连理工大学 ， 机械设计及理论， 2005， 硕士

【摘要】 浇注型聚氨酯弹性体(CPUE)俗称液体橡胶,它可以在很高的硬度下保持弹性、优良的耐磨性和良好的机械强度,具有化学稳定性、身体适应性和血液相容性等性质,而且,可流动性好,可以制成几何形状复杂的制品。因此,在军事、医学、工程等方面得到了广泛的应用。 但是,其收缩率较大,而且变化较大,尤其对大型(1000mm以上)聚氨酯弹性体而言。因此,分析聚氨酯弹性体产生收缩的原因,研究收缩率同工艺条件和制品结构之间的定性和定量关系,对浇注型聚氨酯弹性体的模具设计具有重要意义。 为了定性分析各个因素对收缩率的影响,本文以正交实验为基础,系统研究了各个因素对大型聚氨酯弹性体制品收缩率的影响,说明了各因素对收缩率的影响趋势,并对其原因做了详细分析。文章指出制品结构对收缩率影响最大,工艺条件对收缩率变化也有较大的影响,而环境条件的波动则是收缩率后期波动的原因。并测出了收缩率随时间的变化曲线。而且,在流动路径长度实验的基础上,本文还研究了流动路径长度对收缩率的影响。 由于正交实验没有考虑制品的形状和浇口的位置,更没有考虑各个因素之间的耦合关系,所以正交实验方法不能对收缩率进行定量预测。因此,本文提出了基于人工神经网络的大型浇注型聚氨酯弹性体制品收缩规律的预测方法。利用人工神经网络能同时处理多变量、通过自我调整权值和阈值就能建立输入和输出之间的定量关系的优势,把人工神经网络应用于大型浇注型聚氨酯弹性体收缩规律的研究上,实现了聚氨酯弹性体制品收缩率的定量预测。 为了使神经网络能对制品的结构进行学习,进而对复杂结构制品的收缩率进行预测,本文初步建立了制品的结构模型,并作为神经网络的一个输入参数。实现了对不同结构制品收缩率的定量预测。 最后,使用未经人工神经网络学习的数据对此收缩率模型的精度进行检验,预测结果和实验数据的对比表明此方法是可行的。更多还原

【Abstract】 Casting polyurethane elastomer (CPUE ) is commonly called as liquid rubber. It can keep the elasticity, fine wearability and good mechanical intensity with very high hardness. It also have such characteristics as chemical stability , health adaptability and blood compatibility ,etc, and it can be made to the products with complicated structure for its good mobility. So it is widely used in military, physic and engineering.However, its shrinkage is big, and varies a lot, especially for large(above 1000mm ) polyurethane elastomer. So it is significant to the mold design of CPUE to analyze the causation for that the CPUE products shrink and to study the qualitative and quantitative relations between shrinkage and every factor.In order to analyze the effect of every factor on the shrinkage in characteristics,this article researches the effects of every factor on the shrinkage of large CPUE products on the basis of experiments. It gives the trend of the effect of every factor on the shrinkage,and analyzes the causation in detail. It points out that product structure and technical conditions have great effect on the shrinkage, and the variety of environment factors causes the subsequent fluctuation of the shrinkage. The curve in that the shrinkage varies with time is provided also. This article researches the effect of the length of flowing path on shrinkage, too.Because not considering the shape of products, the position of injection gate and the relation of every factor, orthogonal experiments can not used to predict the shrinkage quantificationally. So this article presents a method to predict the shrinkage based on artificial neural network. It makes full use of the advantage of artificial neural network to realize the quantitative prediction of the shrinkage.In order to make artificial neural network learn about product structure and predict the shrinkage of compound products, a structure model is put forward as an input parameter of artificial neural network. After learning, it can express the rule for the changes of product structure.In the end, by comparing the result of artificial neural network with the data of experiments, it is proved feasible on the shrinkage prediction.更多还原