【作者】 张海涛；

【导师】 王洪艳；

【作者基本信息】 吉林大学 ， 应用化学， 2010， 博士

【摘要】 耐水性是水性涂膜研发及大规模生产环节中重要的性能指标。本论文主要研究了聚丙烯酸酯水性涂膜及涂料体系的耐水性建模问题。利用人工神经网络、分子动力学、数理统计多元回归等手段,对材料的耐水性进行了预测与分析,为提高水性涂膜材料的耐水性提供有利的支持。主要内容如下:(一)利用人工神经网络多目标同时预测的方法,从涂料体系配方中的三种单体(丙烯酸丁酯、甲基丙烯酸甲酯和苯乙烯)及两种颜料(二氧化钛和碳酸钙)的用量出发,建立对涂料的硬度、附着力、耐冲击性和反射率四种性质的预测模型。经过数据的预处理及对网络参数的优化,得到了98%的总体预测准确率。(二)从分子动力学模拟出发,通过计算涂膜聚合物分子与水分子间及界面间的相互作用,建立了能量与饱和吸水率之间的拟和关系;考察了聚合物内部氢键作用对水分子在内部的扩散系数的影响;以扩散系数、水性官能度及其相互作用三个因素建立起预测聚丙烯酸酯水性膜材料饱和吸水率的模型,其预测误差约5%,为后续的多尺度建模研究提供了参数。(三)研究了乳液与颜料粒子之间的相互作用。通过实验测定接触角与分子模拟得到的相互作用能量比较,找到了它们之间的对应关系,为涂膜多尺度建模中含有颜料的部分提供相应的数值基础。(四)建立了涂膜耐水性的多尺度关联模型,提出平衡时间tB和渗透平均速率SV两个参数用于表征膜材料耐水性。考察了各种宏观因素及微观变量对材料耐水性的影响,与实验的结果一致,证明了模型的可行性及可靠性,为在特定环境下使用的膜材料的生产工艺及选用提供指导依据。(五)将之前考虑的各部分内容综合总结,建立可视化界面系统,为本研究在更大范围内使用提供有力支持。更多还原

【Abstract】 Paper mainly discussed the properties of polyacrylate waterborne coating and the water resistance of it. The property of water resistance is a key index of the application of waterborne coating. Materials which had poor water resistances, the water molecules in the environment were easily eroded them, and then changed the internal structures, and change some other properties in further. In order to improve coatings water resistances, it was needed to modify the materials in many aspects. However, the method which was most used was still single factor rotation. Because there were so many factors would affect the water resistance, it was a very complicated work for choosing a better formula. Developing a method for predicting the water-resistant of coating quickly and accurately It became an important task for developing a quickly and accurately method to predict water-resistant of waterborne coatings. In the paper, artificial neural networks, molecular dynamics, multi-scale modeling and other methods were used for studying the water resistance of coating.In the second chapter, by the method of artificial neural network simultaneous prediction, the properties of coatings were predicted by the formula. The network model established which includes five input nodes and four output nodes. The five input nodes corresponding to the amounts of three kinds of coating monomers (butyl acrylate, methyl methacrylate and styrene) and two kinds of pigments (TiO2 and CaCO3). The four output nodes corresponding to the four types of coating properties, which were hardness, adhesion, impact resistance and reflectivity. After data preprocessing, the choice of hidden layer nodes, options of the hidden layers number, learning rate and transfer function and other works, the optimal network structure were confirmed. Base on the structure of network, the weights and thresholds values were trained by the experiments results. Then 9 samples’properties were predicted by the network and compared with the measured values. The final prediction accuracies of four properties were calculated. Adhesion and impact resistance have the perfect results. The errors of reflectivity and hardness were 0.16% and 8.02% respectively. The results showed that, BP neural network prediction was a feasible method for predicting the coatings’properties. It could be used to guide the experiments and productions.In Chapter III, saturated water absorption which was a key index in the study of coatings was studied by experiments and simulations together. The discussion was focused on the affect materials ratio on saturated water absorption, and the binding energies between water molecules and materials were calculated. The relationship between saturated water absorption and binding energies was confirmed. At the same time, the hydrogen bonds energies of water molecules in the materials were calculated. And the diffusion coefficients of water molecules which were affected by the hydrogen bonds energies were simulated by the method of molecular dynamics (MD). The results confirmed that the diffusion coefficients were affected by the monomer ratio closely. Increasing the ratio of hydrophilic monomers, the water molecules diffusion coefficient would decrease, and the saturated water absorption would increase. Because of this relationship, a parameter of aqueous functionality of system (φ) was proposed which was stood for the proportion of hydrophilic groups in the total amount of monomer. Then a prediction model of saturated water absorption was built up by diffusion coefficient, aqueous functionality and theirs interaction. The average error of this model was about 5%. It proved that the model could be used to predict the saturated water absorption which would be used in the final multi-scale model of membrane water resistance.In Chapter IV, pigments which played an important role in the coatings were discussed. The wetting effects between kinds of pigments and latexes were considered. By the experiments, the contact angles of pigments and latexes were measured. Then the energies of interfaces between pigments and latexes were calculated by MD method. By the comparison, the relationship between two energies was founded. So the water resistance indexes of pigments were defined which were stood for the properties of pigments. And the indexes were used in Chapter V and Chapter VI.In Chapter V, a multi-scale modeling of water resistance was established base on Fick’s second law for measuring the water resistance of membranes quantitatively. Two indexes, balance time (tB) and seepage velocity (SV) were proposed for the characterization of membrane water resistances. Macro-scale factors such as such as the thickness of membranes (H), the ambient temperature (T) and monomer ratio were picked out for discussion. Micro-scale variables, such as water diffusion coefficient (D), membrane saturation water absorption (CB0) and of water molecules escape concentration (CB1) which were affected by the macro-scale factors and influenced the water resistance indexes such as tB and SV. The agreement between the experimental results and calculated results proved the feasibility and reliability. The two indexes could be used for measuring the membranes water resistances, and could guide the producing in further.In Chapter VI, a visual system was programmed combing the discussion in Chapter III, IV and V. By the operations of click and select, the water resistance indexes of coatings which had different monomers ratios, different macro-scales parameters, different kinds and amount of pigments could be calculated quickly. Researchers could use this software design the better coatings with higher water resistance.更多还原