【作者】 吴莉莉；

【导师】 管国锋；

【作者基本信息】 南京工业大学 ， 化学工程， 2006， 硕士

【摘要】 针对近沸程或共沸物体系萃取精馏萃取剂的设计问题,本文提出了一种新的分子设计策略——计算机优化设计方法,主要分为两个部分:计算机辅助分子设计方法和计算机优化筛选方法。以基团贡献法为基础,采用计算机辅助分子设计方法中的遗传算法设计了链状、环状和单环芳烃三类萃取剂;并对设计出的萃取剂采用计算机优化筛选方法进行模糊综合评价。通过气相色谱法测定水-醋酸体系在较优萃取剂中的无限稀释活度系数,将计算得到的无限稀释相对挥发度与本文预测值相比较,验证设计方法的准确性。针对遗传算法搜索效率低和不易得到全局最优解等问题,本文设计了一套新的分子设计方法,整个设计的核心为:编码方案的确定以及遗传算法操作算子的改进。采用基团的子基团号组合直接表示分子的新编码方案,生成的分子形式简单,且基团的各种参数均通过子基团号链接,大大增加了算法的执行效率。采用跨代精英选择方法,克服了传统选择算子选择机制产生的随机误差。提出错位交叉算子,避免了交叉产生个体等位基因上的值相同或者相近,大大增强了种群的多样性。采用突变算子,在种群进化后期增加对种群的扰动,防止提前收敛。为了全面评价设计结果,本文建立了模糊综合评价的优化筛选方法,得到包含两组权重分配集,并考虑选择性、溶解性、粘度、比热、相对分子量、毒性、腐蚀性和沸点等性质的综合评价模型,使评价结果更加准确可靠。为了考察本文设计方法以及所编程序的准确性,对已有文献报道的丙酮-甲醇、乙醇-乙酸乙酯体系的萃取精馏萃取剂进行了验证设计。对于丙酮-甲醇体系,较好的萃取剂有水、乙二醇以及二甲亚砜等;对于乙醇-乙酸乙酯体系,较好的萃取剂有三甲苯、丙三醇和氯苯等。设计结果、评价结果与文献报道基本一致,且预测精度高于文献预测精度。对未见文献报道的水-醋酸体系,设计和优化筛选出了N-甲基吡咯烷酮、N-甲基乙酰胺、二甲亚砜和2-吡咯烷酮等萃取剂。为了验证设计结果的准确性,采用气相色谱法测定了水和醋酸在N-甲基吡咯烷酮和N-甲基乙酰胺中的无限稀释活度系数,将计算得到的无限稀释相对挥发度与本文预测值比较,相对误差小于10%,从而表明本文提出的计算机优化设计方法,即与Mod.UNIFAC基团贡献法更多还原

【Abstract】 A novel strategy for designing extractants by computer optimization design method was presented. It was composed with the computer-aided molecular design and computer optimization screening method. The chain compounds, compounds ring and mononuclear aromatics were designed by combining Mod.UNIFAC group contribution method with modified genetic algorithm. And the obtained extractants were estimated by the computer optimization screening method. The accuracy of the method was verified by determining the infinite dilution activity coefficient of the water + acetic acid system in the superior extractants by gas-liquid chromatography method.A method based on the improved genetic algorithm for the computer-aided molecular design for extractive distillation was presented. A number of enhancements had been comprised. In order to greatly enhance the algorithm performing efficiency, a new encoding scheme was developed. Groups were represented by their own group number, and the molecular structure was described by string figures. To prevent the random errors of selecting molecule by roulette wheel rules and losing good solutions in generation reproduction, a cross generation elitist selection method was used. Dislocation crossover was developed to make molecules keep good diversity, and overcame the unfavorable defect of traditional crossovers which caused GA get into prematurely. In this work, mutation operator was just used if the algorithm almost achieved constringency stage. Here mutation was used to disturb the population and do make improvement.In order to estimate the obtained extractants comprehensively, a model for the comprehensive judgment of extractants in which eight factors including selectivity, solvent power, boiling point, molecular weight, viscosity, specific heat capacity, toxicity and corrosion were considered. And two groups of weights were allocated to assessment factors, and the weights were chosen by the user to suit the specific problem.The extractants of acetone+ methanol system and ethanol+ ethyl acetate system更多还原