【作者】 张学岗；

【导师】 宋海华；

【作者基本信息】 天津大学 ， 化工过程机械， 2008， 博士

【摘要】 特殊精馏与液液萃取是分离液态难分混合物的重要方法,而质量分离剂的选择是实施分离的关键问题。本文就质量分离剂的选择展开研究,以Mod. UNIFAC基团贡献法和改进的遗传模拟退火混合算法(MGASA)为基础,提出了一种新的计算机辅助分子设计方法,解决了CAMD中的若干难点问题,具有重要的理论和应用价值。化工分离过程质量分离剂的计算机辅助分子设计方法(CAMD),一般是以基团贡献法为基础,先预选一定的基团,然后按照某种规则将基团组合成分子,并依据指定目标性质进行筛选。目前CAMD中存在两个难点,一是数学优化方法,即如何解决基团组合爆炸问题;二是分子合成方法,即如何避免基团组合过程中大量结构不完整、性质不稳定分子的生成问题。本文就上述两个方面提出了新的解决方法和比较完整的解决方案。对于数学优化方法,现行的CAMD方法中已采取了一些优化算法来避免组合爆炸问题,如遗传算法(GA)、模拟退火算法(SA)及禁忌搜索算法(TS),但上述方法均有缺陷,运行参数的选择直接影响到算法的运行结果。本文采用了遗传算法与模拟退火算法混合的MGASA算法,弥补了单种算法的缺点,并结合分子设计的特点,对GA中的交叉、变异算子及SA中的新状态产生函数进行了改进,提出了基于知识的交叉因子,增加了不合法分子的修复算子,及防止最优解丢失的记忆算子。改进后的MGASA算法具有GA算法的优化时间性能和SA算法最终趋于全局最优的优点,克服了GA算法“过早收敛”问题和SA算法优化时间性能较差的缺点,是一种并行而且具有自动保优功能的算法;MGASA算法利用GA和SA各自不同的邻域搜索结构相结合,使得算法在解空间中的搜索能力所增强,优化效率得到提高;MGASA算法的多点搜索削弱了SA算法对初值的依赖性,同时利用GA算法不影响平稳分布的特性,提高了算法的鲁棒性。编码策略是算法实施的最基础问题,也是最重要的问题,编码策略中最大的难点在于直链烃、环烷烃与芳香烃的同时编码,虽然经过长时间的研究,目前国内外还没有形成很好的解决方案,本文提出了基于分子组成的基团数目的十进制编码策略,具有以下优点:分子表达简单,编码、解码操作简单易行;实现了直链烃、环烷烃及芳香烃的同时编码,同时搜索,保证了搜索空间的完整性;取消了其它设计方法中组成分子的基团数不能大于8的限制,削弱了基团总数的限制,拓展了搜索空间;消除了解空间的冗余,节省了计算时间。关于分子合成方法,基团组合成分子应满足完整性要求和稳定性要求。本文完善了分子完整性定理,并以此为依据,对不符合分子完整性要求的分子进行合法性修复;本文通过采取新的基团分类方法,可以比较容易地判断分子的稳定性和完整性;采取将分子稳定性作为适应度评价函数之一的策略,避免了分子完整性和稳定性同时修复带来的难度,减少了计算时间的消耗。本文将基于MGASA的CAMD方法应用于萃取精馏、共沸精馏及液液萃取过程,完善了溶剂选择的要求及相应适应度函数的计算,采用Matlab编程形成了一套通用的计算软件。采用实际物系对本文提出的方法和形成的软件进行了检验,计算过程的追踪表明算法过程为随机优化过程,总体趋于优化;将筛选结果与文献对比,表明本文筛选出来的溶剂适应度在本文的评判标准下普遍高于文献中溶剂,为工程应用提供了更好的选择,由此表明本文提出的质量分离剂的计算机辅助分子设计方法具有很好的准确性和实用性。更多还原

【Abstract】 Special distillation and liquid-liquid extraction are important ways to deal with liquid mixtures which are difficult to be separated. Selection of solvents is the key issue. This paper is focus on the selection of solvents. A novel general method of Computer aided molecular design based on improved MGASA hybrid algorithm, combined with Mod. UNIFAC Group Contribution method is put forward, and used to select solvents for extraction distillation, azeotropic distillation and liquid-liquid extraction.CAMD of chemical separation solvents is generally based on group contribution method. Some groups with certain structure are selected firstly, and then combined with one another to form a molecule according to certain rules. The one of the two difficult issues is mathematic optimization method, designed to deal with the problem of‘combination explosion’. The other is the method of molecular compounding, which has to be thought to avoid production of rubbish molecules.On the issue of mathematic optimization, though some optimization algorithm, such as Genetic algorithm, Taboo algorithm and simulated annealing algorithm are adopted in current CAMD, some disadvantages exist and the results relies on the running parameters firmly. Improved MGASA hybrid algorithm is put forward in this paper, which combined with the advantages of GA and SA, avoiding their disadvantages. Some operations in GA and SA are improved according to the need of molecular design.Firstly, operations including crossover, mutation and new state producing which are based knowledge is founded. Secondly, recover operation is added to MGASA to reduce the amount of illegal molecule which produced in general MGASA operations. Thirdly memory operation is added to avoid the optimization result lost and reduce the degree reling on convergence.This paper puts forward a novel code strategy, which coding the number of groups compounding a legal molecule in binary system.Here called group number binary cluster for short. This coding strategy has several advantages.Molecule can be coded simply. Including and decoding operations can be carried out easily. Paraffin, cycloparaffin and aromatic groups are coded together and searched synchronously in this coding strategy.This strategy guarantee the complete search for solves because of getting rid of the limits on the number of each kind of groups and the category of groups. Constracting with former strategies, this strategy extend the searching field.While it saves computing cost with omitting the redundancy.On the issue of molecule compounding, legal molecule should satisfy the demand of molecular integrality and chemical stability. About molecular integrality, Yangzhensheng has put forward a theorem. This paper amends the theorem and recovers the illegal molecule according to the amended theory. Chemical stability is put in fitness function.This reduces the complexity of recovering illegal molecule according to molecular integrality and chemical stability synchronously. Computing cost is saved at the same time.MGASA method is applied for separation solvents of extraction distillation, azeotropic distillation and liquid liquid extraction. Rules for adaptable solvents and fitness functions are amended. Matlab programs for searching solvents for the above three kinds of processes are compiled. And the results of application on practical systems show excellent searching ability. Some useful solvents are founded besides those mentioned in literature. In a word, the CAMD method put forward in this paper has high level of rightness and can be used in practical chemical processes.更多还原