The Application Of Genetic Algorithms To Computer-aided Design Of Petlyuk Distillation Column

The traditional energy dries up day by day, while the technology of utilizing the new energy is not mature; People's consciousness of environmental protection strengthens day by day, and our environment can benefit from the energy conservation; it is much higher than the world average level in the same output value versus energy consumption in our country. Under such background, the researchin energy conservation appears extremely important.There is an active research direction, process intensification in chemical engineering, with the goal, energy conservation. It is a large proportion of energyconsumption by the distillation process to the whole chemical industry. Therefore, the practice of process intensification in distillation process is very important to energy conservation in chemical industry.The Petlyuk distillation is one kind of process intensification technology. It both conserves energy and saves the equipment investment, but also has other merits, thus it has the extremely attractive prospects for development. Therefore, it attracts many eyes from academic and industry circles. Although the Petlyuk column has many attractive merits, but the lack of one kind of reliable design technology, as well as the immaturity of control technology, for this kind of column, causes it applys widely with difficulty.The design of Petlyuk column involves a very complex optimization. This dissertation performed two times transform of problem to reduce the complexity, but the problem is still complex. The constraint is extremely complex. The variables which are to be optimized are of different data types; moreover, there are complex relations between them. The combination optimization and the function optimization interweave in the same problem, simultaneously, it also is the multiobjective optimization. The computation of individual fitness is time-consuming. Considering the genetic algorithms has huge potential in the constraint optimization, the combination optimization, the function optimization as well as the multiobjective optimization, this subject attempts to solve optimization problem in the Petlyuk column design with the genetic algorithms. When computing individual fitness, this dissertation uses Aspen plus(?), the well-known software in the domain of chemical engineering process simulation, to solve the distillation model of the Petlyuk column.According to the characteristic of the problem, this dissertation adopts the most natural encoding, introduces individual sex in a new way to GAs to enforce the balance in individual evolution concerning different objectives, the population size is dynamic, and gene linkage is utilized when recombines. Modifications of the tournament selection, roulette wheel selection and stochastic universal sampling selection are practiced to adapt the situation where explicit sex definition is made and the multiobjective is to be handled. The operation that control the proportion of male and female individuals and the operation that control the population size are introduced to genetic algorithms to meet the need of the modification to the standard genetic algorithms.The time-comsuming computation of individual fitness brings much inconvenience to the debug of program and the test of algorithms. Therefore, only a few of result data is available in this dissertation.