【作者】 何莉；

【导师】 樊希山；

【作者基本信息】 大连理工大学 ， 化学工程， 2003， 硕士

【摘要】 分离序列综合指：给定一进料流股，已知流股状态（流量、温度、压力和组成）系统的化的设计出能从进料中分离出所要求的产品，并使总费用最小。 C₄分离序列由于其组分沸点接近，较难分离。本文结合某厂C₄分离系统节能改造，通过该序列的优化综合，实现生产系统的节能。 本文的主要工作简述如下： 1．利用直观推断法和能量集成的策略，针对原系统拟定三个不同特点的初始节能分离序列。 2．利用化工模拟软件对该系统进行了模拟。由于该分离系统结构较复杂，且各个分离单元的分离任务和分离条件差别较大，本文对不同的分离单元采用了多个热力学方程和多种方法的尝试，最终通过模拟结果与现场数据的吻合，确认各分离单元的热力学方程。 3．应用确认的热力学模型，对初始方案进行模拟，并利用灵敏度分析和轮换变量法对初始方案中分离序列中关键操作参数进行讨论和优化，确定优化参数。结果表明三个用能优化序列均可完成分离任务，且具有一定的节能效果。 4．上述节能序列均从该系统内一子系统用能情况来考虑。但若从全系统考虑，仍然存在较大的节能潜力，后续工段中的一乙腈回收塔所分离物系物性稳定，有利于提压操作，提压后塔顶蒸气可以作为其它塔底再沸器热源，同时塔底釜液热量也要进行回收。通过全系统热集成可显著提高以上序列的用能完善程度。 5．对BP人工神经网络的设计参数和训练过程进行分析，并结合任务分解法的思想，利用BP人工神经网络对其中一用能优化序列建立黑箱数学模型。神经网络模型预测结果与ASPEN模拟结果的误差表明该模型能够满足模拟该过程的要求。 6．分析遗传算法的机理，用MATLAB建立改进的遗传算法并通过测试函数的测试。用该算法对人工神经网络所建立的数学模型进行优化。比较遗传算法和轮换变量法优化结果，表明神经网络和遗传算法相结合能够对分离序列进行优化，对工程应用具有一定的指导作用。更多还原

【Abstract】 Separation process synthesis is a manipulation to deal with mixtures to be separated at the low expense of energy consumption and capital cost. And it is often used in the pretreatment of raw material, the purification of the product, the disposal of scrap, etc.Due to the close boiling points, C4 hydrocarbons are difficult to be separated. In this paper, combining with the refit plan in a factory, the energy efficiency of the whole separation system is increased through synthesis and optimization of the separation sequence.The six points are summarized as follows:1. The draft is designed by means of the tool of heuristic rules and technique of energy integration.2. The C4 hydrocarbons separation system is simulated using the simulator ASPEN PLUS. For each unit, the thermodynamics function is confirmed.3. Based on the simulation results and energy analysis of the system, the plans are simulated with the established thermodynamics functions, and the operation parameters are optimized by the sensitivity analysis. Calculation shows that all of the proposed plans are applicable to effectively separated the mixtures at lower expense of energy consumption.4. The acetoniltrile recovery column is considered to increase the operation pressure. The energy integration plan is designed to further increase the energy efficiency of the whole system.5. Mathematical model of the thermally coupled distillation is established with artificial neural network. Calculation shows that the model can simulate the process rigorously.6. A modified genetic algorithm is used to optimize the model by artificial neural network. It shows that the separation process can be optimized with the established model and optimization algorithm.更多还原