摘 要稠油是一类较为特殊的石油资源，在全世界的储量不是很多，在位于新疆克拉玛依地区的油田中，先后发现多个大的油田，发现含有大量的稠油资源含量，已探明相应稠油资源量达数亿吨，这为稠油常减压蒸馏装置的原料供给提供了充足的来源。 原油常减压蒸馏工艺是石油化工生产中最常见、最重要的一种分离工艺，常减压蒸馏装置在大型炼油厂及大型石油化工厂一般被称为是龙头装置，而常减压蒸馏塔则是实现蒸馏工艺过程的核心装置设备,其平稳、高效的操作水平是可以直接提高全厂的经济效益,并且其运行不仅对常减压蒸馏装置十分重要，对下游装置的高水平运行也有重要的影响,在提高整个企业的经济效益方面是有着非常重大的作用。本文采用流程模拟软件对中石油克拉玛依石化公司的一套稠油常减压蒸馏装置开展了流程建模的研究工作，通过科学合理的流程模拟和采用神经网络技术，运用Aspen Plus流程模拟软件和基于Matlab数学计算软件的RBF神经网络技术，通过运用稠油常减压蒸馏装置的各项实际运行参数和装置所加工原油及各侧线性质的化验分析数据，使用用上述软件工具创建稠油常减压蒸馏装置的流程模型及其质量预测模型。采用了三种通用的适用于炼油过程的物性方法，并从中优选出最适合运用于稠油常减压蒸馏的物性计算方法。在创建好相应的流程模拟模型的基础，采取在 Aspen Plus 软件中集成的相关分析工具进行分析，并在此基础上，得到了稠油常减压装置的不同操作参数的变化对装置的实际运行效果变化的影响结果。分析了RBF神经网络技术在常减压蒸馏装置上的应用情景，同时采用遗传进化算法来对其进行优化。在工艺流程和原油蒸馏原理的分析基础上,讨论了影响侧线产品质量的多种不同的因素,从而使用遗传算法优化的RBF神经网络，通用分别建立减压四线的粘度和闪点两个不同的预测模型来预测装置运行所带来的质量变化情况。最后，对装置的运行提出了一个可用于改进的方案：针对性的提高减四线拔出率的一种改进方案。
Heavy oil is a special type of oil resources in the world's reserves are not many, in the Xinjiang area of Karamay oilfield, has found a large oil field, was found to contain large content of heavy oil resources, has proven the heavy oil resources amounted to hundreds of millions of tons of heavy oil, the atmospheric vacuum distillation unit of raw materials supply the adequacy of the source.Crude oil distillation process is the most common in the petroleum chemical industry production, a separation process is the most important, atmospheric and vacuum distillation unit in large refinery and a large petrochemical plant is generally known as the leading device, and vacuum distillation tower is the core to realize distillation process equipment, the stable and efficient the operation level is to improve the economic benefits of the whole plant directly, and the operation is not only in the atmospheric vacuum distillation unit is very important, high level operation of the downstream device also has an important influence, has a very important significance in improving the econmic efficiency of the whole enteprise.Through the simulation and research of PetroChina Karamay Petrochemical Co., Ltd. oil atmospheric vacuum distillation unit, through scientific and reasonable process simulation and using neural network technology, using the Aspen Plus process simulation software and software based on Matlab mathematics RBF neural network technology in the heavy oil of atmospheric and vacuum distillation unit for the calibration data based on Simulation the quality and prediction of atmospheric and vacuum distillation unit model.Three recommended for property method in the refining process of choice, comparative analysis of the simulation results of different methods. The physical method to selct the suitable. At the same time in the Aspen Plus based on the analysis of sensitivity analysis tools, the influence of variation of different operating parameters that decompression device operation effect of the device changes.Analysis of the application situation of RBF neural network technology in the atmospheric vacuum distillation unit, while using genetic algorith to optimize RBF neural network. Based on analyzing the process and principle of crude oil distillation, the various factors affecting the quality of sideline products different discussion, thus using the RBF neural network optimized by genetic algorithm, the general was established decompression four line viscosity and flash point of two different prediction models to predict the quality changes brought about by the operation of the device.Finally, an improved scheme is proposed to improve the operation of the device: an improved scheme for improving the four line pull out rate
