【作者】 周猛飞；

【导师】 王树青；

【作者基本信息】 浙江大学 ， 控制科学与工程， 2010， 博士

【摘要】 随着世界石油资源供应日趋紧张,以及原油品质的劣质化、重质化趋势日益明显,重油加工技术已成为提高炼油企业效益和竞争能力的主要手段之一。延迟焦化技术具有原料适应性强、投资和操作费用较低等特点,是炼油企业主要的重油加工工艺。但由于延迟焦化装置的原料来源复杂,各个单元变量之间的关联性强,特别是延迟焦化过程的连续/间歇操作特性,使延迟焦化装置成为最难操作和控制的炼油装置之一。因此,如何在现有工业装置的基础上,保证生产过程的平稳操作,进而通过先进控制技术提高该装置的控制和管理水平,是一项具有重要实际应用价值的研究。本论文的研究内容包括：1.对延迟焦化工业生产过程和延迟焦化常规控制系统进行讨论和分析。重点针对延迟焦化装置的间歇特性展开深入的讨论和分析,主要分析延迟焦化连续装置和间歇装置之间的关系,并从工程应用的角度,提出了一种延迟焦化间歇操作特性的检测模型,该模型可以有效地对延迟焦化装置的间歇操作状态进行实时检测。2.针对延迟焦化装置工况的多变性和扰动的时变性,研究了时不变和时变扰动过程的PID控制器性能评估方法。对于时不变过程的PID控制器,提出了一种双层结构的综合评估方法。该方法对PID控制器的随机性性能和确定性性能两方面进行有效的评估。对于时变扰动过程的PID控制器,提出了一种基于权衡曲线的PID控制器性能评估方法,该方法综合考虑系统受到时变扰动时的全局性能。根据PID控制器的权衡曲线,可在克服暂态扰动和常态扰动之间权衡,选取合适的基准,有效地对控制器进行评估和维护。同时将所提出的方法分别应用于延迟焦化装置的时不变过程和时变扰动过程的PID控制器的性能评估,对评估结果差的控制器进行整定或维护,从而提高常规PID控制系统的性能。3.针对延迟焦化连续/间歇操作特性,提出了一种基于事件驱动的迭代学习预测控制算法。该算法通过间歇操作特性模型对事件的检测,及时触发对重复扰动的估计和学习,从而在MPC结构中增加对扰动的迭代估计,以克服间歇事件的周期性干扰。同时,针对延迟焦化工业过程先进控制系统的实际工业应用,对延迟焦化先进控制目标、控制器结构、约束条件、工程实施等方面进行详细的分析。实际应用结果表明：延迟焦化装置先进控制系统,提高了延迟焦化操作过程的控制品质和稳定性。4.在分析工业生产过程控制系统经济性能评估方法的基础上,将基于性能评估的最小方差引入到多变量模型预测控制软约束调整与经济性能的协调中,从而提高先进控制系统的经济性能。同时将该方法用于评估和提高延迟焦化装置加热炉先进控制系统的经济性能,从而更合理地提高加热炉的热效率。更多还原

【Abstract】 With the increasingly tight supply of oil resources, petrochemical industry nowadays involves more and more heavy oil based applications. Heavy oil upgrading has been an important method for refineries to enhance the economic benefits and competitive capacity. Delayed coking has become a major processing technology for heavy oil and has the advantages of low investment and operating costs, handling different kinds of heavy oil etc.Delayed coking, however, is one of the most difficult units of refinery to operate and control, because of its strongly coupling, the semi-batch nature and its source complexity. Therefore, It is important and of practical significance to ensure smooth operation and improve the performance of control and management through the advanced control technology.The main contents of this thesis are as following:1. The technological process and conventional control system design for delayed coking units are introduced. Specially, Semi-batch nature of delayed coking units is analyzed and the relationship between continuous processes and batch processes is discussed in depth. A practical model for batch operation nature is proposed for on-line monitoring the operation status of batch processes.2. The operating conditions of delayed coking units are time-variant due to the processing different raw oil. Besides, delayed coking process is also subject to time-varying disturbance due to its semi-batch nature. Therefore, the poorly/less optimally tuned PID control performance is usually encountered in the delayed coking industry. The PID performance assessment in delayed coking units is important for maintaining and improving the control performance of PID controllers. For the time-invariant process under PID control, a two-layer structure method is provided, which enables performance analysis of PID controller by integrating both stochastic performance and deterministic performance of PID controller. For the process involving with time-varying disturbance, a PID trade-off curve is proposed, which represents the global optimum for abrupt disturbance and other representative disturbance. According to the trade-off curve, one can choose alterative benchmark to assess the PID control performance. The two developed performance assessment methods are applied to the delayed coking units.3. Focusing on the continuous/batch characteristics of the delayed cooking units, an event-driven iterative learning predictive control (ILMPC) is proposed to the delayed coking units, which is a class of continuous/batch processes characterized by the operations of batch processes generating periodic strong disturbances to the continuous processes. Thus, traditional regulatory controllers are unable to eliminate these periodic disturbances. ILMPC integrates the feature of iterative learning control (ILC) handling repetitive signal and the flexibility of model predictive control (MPC). By on-line monitoring the operation status of batch processes based on the model of batch operation characteristic, an event-driven iterative learning algorithm for batch repetitive disturbances is initiated and the soft constraints are adjusted timely as the feasible region is away from the desired operating zone. Besides, the choice of main controlled variables and manipulated variables, constraint conditions, the structure of controller etc are analyzed. The results of an industrial application show that the proposed ILMPC method is effective for the delayed coking units and improves the stability and enable operations closer to constraints of the delayed coking units.4. The economic assessment of advanced control for industrial processes is introduced. The minimum variance based performance assessment is coordinated into the compromise between the soft constrains adjustment of MPC and the economic performance of the system, in order to enhance the economic performance of advanced control system. The economic assessment is illustrated by an industrial furnace, which is a key unit for the delayed coking process.更多还原