【作者】 陈军；

【导师】 赵众；

【作者基本信息】 北京化工大学 ， 控制科学与工程， 2022， 硕士

【摘要】 随着市场竞争的日益加剧以及环境保护要求日益提高,化工企业的生产理念也在发生着巨大的变化。过去,装置的平稳运行可能是企业的第一关注点,从而一味地抓产量而牺牲了能耗和环境。现如今,企业都在不断的从各个方面提升自己的综合竞争力,对节能降耗、低碳减排等有了新的认识。生产过程的操作优化是一种提升企业竞争力的有效手段,实施操作优化的装置能够在面对干扰和其它时变特性下始终保持系统的运行状态接近最佳运行目标。传统的生产过程的操作优化往往假设调优变量与目标函数之间存在单一的静态关系,并且也没有考虑过程中存在的其它干扰,导致对干扰十分敏感且难以处理动态过程的实时优化。相关积分优化方法,重构了生产过程稳态优化的提法,将时间项和干扰项考虑进去,使得算法具有很强的抗干扰能力。同时,基于数据驱动,利用优化变量以及目标函数变量的实时数据工作,无需建立精确地机理模型。然而,传统的相关积分优化方法存在着对不同工况的适应性较差、没有考虑过程中存在的实时约束等局限性。本文针对传统相关积分优化方法存在的局限性,将相关积分与预测控制相结合,采用控制方法来解决实时优化问题。针对传统算法没有考虑实时约束以及对不同工况的适应性较差等局限性,采用鲁棒预测控制方法来对表征系统是否还有优化裕度的中间变量梯度进行控制。针对线性预测控制在强非线性系统控制效果变差的缺点,采用Volterra非线性模型预测控制对梯度进行控制。最后,对改进的相关积分优化算法进行了仿真验证,并开发了相关软件,在中石化某炼厂二甲苯加热炉热效率调优过程中进行了工业应用测试,验证了改进算法的可行性和有效性。更多还原

【Abstract】 With the increasing competition in the market and the increasing requirements for environmental protection,the production philosophy of chemical companies is also undergoing great changes.In the past,the smooth operation of the plant may be the first concern of the enterprise,thus focusing on the production at the expense of energy consumption and environment.Nowadays,enterprises are constantly improving their comprehensive competitiveness from all aspects,and have a new understanding of energy saving,low carbon emission reduction,etc.Operational optimization of production processes is an effective means to improve the competitiveness of enterprises,and the implementation of operational optimization can keep the system operating close to the optimal operating target in the face of disturbances and other time-varying characteristics.Traditional process optimization often assumes a single static relationship between the tuning variables and the objective function,and does not take into account other disturbances in the process,making it very sensitive to disturbances and difficult to handle real-time optimization of dynamic processes.The correlation integral optimization method,which reconstructs the formulation of steady-state optimization of production processes,takes into account the time and disturbance terms,making the algorithm highly resistant to disturbances.At the same time,it is data-driven and works with real-time data of the optimization variables as well as the objective function variables,without the need to build an exact mechanistic model.However,the traditional correlation integral optimization method has limitations such as poor adaptability to different working conditions and not considering the real-time constraints existing in the process.This paper addresses the limitations of the traditional correlation integral optimization method by combining the correlation integral with predictive control and using a control method to solve the real-time optimization problem.To address the limitations of the traditional algorithm that does not consider real-time constraints and poor adaptability to different operating conditions,robust predictive control is used to control the gradient of intermediate variables that characterize whether the system still has optimization margin.To address the drawback that linear predictive control becomes less effective in strongly nonlinear systems,Volterra nonlinear model predictive control is used to control the gradient.Finally,the improved correlated integral optimization algorithm is validated by simulation,and the related software is developed and tested for industrial application in the process of thermal efficiency tuning of a xylene heater in a Sinopec refinery to verify the feasibility and effectiveness of the improved algorithm.更多还原