【作者】 姜景杰；

【导师】 潘立登；

【作者基本信息】 北京化工大学 ， 控制理论与控制工程， 2007， 博士

【摘要】 苯乙烯(SM)是一种重要的基本有机化工原料，用途十分广泛。从总体上看，世界苯乙烯的生产能力已经出现了过剩的态势，而亚洲尤其是中国的消费量却不断增加，这样势必导致更多的外国产品流入我国，对我国苯乙烯行业今后的发展造成一定的威胁。因此应尽快采用先进技术对现有装置进行技术改造，逐步提高装置的生产规模和工艺技术水平，以增加产量，降低生产成本，增强我国苯乙烯在国内外市场中的竞争力。为了提高苯乙烯装置的经济效益，很多企业购买了先进控制软件，这些先进控制软件在投入使用初期效果明显，但能长期投入运行的却是寥寥无几，分析其原因，这些先进控制软件的鲁棒性有限，技术维护复杂，随着装置的运行，模型发生变化，而先进控制软件所使用的模型不能在线及时更新或难于更新，结果不得不放弃这些先进控制软件。如何解决先进控制软件技术维护复杂，不能长期投运的问题，是本文研究的重点。要保证基于模型的先进控制长期高效的运行，最基本的一点是保证对象模型辨识的准确性，同时为减少技术维护的复杂性，就要研究和解决如何以一种简单有效的方法进行对象辨识和优化求解问题。为此本文提出了在对象模型辨识基础上的基于内部模型控制(IMC)原理的IMC-PID和模型PID先进控制技术。实施基于模型的先进控制最关键的一点是获取准确的对象模型，获取对象模型最简单的方法是获取对象的动态响应数据，然后使用优化方法求取对象参数。为了以简单高效的方法获取对象模型，本文采用OPC／DDE数据通信技术进行数据采集，并向回路施加测试信号以获取回路动态特性；为了减少算法的复杂性，文中使用NLJ算法求取辨识参数。在对象辨识方法的研究中，针对工业装置一些大干扰或滞后时间较大的而造成过渡时间过长，很难达到稳态的系统，提出了一种新的闭环对象辩识方法——对象参数辨识与系统状态方程初值同时辨识的新方法(Model Parameters and State Equation Initial Values Identified simultaneously----简称MPSEIVI)；对于纯滞后较大对象的求解出现辨识对象误差问题，提出了新的纯滞后对象的求解算法——M-Pade法。为了确保先进控制的长期运行，减少投资和充分挖掘原有控制系统的潜能，本文提出了基于常规PID的模型PID(MOD-PID)技术，该技术使用原来的PID控制器，因此不需要对原来的控制系统进行改造，只要能辨识出对象模型，就可求得PID参数。由PID的控制性能可知，对于一些较复杂的回路，PID效果并不是很好。针对这些回路，基于IMC控制原理，前人提出了基于IMC控制的IMC-PID控制。本文在应用过程中发现，在应用IMC控制器求取IMC-PID控制器参数时，由于其推导过程相当繁杂，并且对不同的对象类型、控制器类型、纯滞后类型都要进行公式推导，而最终得到结果，有可能是多重解，这样就很难判断哪一个解最优。为了解决这个问题，基于阶跃响应相同的原理，本文提出了由IMC控制到IMC-PID控制的新解法。IMC-PID控制需要对原来的PID控制器进行改造，但有些控制系统无法实施IMC-PID改造，针对无法实施IMC-PID改造的控制回路如何获取最优控制的问题，提出了基于IMC的模型PID(IMC-MOD-PID)新解法。在以上研究的基础上开发了苯乙烯装置先进控制软件包，并在苯乙烯装置上取得了成功应用。实际应用结果表明，以NLJ算法、IMC和MPSEIVI辨识法为基础的开发的IMC-PID及IMC-MOD-PID先进控制软件，操作使用简单明了，效果显著，维护量小，为苯乙烯装置的先进控制和工业控制系统PID回路的优化改造提供了一种简洁实用的方法。更多还原

【Abstract】 Styrene is an important organic chemical material and has wide applications. On the whole, the world styrene production capability is superfluous, but the consumption in china is increased. As a result, more foreign products will go to our country, and this will threaten the styrene development in china. Hence, the scale and technology of the plants should be enhanced using the advanced technology. Many factories have purchased the advanced process control software package (APCSP) for styrene plants to improve economic benefit. These plants had good performance by applying those packages in early days, but the APCSP can’t be used for a long time in china. By analyzing the reason, we find that the robust of the APCSP is not strong, and its maintenance is complicated. In fact, the plants have changed after a long time, but the model in those APCSP can’t be updated on time or is difficult to be updated. Finally the APCSP can’t be used.How to resolve the above problems is the emphasis in this thesis. The model accuracy is the key for the long time and effective running of the APCSP, while the simple and effective model identification method must be performed to reduce the complexity of maintenance. Hence the IMC-PID and MODEL-PID method based on the model identification are proposed. The accurate model is the keystone for applying the advanced process control, and the simplest method is by using optimal method based on the dynamic response of the plant. To simply and effectively get the model, collect data by using the OPC/DDE data traffic technique and get the dynamic characteristic of the control loop by adding test signal to it; to reduce the algorithm complexity, identify the model parameters by using the NLJ optimization method.For the industry plants with longer time constant which difficultly get the stable state because of the larger disturbance and time-delay, the novel closed-loop model identification method called Model Parameters and State Equation Initial Values Identified simultaneously (MPSEIVI) is proposed; for the model error because of larger dead-time, an improved method for resolve the dead-time called M-Pade method is performed.To ensure effective running for a long time, reduce the investment and sufficiently mine the potential of the original system, the model PID technique based on the classical PID control is presented. It uses the existing PID controllers and doesn’t change them. If only the model parameters can be identified, the parameters of the PID controller can be resolved.The results of the classical PID controller are unsatisfactory for some complex control loops. For them, the IMC-PID method has applied in some literatures. However, when we apply the IMC-PID controller, it is too difficult to choose optimal parameters of the PID controller because many results can be found by using the complex mathematics method. Hence, the new method from IMC controller to IMC-PID controller based on the theory of identical step response is proposed.When applying the IMC-PID control the existing PID controller form should be changed, but some control system can not be changed. For these control loops, the model PID method is performed to implement the optimal control.Based on the above theories and methods, a novel advanced process control software package for the Styrene plant is developed, and has a successful application on styrene process. Practical applications show that the IMC-PID and the IMC-MOD-PID APC software based on NLJ、IMC and MPSEIVI algorithms are advantageous of being simple, practicable, economical and easy to perform for the PID optimization of the complex process system.更多还原