【作者】 周国雄；

【导师】 吴敏；

【作者基本信息】 中南大学 ， 控制科学与工程， 2010， 博士

【摘要】 在炼焦生产过程中同时伴随产生大量的副产品——焦炉煤气。集气过程回收利用副产煤气,不仅节约能源,而且减少环境污染,是钢铁生产的重要环节。集气管压力稳定与否,影响到煤气质量、设备寿命以及生产环境。研究多座不对称焦炉集气管压力控制对钢铁工业生产有着重要的意义。多座不对称焦炉煤气集气过程是一个高度复杂的工业生产过程,具有多变量、强耦合、不对称、非线性、时变、难以建立数学模型等控制难点,采用传统的控制方法或者单一的智能化技术很难达到理想的效果。本文提出一种智能解耦与优化控制策略,为有效解决焦炉集气管压力的优化控制问题提供一种新的途径。论文的研究成果主要包括下述四个方面：(1)基于灰色预测和BP神经网络的集气管压力预测模型针对焦炉煤气集气过程是一个高度复杂的工业生产过程,难以获得焦炉集气管压力的精确数学模型,提出一种基于灰色预测和BP神经网络的集成建模方法。通过对焦炉的过程机理和数据进行分析,将集气管压力过程工况分为两种,采用灰色预测和BP神经网络方法建立的集气管压力预测模型,可以更加准确地预测焦炉集气管压力。(2)基于耦合度分析的焦炉集气管压力模糊解耦算法针对焦炉容量不同、管道布局不同等不对称特性而造成焦炉煤气集气过程具有强耦合特性的特点,提出一种基于耦合度分析的模糊解耦方法。该方法从概率统计的角度出发,采用耦合度分析的方法实时确定不对称焦炉集气管压力的强弱耦合组,降低控制的输入维数；然后分别选用强、弱模糊解耦控制器对相应的集气管压力进行解耦控制,实现焦炉集气管压力的解耦。(3)焦炉集气管压力智能优化控制算法针对集气管压力扰动变化激烈且幅度大的情况,提出一种基于粒子群优化的变结构模糊控制方法。为获得切换的平稳性,根据已建立的集气管压力预测模型,选用模糊切换策略实现。该方法针对集气管压力的不同波动范围,采用不同的控制规则设计两个模糊控制器。针对模糊量化因子调节的困难,采用粒子群优化惯性系数的自适应调整机制,以寻优模糊控制器量化因子。根据蝶阀的流量特性曲线设计蝶阀专家控制器,使蝶阀更好的响应控制量,提高控制品质。针对炼焦生产过程中集气管压力难以根据不同工况进行实时调整的问题,提出采用基于线性回归与RBF神经网络方法建立焦炉集气管压力设定值优化模型,实现了不同工况下的集气管压力优化设定,保证焦炉的稳顺运行。(4)智能优化控制系统及工业应用从控制实时性、数据的采集和获取、系统的可靠性和安全性考虑,设计系统的整体框架,采用OPC通信技术实现应用软件与基础自动化系统的通信,从而实现焦炉集气管压力的智能解耦与优化控制策略。控制算法在某钢铁公司的三座焦炉煤气集气过程中得到实际应用。运行结果表明,算法具有简便、易行、可靠、易扩充及抗干扰能力强等优点,实现了多座不对称焦炉集气管压力的稳定,为多座不对称焦炉煤气集气过程的解耦与优化控制提供了一种有效的途径。更多还原

【Abstract】 Coke is the main raw material in the metallurgical industry. In the coking process, large amount of by-product gas will be generated from coke-ovens, and the process of recycling by-product gas is called gas collecting process. It is an important link in the iron and steel production, which not only saves energy but also reduces environmental pollution. The stability of gas collector pressure directly influences the gas quality, the life-time of coke-ovens and the producing environment. Therefore, it is a vital significance for the iron and steel enterprise to study on the control of gas collecting process of multi asymmetrical coke ovens.The gas collecting process of multi asymmetrical coke ovens is a highly complex industrial process, which is multi-variable, strong coupling, asymmetry and nonlinearity, time-varying and hard in modeling, so it is difficult to acquire good control results with classical control theory or unique intelligent technology. An intelligent decoupling and optimization technology is proposed in this paper and a valid way is presented for the optimization problem in the gas collector pressure. The main achievement in this paper includes several aspects as follows:(1) Prediction model of gas collector pressure based on gray forecasting and BP neural networkAs the gas collecting process of multi asymmetrical coke ovens is a highly complex industrial process, which is difficult to obtain accurate mathematical model, an integrated modeling method which incorporated gray forecasting and BP neural network is proposed. Through analysis of the process mechanism and data of coke-ovens, the process of the gas collector pressure conditions has been divided into two, and the prediction modle of gas collector pressure based on gray forecasting and improved BP neural network is established, whichi can obtain accurate predictions of gas collector pressure.(2) Fuzzy doucoupling method based coupling analysis of gas collector pressure of coke-ovensThe gas collector pressure of a coke-oven with asymmetry has different capacity and its steel enterprises has different channels, which make the gas collecting process strong coupling, and based on this characteristics, a fuzzy decoupling control approach based on coupling analysis is developed. From the perspective of probability and statistics, the strong or weak coupling group for the asymmetrical gas collector pressure is real-time determined by coupling analysis method, which can reduce the inputs dimension of the follow-up control. And then strong or weak fuzzy decoupling controller is chosen for the corresponding gas collector pressure, which can realize decoupling of the gas collector pressure.(3) Intelligent control method of gas collector pressure of coke-ovensThe gas collector pressure of a coke-oven fluctuates sharply and changes suddenly, and based on this characteristic, a variable structure fuzzy control approach based on particle swarm optimization method is developed. In order to obtain a smooth switch, according to the established prediction model of gas collector pressure, fuzzy switching strategy is seclected. Two fuzzy controllers which use different control rules are designed according to different ranges of gas collector pressure. For the situation that fuzzy controller quantifiable factors are difficult to be modified, the adaptive tuning laws of PSO inertia coefficient are adopted to obtain the preferable fuzzy controller quantifiable factors. And butterfly valve controller is developed to adapt to the characteristics of butterfly valve so as to enhance the control quality. In coking plant production process, the target of gas collector pressure is difficult to adjust timely according to different conditions, the linear regression and RBF neural network are integrated to build dynamitic model for the gas collector pressure, which can obtain the different set point of the gas collector pressure according to different conditions, thus the cock-ovens can be ensure good run.(4) Intelligent decoupling and optimal control system and its applicationThe system frame is designed after the real-time, data collection and acquisition, reliability and safety are considered. The communication between the application soft and basic automation is realized with OPC and intelligent decoupling and optimization control technique for multiple gas collector pressure in coke-ovens with asymmetry is realized. The control algorithm has been applied in gas collecting processes of three coke ovens in an iron and steel enterprise. The actual running result shows that the control algorithm is simple, feasible, reliable, and easy to expand and has strong anti-disturbance ability. The control algorithm has realized the pressure stability of the gas collecting process of multi coke-ovens. In a word, it provides an efficient method for the decoupling and optimazation control of the gas collecting process of multi asymmetrical coke ovens.更多还原