水泥联合粉磨系统的建模与预测控制研究

发布时间：2018-05-08 15:32

本文选题：联合粉磨系统 + 水泥粒度　；参考：《济南大学》2016年硕士论文

【摘要】：水泥是我国基础建设和经济发展必不可少的基本原材料,其生产中重要的一个环节为水泥联合粉磨。该环节是由稳流仓、辊压机、打散机(V型选粉机)、磨机、选粉机以及主排风机等设备组成的复杂系统,其中稳流仓和水泥粒度对水泥生产稳定及质量有着重要的影响。为此,本文围绕稳流仓和水泥粒度两个重心,开展水泥粉磨建模与预测控制研究工作。具体研究工作如下:(1)针对带有水泥粒度的联合粉磨系统建模问题,给出一种分工况的联合粉磨粒度建模方法。依据联合粉磨工艺和在线粒度分析仪,分析关键变量之间的相互影响关系,并通过历史数据划分了水泥粒度工况模板(两个典型工况区间);采用滑动平均滤波方法降低历史数据噪声对建模的影响;针对典型工况1,采用回归分析算法建立多输入单输出的粒度模型;针对典型工况2,采用最小二乘支持向量机(LS_SVM)进行了相应建模;仿真结果说明基于水泥粒度工况模版所建立的模型能够较好地描述水泥粒度动态变化过程。(2)为实现联合粉磨稳流仓的稳定控制,给出了基于神经网络极限学习机(ELMNN)的建模以及内模控制。基于联合粉磨系统工艺和变量关系分析,确定了喂料量为影响稳流仓料位主要因素;采用滑动平均滤波对数据进行降噪;利用ELMNN建立稳流仓内部模型,通过Taylor级数设计了稳流仓内模控制器,并分析了闭环系统的稳定性;仿真结果说明所提出的建模方法和控制器能够实现稳流仓料位稳定控制。(3)针对带有非线性的水泥联合粉磨粒度(45μm筛余)稳定控制问题,给出一种基于模型的广义预测粒度控制方法。基于(1)中所建立的典型工况1模型,通过受控自回归积分滑动平均模型(CARIMA)和长时段的优化性能指标,设计了广义预测粒度控制器;借助粒度闭环传递函数,将粒度闭环系统转换为粒度内模结构形式,并分析了闭环系统稳定性;仿真结果证明了所提出方法的有效性。(4)在(1)~(3)的研究成果基础上,将专家系统、Bang-Bang控制、内模控制以及广义预测控制算法相结合,构建了水泥联合粉磨系统的自动控制软件平台。工程应用说明该平台具有良好的运行效果。
[Abstract]:Cement is an essential raw material for infrastructure construction and economic development in China. Cement grinding is an important link in its production. This link is a complex system composed of steady flow bin, roller press, dispersing machine (V type separator, grinding machine, separator and main exhaust air machine), in which steady flow bin and cement particle size have important influence on the stability and quality of cement production. Therefore, the modeling and predictive control of cement grinding are carried out in this paper, focusing on the two centers of gravity of steady flow bin and cement granularity. The specific research work is as follows: (1) aiming at the modeling problem of combined grinding system with cement particle size, a modeling method of combined grinding particle size is presented. According to the combined grinding process and the on-line particle size analyzer, the interaction between the key variables was analyzed. According to the historical data, the cement granularity working mode template (two typical working conditions) is divided, and the influence of the historical data noise reduction on the modeling is reduced by the sliding average filter method. For typical condition 1, regression analysis algorithm is used to establish the granularity model of multiple input and single output, and for typical condition 2, the corresponding modeling is carried out by using least square support vector machine (LSSVM). The simulation results show that the model based on cement granularity working condition template can better describe the dynamic change process of cement granularity. The modeling and internal model control of ELMNN based on neural network are presented. Based on the analysis of the process and the variable relation of the combined grinding system, the feeding quantity is determined as the main factor affecting the material level of the steady flow silo; the moving average filter is used to reduce the noise of the data; and the internal model of the steady flow bin is established by using ELMNN. The internal model controller of steady flow bin is designed by Taylor series, and the stability of closed loop system is analyzed. The simulation results show that the proposed modeling method and controller can realize the stable control of the material level of the steady flow silo. A generalized predictive granularity control method based on model is presented. A generalized predictive granularity controller is designed based on the model of typical working condition 1, which is based on the controlled autoregressive integral moving average model (CARIMA) and the optimization performance index for a long period of time, with the aid of the granularity closed-loop transfer function. The closed-loop system is transformed into a granular internal model structure, and the stability of the closed-loop system is analyzed. The simulation results show that the proposed method is effective, and the expert system Bang-Bang is controlled on the basis of the research results. Combined with internal model control and generalized predictive control algorithm, the automatic control software platform of cement combined grinding system is constructed. The engineering application shows that the platform has good running effect.
【学位授予单位】：济南大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TQ172.63;TP273

【参考文献】