基于典型工况模型的篦冷机预测控制研究

发布时间：2018-06-30 01:34

本文选题：篦冷机 + 预测控制　；参考：《济南大学》2017年硕士论文

【摘要】：篦冷机作为熟料冷却系统的关键设备,冷却效果的好坏不仅关系到水泥熟料的质量,还会对水泥生产的其他环节产生间接影响。当今国内水泥厂篦冷机的控制主要以人工控制和半自动控制为主,人工控制主观性强、随意性大,而且篦冷机自身存在非线性、大滞后、工况波动频繁等诸多问题,对篦冷机的控制研究正由传统控制算法向先进智能控制算法研究转变,因此,本课题对水篦冷机进行预测控制研究是很有必要的。本文在深入了解水泥篦冷机工艺的基础上,结合操作员的操作经验,以典型工况为切入点,采用动态矩阵控制算法,提出了基于典型工况下的篦冷机预测控制方案,最后采用模块化思想对篦冷机预测控制软件进行设计与开发,经现场调试后,应用于现场实际控制中,通过实际应用效果表明,该控制软件取得了不错的控制效果,具有较强的实用性。本文的详细工作内容如下:(1)建立水泥篦冷机典型工况模板。针对篦冷机自身存在的非线性、工况波动频繁等问题,采用K-means聚类方法对历史数据进行聚类分析,划分出三种典型工况,并根据主导变量(篦压)进行划分得到其他参变量的工况基准值点,所划分的典型工况占到了整个篦冷机工况的90%左右,通过实际应用也验证了该方法划分工况的合理性和正确性。本课题是在典型工况背景下对篦冷机进行研究,该研究也为下一步的研究奠定基础。(2)典型工况下篦冷机预测模型研究。针对所划分出的典型工况,选取现场典型历史数据,考虑到现场干扰众多,数据难免会有误差,因此,选取均值滤波方法对典型历史数据进行均值滤波处理,然后采用最小二乘法建立篦冷机预测模型。由于篦冷机长时间运行过程中存在工况的切换,当工况变化时,势必会牵扯到模型的切换。针对不同工况间模型的切换,采用模糊加权的方式进行平滑切换,以此来保证切换过程的平稳性。(3)典型工况下篦冷机DMC预测控制器的研究。根据建立的预测模型设计篦冷机DMC控制器,通过MATLAB仿真确立了各典型工况下最佳的控制参数,考虑到现场实际应用,针对控制器之间的切换,采用模糊加权的方式实现平滑切换,并对所设计的控制器稳定性性能进行分析。(4)篦冷机预测控制软件的工程实现。根据以上研究内容,采用模块化思想开发篦冷机预测控制软件,该软件主要由数据采集系统、数据库存储系统、控制算法系统等三部分组成。在实际生产现场,该软件以外挂包的方式安装在工程师站,通过现场的实际运行,取得了不错的控制效果,具有较强的实用性,同时也验证了本课题的研究意义所在。
[Abstract]:Grate cooler as the key equipment of clinker cooling system, the cooling effect is not only related to the quality of cement clinker, but also has an indirect effect on other aspects of cement production. At present, the control of grate cooler in domestic cement plant is mainly controlled by manual control and semi-automatic control. Manual control is subjective and random, and the grate cooler itself has many problems, such as nonlinear, large lag, frequent fluctuation of working conditions, and so on. The control research of grate cooler is changing from traditional control algorithm to advanced intelligent control algorithm. Therefore, it is necessary to study the predictive control of grate cooler in this paper. On the basis of deep understanding of cement grate cooler technology, combined with operator's operation experience, taking typical working condition as the breakthrough point and adopting dynamic matrix control algorithm, the predictive control scheme of grate cooler based on typical working condition is put forward in this paper. Finally, the predictive control software of grate cooler is designed and developed by modularization idea. After debugging, it is applied to the actual control on the spot. The result of practical application shows that the control software has achieved good control effect. It has strong practicability. The detailed work of this paper is as follows: (1) the typical working condition template of cement grate cooler is established. Aiming at the nonlinearity and frequent fluctuation of the grate cooler itself, K-means clustering method is used to cluster the historical data, and three typical working conditions are divided. According to the main variable (grate pressure), the reference points of other parameters are obtained. The typical working conditions are about 90% of the total grate cooler working conditions. The rationality and correctness of the method are verified by practical application. In this paper, the grate cooler is studied under the background of typical working conditions, which also lays a foundation for the next research. (2) the prediction model of grate cooler under typical working conditions. In view of the typical working conditions, the typical historical data is selected. Considering the numerous field interference, the data will inevitably have errors. Therefore, the mean value filtering method is selected to process the typical historical data. Then the prediction model of grate cooler is established by least square method. Due to the switching of the operating conditions during the long time operation of grate cooling units, the model switching will be involved when the operating conditions change. In order to ensure the smoothness of the switching process, the fuzzy weighting method is used to smooth the model switching between different working conditions. (3) the research of DMC predictive controller for grate cooler under typical working conditions. The DMC controller of grate cooler is designed according to the established prediction model. The optimal control parameters under typical working conditions are established by MATLAB simulation. Considering the practical application in the field, the switching between controllers is considered. The fuzzy weighting method is used to realize smooth switching and the stability performance of the controller is analyzed. (4) the engineering implementation of predictive control software for grate cooler. According to the above research contents, the predictive control software of grate cooler is developed by modularization. The software consists of three parts: data acquisition system, database storage system and control algorithm system. In the actual production site, the software is installed in the engineer station by the way of external package. Through the actual operation of the field, the software has achieved good control effect and has strong practicability. At the same time, it has verified the significance of the research of this subject.
【学位授予单位】：济南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TQ172.622.4;TP273

【参考文献】