填料塔液泛在线监测的改进主元分析方法研究

发布时间：2018-07-25 06:49

【摘要】：液泛气速是填料塔操作的上限气速,若为避免液泛而采用过低的操作气速又会造成气液分布不均及投资浪费。实时监测填料塔运行状态,不仅能够保证其运行于正常工况下,也有利于提高设备效率,具有重要的现实意义及工业应用前景。本文首先综述了液泛监测的研究现状,提出将数据驱动的方法应用于填料塔液泛在线监测过程中。针对单一主元分析(PCA,Principal Component Analysis)监测模型难以满足复杂工况的监测要求,结合集成学习的策略,提出改进的动态主元分析方法(Enhanced-DPCA,Enhanced Dynamic Principal Component Analysis),并通过设计不同的实验进行验证。最后,开发出基于改进主元分析方法的液泛实时监控软件,实现填料塔运行状态的实时监控。主要创新点如下:(1)针对传统液泛监测模型难以获得广泛认可的缺点,同时考虑过程数据未得到充分利用的现状,将PCA及其改进方法应用于填料塔液泛在线监测中。通过对监测结果进行分析,针对建模数据的多样性问题,及单一监测模型难以满足复杂操作工况的监测要求,提出一种基于模糊C均值(FCM,Fuzzy c-Means)聚类的监测方法,并通过贝叶斯加权实现监测结果的集成。文中以DPCA为例,建立Enhanced-DPCA监测模型,经实验验证,集成监测模型能够更好地利用不同特征数据的信息,有利于提高监测效果,能够保证填料塔运行于安全且高效的操作气速之下。(2)结合数据采集软件,开发出基于Enhanced-DPCA方法的填料塔实时监控软件。首先,根据实时数据的特征,对监测模型参数进行调整,使其更加适合实时过程数据特征,之后,实现不同喷淋密度下填料塔运行状态的实时监测。结果表明,通过该软件能够保证填料塔的操作气速达到液泛气速的80%,相对于传统工业过程中所确定的60%有明显提高,有利于提高设备运行效率。最后,在监测软件中加入控制模块,实现填料塔自动操作,并且能够在监测到液泛后,自动调整风速,使其恢复至正常状态。
[Abstract]:Liquid gas velocity is the upper limit of gas velocity in packed column operation. If the gas velocity is too low to avoid flooding, it will lead to uneven distribution of gas and liquid and waste of investment. The real-time monitoring of the running state of the packed tower can not only guarantee its operation under normal working conditions, but also improve the equipment efficiency. It has important practical significance and industrial application prospect. In this paper, the research status of liquid flooding monitoring is reviewed firstly, and a data-driven method is proposed to be used in the process of liquid flooding monitoring in packed column. Considering that the single principal component analysis (PCA-Principal Component Analysis) monitoring model is difficult to meet the monitoring requirements of complex working conditions, an improved dynamic principal component analysis (Dynamic Principal Component Analysis),) method is proposed and verified by designing different experiments in combination with the strategy of integrated learning. Finally, the software of liquid flooding real-time monitoring based on improved principal component analysis method is developed to realize the real-time monitoring of the running state of packed tower. The main innovations are as follows: (1) aiming at the shortcomings of traditional liquid flooding monitoring model which is difficult to obtain wide recognition and considering the current situation that process data are not fully utilized, PCA and its improved method are applied to the flooding monitoring of packed column. By analyzing the monitoring results, aiming at the diversity of modeling data and the difficulty of single monitoring model to meet the monitoring requirements of complex operating conditions, a monitoring method based on fuzzy C-means (FCM) fuzzy c-Means clustering is proposed. The integration of monitoring results is realized by Bayesian weighting. Taking DPCA as an example, a Enhanced-DPCA monitoring model is established. It is proved by experiments that the integrated monitoring model can make better use of the information of different characteristic data, which is helpful to improve the monitoring effect. It can ensure that the packed tower can operate safely and efficiently under the operating gas velocity. (2) combined with the data acquisition software, the real-time monitoring software of the packed tower based on Enhanced-DPCA method is developed. Firstly, according to the characteristics of real-time data, the parameters of the monitoring model are adjusted to make it more suitable for real-time process data features. Then, the real-time monitoring of the running state of packed tower under different spray density is realized. The results show that the operating gas velocity of the packed tower can reach 80% of the liquid gas flooding velocity by the software, which is obviously higher than the 60% determined in the traditional industrial process, which is beneficial to improve the operation efficiency of the equipment. Finally, the control module is added to the monitoring software to realize the automatic operation of the packed tower, and the wind speed can be automatically adjusted to normal state after the liquid flooding is monitored.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TQ053.5

【相似文献】