基于PCA的工业锅炉运行监控与优化研究

发布时间：2018-08-14 19:57

【摘要】：我国工业锅炉数量众多,运行效率普遍较低,导致能源浪费严重,对锅炉的运行状态进行有效的监控和优化,提高锅炉的运行水平能够产生显著的能源和环境效益。工业锅炉是多输入、多输出的复杂系统,其参数间具有强耦合性,难以通过建立精确的数学模型对其运行状态进行监控与优化。本文提出一种利用主元分析法(Principal Component Analysis, PCA)从锅炉的历史数据中提取有效信息,确定影响锅炉效率的主要参数,进而指导操作人员调整运行参数,实现锅炉运行状态优化的方法。此外,对锅炉数据采集过程中出现异常数据的甄别与恢复等预处理方法进行了研究。本文主要的工作如下：首先针对工业锅炉运行数据中存在的异常数据,用回归分析法对锅炉运行数据进行实时处理。利用锅炉中正常的历史数据建立多项式回归模型,预测下一时刻的值并与测量值比较,判别测量值是否为异常数据,并对异常数据予以修正。其次,将主元分析法引入到工业锅炉运行监控及优化过程,利用锅炉热效率较高的历史数据建立主元模型,获得Q和T2统计量的控制限,将锅炉实时数据代入主元模型计算出Q和T2统计量,并与其控制限比较,达到对锅炉运行监控的目的。当锅炉处于热效率较低的工况时,利用基于Q和T2贡献图的方法确定导致低效的主要参数,指导操作人员进行有针对性的操作。最后采用B／S模式,建立了一个具有数据采集、数据处理和数据发布的网络化平台,把复杂的数据分析与处理工作移植到专用的数据中心,然后将操作指导结果通过网络反馈给操作人员,有利于计算、存储资源的集约化应用。系统在某工业现场投入实用,产生一定的节能效益,证明了该系统方案和优化方法的可行性和有效性。
[Abstract]:There are a large number of industrial boilers in our country, and the operation efficiency is generally low, which leads to serious energy waste. To effectively monitor and optimize the boiler operation state and to improve the boiler operation level can produce significant energy and environmental benefits. Industrial boiler is a complex system with multi-input and multi-output, and its parameters are strongly coupled. It is difficult to monitor and optimize the operation state of industrial boiler by establishing accurate mathematical model. This paper presents a method of extracting effective information from boiler historical data by principal component analysis method (Principal Component Analysis, PCA), determining the main parameters affecting boiler efficiency, and then instructing operators to adjust operation parameters and realize boiler operation state optimization. In addition, the preprocessing methods such as discrimination and recovery of abnormal data in boiler data acquisition are studied. The main work of this paper is as follows: firstly, according to the abnormal data of industrial boiler operation data, the real-time processing of boiler operation data is carried out by using regression analysis method. A polynomial regression model is established by using the normal historical data of the boiler to predict the value of the next moment and compare it with the measured value to judge whether the measured value is abnormal data and to correct the abnormal data. Secondly, the principal component analysis method is introduced into the process of industrial boiler operation monitoring and optimization, and the principal component model is established by using the historical data of higher boiler thermal efficiency, and the control limits of Q and T2 statistics are obtained. The statistics of Q and T2 are calculated by replacing the real time data of boiler into the principal element model and compared with their control limits to achieve the purpose of monitoring the operation of the boiler. When the boiler is in the condition of low thermal efficiency, the method based on Q and T2 contribution diagram is used to determine the main parameters leading to the low efficiency and guide the operator to carry out the targeted operation. Finally, a network platform with data acquisition, data processing and data publishing is established by using the B / S model, and the complex data analysis and processing work is transplanted to the special data center. Then the operation guidance result is fed back to the operator through the network, which is advantageous to the intensive application of computing and storage resources. The system has been put into practice in a certain industrial field and has produced certain energy saving benefit, which proves the feasibility and effectiveness of the system scheme and optimization method.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TK227

【相似文献】