| The power industry is an important energy production industry of the national economy.It has a close relationship with human life and work.Coal-fired power generation is the most important part of the power production system,and every technological advancement will bring huge economic and social benefits.Once the system fails,it will not only affect the people’s life,but also hinder industrial development in the region.With the advancement of technology in recent years,the scale of power generation data has become more and more huge,which brings convenience to the research of process monitoring methods based on big data.In this paper,we used process data from a power plant in Zhejiang and conducted a plant-wide process monitoring method based on distributed statistical methods.The research content includes:The second chapter of the paper introduces the basic principles of modern coal-fired power generation and the basic methods involved in distributed plant-wide process monitoring,including a module division method based on mutual information,a fault detection method based on principal component analysis,Bayesian fusion method and Granger Causal analysis.Based on these methods,the plant-wide distributed process monitoring framework is proposed.The third chapter of the paper carried out plant-wide distributed process monitoring research,where the data come from unit # 6 of a power plant in Zhejiang.First,the mutual information method is used to divide the process variables of the power plant into multiple sub-modules;and in each sub-module,the process monitoring model is established using the principal component analysis method,then used Bayesian inference to monitor the fault result and used contribution plot method to isolate fault variables;Finally,Granger causality analysis method is used to find out the root cause of the fault.Four typical faults are used to verify the feasibility of the plant-wide process monitoring method proposed in this paper in industrial systems.The fourth chapter of the paper used MATLAB to design a multivariate statistical analysis toolbox.The main functions include data loading,data analysis,data classification,data regression and plant-wide monitoring modules.The toolbox realizes the transformation from theoretical method to practical application,which greatly facilitates the process monitoring process of power plant data.Finally,the fifth chapter of the paper summarizes the research content and innovation of the full text,and prospects the follow-up research of the subject. |
