Research On Fault Diagnosis Technology Based On Cointegration Coefficient Matrix

As an important part of process control system,fault diagnosis technology is a powerful cornerstone for industrial systems to achieve a high degree of intelligence.At present,the research of fault diagnosis technology is mostly based on the assumption that the process is stable.However,due to production demand,load adjustment,external disturbance and other reasons,industrial production process often shows nonstationary characteristics.The results of existing fault diagnosis technologies are not very effective in the fault diagnosis of industrial production process.In order to solve the problem of fault diagnosis of non-stationary process,the cointegration coefficient matrix theory is introduced into the research of fault diagnosis of non-stationary process,and the control system is taken as the research object.The specific research of this paper are summarized as follows:(1)Aiming at the problem that traditional methods can't deal with the nonstationary process data,the cointegration coefficient matrix theory is introduced.The principle of cointegration coefficient matrix theory is introduced and applied to the process of extracting hidden information from non-stationary systems.(2)Based on the cointegration coefficient matrix theory,a research method for the degradation degree of electric control valve in industrial process feedback control loop is proposed.After the collected non-stationary data are tested by ADF,the cointegration model is established by all variables which are first-order single integration.Establish a cointegration model and obtain a matrix of cointegration coefficients to obtain a stable residual sequence for calculating monitoring statistic to evaluate the degradation degree of the valve.The simulation model of electric regulating valve built by Simulink is used to verify the feasibility of the method.(3)Considering the mixed characteristics of stationary and non-stationary variables in the industrial process,this paper adopts the fault diagnosis method based on the combination of PCA and cointegration coefficient matrix theory.Firstly the data is modeled locally,and then the whole is modeled to filter out the fault information contained in the hybrid system.The fault diagnosis of the control system is realized by calculating the monitoring statistics,and the contribution rate of each variable to the monitoring statistics is analyzed to determine the source of the fault by using the contribution graph method.Finally,the above method is applied to the monitoring platform of a municipal solid waste power generation system.The reliability of the method are verified with the help of superheated steam temperature control system.