Study On The Method Of Data Correction And Significant Error Diagnosis In Thermal Process

The generating set of the thermal power plant collects process information through various measuring points,so as to realize the monitoring and automatic control of the thermal process.However,the failure of the measuring point will produce abnormal signals,which will greatly reduce the reliability of the unit automation and information system.Therefore,it is necessary to detect and diagnose the running status of the measuring point.This article focuses on thermal process data correction and significant error diagnosis methods.The main contents include:(1)A new robust function is proposed based on the principle of robust data correction,and the effectiveness of the algorithm is verified through numerical simulation examples.The mechanism model of the steam turbine regenerative system is constructed,and the validity of the algorithm is verified by the data of Heat Consumption Acceptance rate operating conditions.The method is further applied to the correction of the measured flow,and the Constraint equation test method and measurement data test method is combined to realize the fault location.The results show that the method effectively suppresses error transmission,can correct the flow rate to a reasonable range,and improve the credibility of the data.(2)There are a large number of dynamic changes in the thermal process,and it is difficult for conventional static algorithms to accurately monitor the dynamic process.Based on the principal component analysis method,this paper introduces the dynamic principal component analysis method based on augmented matrix,and analyzes and compares the advantages and disadvantages of the two methods.Numerical examples prove the advantages of dynamic principal component analysis in fault detection of dynamic processes.At the same time,it is also found that dynamic principal component analysis has the problems of large noise amplitude and difficulty in fault separation,which requires further study.(3)Aiming at the failure of the diagnosis index in the fault separation process,a new fault diagnosis index is proposed to eliminate the problem of diagnosis failure.At the same time,an improved reconstruction contribution map is proposed based on the conventional reconstruction contribution map method.The method effectively suppresses the “residual effects” problem.The method is further applied to the dynamic principal component analysis method,and the simulation example shows that the improved reconstruction contribution graph method can effectively suppress the problem of “residual effects”.(4)The fault diagnosis method based on sparse principal component analysis is studied,and the advantages of this method in reducing process data noise are verified through simulation examples.For the problem of determining the non-zero elements of the principal component,the forward selection algorithm is combined with the interior point method to reduce the noise of the process data and avoid the problem of “oversparseness”.Combining the advantages of the two algorithms of sparse dynamic principal component analysis and improved reconstruction contribution graph,it is fused into a sensor fault diagnosis method based on dynamic sparse principal component analysis.This method is used in the fault diagnosis of flow in the dynamic process of steam turbine regenerative system.obtained a good outcome.