| As China’s energy-saving and environmental protection standards gradually increase,desulfurization systems and power units simultaneous construction and put into operation in the thermal power plant.Steam-driven induced draft fan,that the induced draft fan combines with booster fan,is energy saving.But it has monitoring parameters,high cost of test equipment failure and fault data less problem.At the same time,data-driven fault warning research stays at the theoretical stage,combined with the site is not close.To solve these problems,this paper introduces the multivariate state estimation technology(MSET)theory to perform field monitoring of field devices and achieves the purpose of early warning in the absence of prior fault information.it can save time for the thermal power plant maintenance,auxiliary equipment monitoring operations personnel advance work.This paper studies the transmission and back-end algorithms field data,to provide a reasonable solution to the problem that app ears in MSET algorithm in localization.Based on the Python development real-time data acquisition interface on-site.The interface to implement fast access,stable operation.In view of previous algorithm studies theoretical results,combined with field c onditions existing memory matrix D extraction optimization.In addition,this paper from data architecture,typical failure mode and actual field data starting situation,etc.,to overcome the previous lack of theoretical research model and field combined defects.In this study,the MSET method is used to monitor the status of the relevant measurement points of the steam induced draft fan,construct a reasonable memory matrix,and use the singular value decomposition method to calculate the generalized inverse matrix,to improve the calculation efficiency.Taking a power plant in Jiangsu province as the object of study,the data set containing surge faults was extracted from the site,and the scheme in this paper was used for status warning of real-time data.The results show that the scheme can reasonably reflect the running state of the equipment,and can capture the early fault trend of the induced draught fan,and can give an early warning about 5 hours in advance for the surge fault. |
