Steam-driven induced draft fan is one of the important equipments for newly built power plants. There are many problems when applying data-driven fault warning scheme to the fans

such as various monitoring parameters

high cost of equipment failure test and few fault operation data. To solve these problems

this paper applies the multivariate state estimation technique(MSET) to monitor the status of the relevant points of the power plant fans. It analyzes a reasonable memory matrix scheme from the perspective of mathematical principle

and uses the singular value decomposition method to calculate the generalized inverse matrix

thus to improve the calculation efficiency. Taking a power plant in Jiangsu as the research object

the data of the surge fault is extracted from the site

and the scheme is applied to perform state alarm real time data warning. The results show that

the above scheme can reasonably reflect the operating state of the equipment

and the surge fault can be warned about 5 hours in advance.