The invention relates to a power plant induced draft fan fault early warning method based on MSET and deviation degree, and the method comprises the following steps: S1, obtaining multivariable data of an induced draft fan in a normal operation state, and building an MSET model; S2, obtaining an estimated vector Xest of the running state of the induced draft fan according to the MSET model; S3, calculating the deviation degree between the observation vector Xobs and the estimation vector Xest of the running state of the induced draft fan by utilizing a deviation degree function; and S4, comparing the deviation degree with a deviation degree threshold value to obtain a residual error, and determining whether to trigger an alarm according to the residual error. An MSET model is established through multivariable data of the induced draft fan in a normal operation state, and an optimal estimation value of model operation is obtained. The deviation degree function is used for judging the deviation degree between the actual value and the optimal estimated value, and the fault development process is reflected visually. The deviation degree is compared with a set threshold value to obtaina residual error, and an alarm is raised if the residual error is too large. Compared with the prior art, the method has the advantages of improving the early warning accuracy and timeliness and the like.