The invention belongs to the technical field of aero-engine fault diagnosis, and particularly relates to a fault diagnosis method of aero-engine exhaust temperature sensors. The control parameters andstate parameters of an engine are acquired. The method includes the following steps: the temperature data of multiple exhaust temperature sensors is acquired; the control parameters, the state parameters and each piece of acquired temperature data are input as input information to a four-layer neural network model; the four-layer neural network model outputs the fault weight of each exhaust temperature sensor according to the input information; each fault weight is combined with the corresponding temperature data according to a weighted average deviation classification algorithm, and a resultabout whether there is a faulty exhaust temperature sensor is output; and if there is a faulty exhaust temperature sensor, the information of the faulty exhaust temperature sensor is output. The method can be used to accurately judge a faulty exhaust temperature sensor under simultaneous working of multiple exhaust temperature sensors.