In consideration of the common sensor faults in aero-engine

a new algorithm was proposed based on support vector regression(SVR) trained by improved particle swarm optimization(PSO)

and was used for sensor fault diagnosis system based on data fusion.A bank of SVR was applied to sensor fault detection

isolation and validation.This fault diagnosis system would isolate the fault sensor relying on the isolation mechanisms

and select the validation module for signal recovery when some fault sensors were detected.According to the simulation experiment of aero-engine sensor faults(impact fault

offset fault and drift fault)

the results show that this sensor fault diagnosis system has high level of precision and is an effective way to sensor fault diagnosis under the conditions of both one and multiple sensor faults.