A fault identification technique for FBG sensors embedded in composite structures

The paper proposes a method for the identification of sensor faults, that can be applied in those vibration control applications where large arrays of sensors are used. Indeed, distributed measurement proves to be very effective in recognizing and suppressing the contribution of different modes to structure vibrations, but its efficiency is dramatically reduced if one or more sensors do not work correctly. This is the case, for example, of fiber Bragg grating sensor chains. For this reason, a sensor fault identification algorithm is introduced. This method, based on the analysis of the residuals of the measurement estimation, allows to identify different typologies of sensor fault or malfunctioning, such as complete fault or additive and multiplicative measurement errors. Once identified, these sensors can be excluded from the feedback loop of the control algorithm in order to avoid unwanted behaviors or instabilities. Numerical and experimental tests have been carried out on a carbon fiber structure considering different fault conditions. Results show that it is possible to identify the faulty sensors and thus improve the signals used in the feedback loop.