robust sensor fault diagnosis by multiple model approach

This paper concerns the sensor faults diagnosis in the presence of unknown input, disturbances, actuator, and component faults using multiple linear model approaches in a group of complicated nonlinear systems. first, the sensor fault by adding a group of dynamic filters in the model output (hereinafter referred to as transfer filters) is arranged in additive form. consequently, using a group of special linear kalman filters a group of decoupled residuals are produced that only are dependent on sensor faults and noises. in continuing to distinguish the current operating point, by generating robust weighing coefficients an adaptive model and a stable adaptive filter are made that estimates the sensor fault in all operating areas. finally, in a simulation environment, the competency of the proposed approach is evaluated.

robust sensor fault diagnosis by multiple model approach

this paper concerns the sensor faults diagnosis in the presence of unknown input, disturbances, actuator, and component faults using multiple linear model approaches in a group of complicated nonlinear systems. first, the sensor fault by adding a group of dynamic filters in the model output (hereinafter referred to as transfer filters) is arranged in additive form. consequently, using a group of special linear kalman filters a group of decoupled residuals are produced that only are dependent on sensor faults and noises. in continuing to distinguish the current operating point, by generating robust weighing coefficients an adaptive model and a stable adaptive filter are made that estimates the sensor fault in all operating areas. finally, in a simulation environment, the competency of the proposed approach is evaluated.