Sensor Placement for Fault Diagnosis Based on Causal Computations

This work develops a methodology to solve the sensor placement problem for fault detection and isolation. The proposed methodology is suitable to handle highly non-linear and large scale systems since it is based on structural models. Furthermore, causality is assigned in those variable-equation relations that the variable can be computed from the equation in order to guarantee the computability of the unknown variables in the residual generation design. Finally, the developed methodology is applied on an air compressor model.