The fault diagnosis inverse problem with Ant Colony Optimization and Ant Colony Optimization with dispersion

This paper is focused on the formulation of fault diagnosis (FDI) using an inverse problem methodology. The FDI inverse problem is formulated as an optimization problem which is solved by bio-inspired algorithms. In this case, the algorithms Ant Colony Optimization (ACO), and its modified version ACO-d have been applied. This approach combines results from FDI area for making an alternative uniqueness analysis of the FDI inverse problem, which is related with detectability and isolability of faults, components of the diagnosis. The proposed approach is tested using simulated data from the Inverted-Pendulum System which is recognized as a benchmark for control and diagnosis. This work also studies the influence of ACO and ACO-d parameters in order to obtain a robust (to external disturbances) and sensitive (to incipient faults) diagnosis. The results show the suitability of the approach. They also indicate that parameters values allowing a greater diversification of the search, yield a better diagnosis. The ACO-d algorithm enables better diagnosis than ACO.