A Unified and Efficient Algorithmic Approach to Model-based Diagnosis and Optimal Sensor Placement

This paper presents a new unified approach for solving diagnosis and sensor placement problems. In previous works [3], we have shown that the main computational challenge in model-based diagnosis, that is, the calculation of minimal diagnosis set, can be represented as the solution of the Hitting Set Problem. We have also shown that this calculation can be mapped onto the Integer Programming problem and have further developed an efficient branch-and-bound technique for its solution. In this paper, we first show that the optimal sensor placement can be also mapped onto the integer Programming problem. We then extend our previously developed branch-and-bound technique for efficient solution of the optimal sensor placement problem. To our knowledge, this is the first systematic (i.e., non exhaustive search) approach for the solution of the optimal sensor placement problem. Our new method can be used both at the design level, where the number and position of sensors are not known, and the case where we want to add more sensors to the system to enhance its diagnosability.