A Review on Optimal Sensor Placement for Health Monitoring

Sensor data provide the foundation for performance and health assessment of most complex systems. An optimal sensor placement is defined as a sensor configuration that achieves the minimum cost while observing prespecified performance criteria. Various optimization algorithms, from random search to heuristic algorithms, have been used for optimizing the sensor placement. A literature search for sensor placement methods yields a large number of publications on various optimization methods. Random search is suitable for a small and simple sensor placement problem since it is straightforward and easily implemented. But it is time consuming and inefficient when dealing with a large system. A variety of heuristic search methods, including Simulated Annealing, Tabu Search, or Genetic Algorithms are available. In large-scale systems consisting of multiple components, a fault may propagate through several components when it occurs. So the solution of sensor placement problem at the system level is required. Then Cause-Effect analysis methods, such as fault tree method, Petri-Net method, and graph theory, were widely used.