Optimal sensor placement within a hybrid dense sensor network using an adaptive genetic algorithm with learning gene pool

This work develops optimal sensor placement within a hybrid dense sensor network used in the construction of accurate strain maps for large-scale structural components. Realization of accurate strain maps is imperative for improved strain-based fault diagnosis and prognosis health management in large-scale structures. Here, an objective function specifically formulated to reduce type I and II errors and an adaptive mutation-based genetic algorithm for the placement of sensors within the hybrid dense sensor network are introduced. The objective function is based on the linear combination method and validates sensor placement while increasing information entropy. Optimal sensor placement is achieved through a genetic algorithm that leverages the concept that not all potential sensor locations contain the same level of information. The level of information in a potential sensor location is taught to subsequent generations through updating the algorithm’s gene pool. The objective function and genetic algorithm are experimentally validated for a cantilever plate under three loading cases. Results demonstrate the capability of the learning gene pool to effectively and repeatedly find a Pareto-optimal solution faster than its non-adaptive gene pool counterpart.