|
Guided waves (GW) are one of the most promising tools for structural health monitoring (SHM). They allow fast inspection of a large area. Thus GW based SHM is finding applications in several fields like aerospace, automotive, wind energy, etc.
Recently, the use of optical fiber sensors and more specifically fiber Bragg grating (FBG) sensors are getting increasing attention from researchers for GW sensing. The FBG sensors are deployed in the edge-filtering configuration for the sensing. The FBG sensors offer several advantages such as light weight and small size. They also offer the chance to be embedded, and multiplexed. Unfortunately, they are passive sensors and show directional sensitivity. Also they are relatively expensive. Thus there is a need to optimize the usage of the sensors to obtain maximum information with minimum cost.
Several different optimization strategies have been proposed in the literature including gradient based as well as evolutionary algorithms. For a non-linear, non-differentiable cost function evolutionary algorithms (EA) such as genetic algorithm, particle swarm optimization, etc. have shown a lot of promise. In this paper the Covariance-Matrix Adaptation Evolution Strategy (CMAES) has been employed for the sensor network optimization for GW based SHM. First, the ability of the algorithm to obtain the global optima is investigated and then the results are compared with the genetic algorithm (GA) which is the most commonly used EA. 3 different criteria for the optimization are used namely, the coverage with 3 actuator-sensor (AS) pairs (cov3), coverage with 1 AS pair (cov1), and the scalarized combination of cov3 and cov1. The results indicate that indeed CMAES is a powerful optimization tool for the sensor placement optimization problem.
|
