Particle swarm optimization of excitation system design of magnetic eddy current sensor

The detection capability of magnetic eddy current and magnetic flux leakage sensors depends on the magnetization level in the test specimen. While low magnetization field intensity makes it difficult to detect defects, higher magnetization levels increase background noise as well as the size and weight of the sensors. Moreover, powerful magnets are used in the magnetization circuit that is difficult to handle and pose potential health and safety hazards. Finite element modelling is widely used for the optimization of the design of magnetization yokes. Modelling softwares are limited in their ability to conduct artificial intelligence-based optimization and require a large number of iterations. This can be time-consuming and computationally expensive. An optimization technique using particle swarm optimization algorithm for designing the excitation system for magnetic eddy current sensors is presented in this paper. Numerical simulation is used to determine the objective function and input variables for the algorithm. A comparative study is carried out to evaluate the algorithm's performance against genetic and artificial bee colony algorithms. The sensor design parameters obtained using the algorithm results are validated through experiments. The results show that the PSO is a fast and computationally efficient algorithm for optimizing the yoke design.