Deep learning-based field-guided optimization method

What is it about?

In this paper, considering the role of physical information in guiding and constraining the optimization method with stochasticity, the establishment of the guided optimization method based on the feedback of physical information was tested on TEAM Problem 35 and achieved very good results. Using the digital twin technique,the function values at the sampled points are used to predict the function values at other points. The variance of the generated magnetic field map is added to the objective function of the optimization algorithm to reflect the guidance of the magnetic field information for optimization. The better population obtained through the particle swarm optimization algorithm is put into the agent model to compute the magnetic field cloud maps generated under different parameters. The standard deviation and variance information of the pixel values of the field maps is calculated to get a relatively uniform solution to guide the optimization of the whole algorithm. The optimization is done through continuous iterations until the Pareto optimal solution set is obtained. The results are compared with those of traditional optimization algorithms to verify the advancement of the method.

Featured Image

Photo by Chris Ried on Unsplash

Photo by Chris Ried on Unsplash

Why is it important?

This article uses deep learning methods to replace traditional numerical simulation methods for fast calculation of electromagnetic fields, and incorporates the variance of magnetic field uniformity as information into the optimization algorithm to solve the optimization problem of TEAM Problem 35.

Perspectives