A New Method for the Robust Design of High Field, Highly Homogenous Superconducting Magnets Using an Immune Algorithm

When designing high-field superconducting magnets, such as the ones used in nuclear magnetic resonance, one is often interested in using the minimum winding volume, in order to reduce the construction costs. On the other hand, a highly homogeneous, highly regulated magnetic field is an essential requirement for this kind of system. Moreover, small variations in the design parameters are unavoidable in the construction of any equipment, and must be taken into account during the design process. Due to these many requirements, the design of magnets with high field homogeneity is a very difficult task for traditional optimization methods. Optimization methods based in the artificial immune systems paradigm are usually able to perform global as well as local exploration of the search space. The local search feature can be used to perform an estimation of the sensitivity of the candidate solutions, thus enabling the algorithm to search for points that represent robust devices at a reasonable computational cost. In this work, we explore the design of a high-field, highly homogeneous superconducting magnets by using an immune-based robust optimization algorithm. The algorithm is described, and applied to the optimization of the device. The analysis of the results shows the validity of the proposed method