Subtractive Clustering–Assisted Surrogate Model to Fast Optimal Design of Traveling Wave Tubes

Abstract

The current techniques in performance improvement of space traveling wave tubes (TWTs) have limitations. It is difficult to consider the balance of several technique indexes, optimized design accuracy, and calculation cost. To tackle such a burden, a multiobjective optimization framework based on the Kriging model is proposed in this research. This framework takes advantage of what Kriging models approximate the responses of the electromagnetic simulation process. It reduces the obstacles caused by multiple-task calculations of TWTs due to the high cost of accurate simulation. In the design of the L-band helix TWT in this research, the predicted values of the model are used as the objective functions, and the multiobjective optimization of its interaction segment is carried out. Also, the proposed infill sampling criterion based on the subtractive clustering method in this research raised the efficiency of building the Kriging model. The numerical results demonstrate that the proposed multiobjective optimization framework is reliable for designing TWTs. It can quickly produce an optimal design scheme, significantly improving the performance of the designed TWTs compared to the original design.