Optimization of Relativistic Backward Wave Oscillator using Parallelized and Cached Particle Swarm Optimization Algorithm

We propose and implement an parallelized and cached particle swarm optimization (PSO) algorithm to assist the design of relativistic backward wave oscillator (RBWO). Parallelization through multiprocessing speed the optimization up, while archiving the simulated device avoids repetitive simulation. The optimization is therefore remarkably accelerated. The objective function is designed to combine the performance in both time and frequency domain. Using the algorithm, an X-band RBWO is auto designed. The optimized device is featured with three stages of non-uniform trapezoidal slow wave structures (SWS). Under the input DC voltage of 500 kV, current of 6.68 kA, guiding magnetic induction intensity of 0.55 T, the device output is 0.96 GW at 11.7 GHz, achieving the power conversion efficiency of 28.7%, much higher than the original one, demonstrating the proposed optimization algorithm effectively promotes the design of RBWO.