改进粒子群算法及其在超导电缆参数优化中的应用

Improved Particle Swarm Optimization with Its Applications to Parameter Optimization of High Temperature Superconducting Cables

提出了一种改进粒子群优化算法,引入平均粒距和差分进化算子改善了传统粒子群算法在高维、多峰函数寻优应用中容易出现“早熟”的问题,并且应用物理加速度的概念,描述了其他粒子的速度信息对某个粒子速度产生的影响,从而加快了粒子群算法的收敛速度.针对高温超导电缆电流分配不均将导致交流损耗增大的问题,应用改进粒子群算法对冷介质超导电缆的绕制方向、绕向角和半径进行了优化.计算结果表明,通过优化调整电缆导电层的结构参数,各层电流可以达到均衡分布,交流损耗显著降低.

An improved particle swarm optimization （PSO） is proposed. To solve the premature of multi-modal function search with PSO algorithm, the average-distance-among-points and differential evolution （DE） operator are utilized. Introducing the conception of acceleration, the effects from the other particles on a single particle are described to accelerate the convergence of PSO. For a high temperature superconducting （HTS） cable, the non-uniform AC current distribution among the multilayer conductors gives rise to the increased AC losses, therefore the structure parameters of the multilayer conductors are optimized with the improved PSO. The results manifest that a uniform current distribution among layers can be realized via the optimization of structure parameters, thus the AC losses reduced substantially.