利用改进粒子群算法优化ERT有限元模型

摘要

为了提高电阻层析成像正问题计算精度,鉴于不同拓扑结构有限元模型对应的正问题计算精度不同,针对电阻层析成像有限元模型优化的特点,提出一种改进粒子群算法.该算法借鉴区间算法与轮盘赌选择生成粒子初始位置,并在算法迭代过程中采取改进精英策略以克服算法早熟收敛.仿真实验结果表明,新算法具有收敛速度快、精度高、稳定性好等优点,满足电阻层析成像有限元模型优化的要求,可有效提高正问题计算精度.%Aiming to improve the solution accuracy of the forward problem,in view of the fact that finite element models of different topologies leaded to different precisions of the forward problem in electrical resistance tomography (ERT),this paper proposed an improved particle swarm optimization algorithm,according to the characteristics of ERT finite element model optimization.Firstly,the new algorithm generated the particle initial position drawing on interval algorithm and roulette wheel selection,and adopted improved elite strategy to overcome premature phenomena during the iteration.Simulation results show that,the proposed algorithm has the advantage of fast convergence speed,high precision and good stability,meets the demands of ERT finite element model optimization,and enhances the solution accuracy of the forward problem.