雷电流波形由波前时间、波尾时间及雷电流陡度因子表征，对电力电子系统的雷电防护具有重要意义。针对含噪雷电流波形的多参数估计问题，提出了一种多群落粒子群优化算法(Nelder-Mead particle swarm optimization，NMPSO)，对雷电流特征参数构成的粒子群整体按粒子群优化算法(particle swarm optimization， PSO)处理，每个子群落按Nelder-Mead单纯形法处理，有效解决了粒子群优化过程中的停滞问题。Heidler雷电流函数的测试表明，当信噪比在5~40 dB内变化时，函数匹配相关系数为0.854 6~0.999 9。此外，NMPSO算法在Heidler雷电流函数参数估计中具有很强的抑噪能力，该文为雷电流波形测量装置提供了一种快速有效的数值算法，因而具有一定的研究价值和工程意义。

Return stroke current waveform (RSCW), characterized by the front time, tail time, and current steepness factor, is of great importance for the lightning protection of electric and electronic systems. Focusing on the multi-parameter estimation of RSCW contaminated by measured noise, the paper puts forward a novel numerical algorithm, i.e. Nelder-Mead particle swarm optimization (NMPSO), in which a particle swarm consisting of lightning current character parameters was tackled in the Particle Swarm Optimization (PSO), and each sub-swarm was processed in the Nelder-Mead (NM) algorithm. Therefore, stagnation can be effectively eliminated in the process of PSO. The simulation results of the Heidler’s lightning current expression showed that correlation coefficients of curve match vary from 0.854 6 to 0.999 9 as the signal-noise ratio increasing from 5 dB to 40 dB. In addition, the application of NMPSO in the parameter estimation of the Heidler’s lightning current expression manifests the super noise-resistant ability. As a result, the proposed numerical algorithm with fast and effective properties can be successfully applied to parameter estimation of lightning current waveform in different expressions and to data compression in lightning measurement systems; it is valuable for engineering evaluation and lightning research.

教育部新世纪优秀人才支持计划项目(NCET-07-0718)。