混合遗传算法在核事故源项反演中的应用

Back-Calculation of Source Terms by Hybrid Genetic Algorithm in Nuclear Power Plant Accident

  • 摘要: 针对国内外普遍关注的核事故源项反演与事故后果评价的问题,运用遗传算法-单纯形法结合烟团模型实现了对源强的反算和释放点位置的快速定位。遗传算法-单纯形法与遗传算法-模式搜索法、遗传算法和单纯形法等3种算法的比较结果表明:遗传算法-单纯形法结合了遗传算法和单纯形法两种算法的优势,也弥补了各自算法的缺陷,其反算值可与期望值较精确符合;扩散模式模块、GA模块和NM模块3者可简单直接的结合,结合所需编写的代码较少,通用性广;GA模块和NM模块的计算花费较少,适用于核电厂对源项的快速估计。

     

    Abstract: To address the issue of nuclear accident’s consequence assessment and source terms inversion which is of common concern at home and abroad, hybrid genetic algorithm combined puff model was used to back-calculate the source terms including the release rate and the location. The results of comparing the genetic algorithm-Nelder Mead (GA-NM) method with the genetic algorithm-pattern search (GA-PS) method, genetic algorithm (GA) method and Nelder Mead (NM) method show that GA-NM method not only combines the advantages of GA method and NM method, but also compensates the shortcomings of the two algorithms.The inverse value can be exactly match the expected one. Dispersion model module, GA module and NM module can be combined straightforward, and the code used to combine them is very simple, so GA-NM method has a wide versatility. As the calculation of GA module and NM module is less costly, GA-NM method can be used for rapid estimation of the nuclear power plant source terms.

     

/

返回文章
返回