论文部分内容阅读
放射性核素大气扩散模型中,基于示踪实验获得的经验扩散参数依赖于具体的实验条件,在事故条件下,由于风场、大气湍流、地表状态等与实验条件存在差异,经验扩散参数难以准确反映实际扩散过程。为了弥补这一不足,可以以经验参数为先验值,使用实际观测数据对其进行实时动态修正。本文基于遗传算法,建立动态修正模型,通过数值模拟得到4种适应度函数对修正结果的影响。结果表明,根据观测误差设置不同权重的适应度函数修正效果更好。在此基础上,使用Kincaid实验数据集进行模型预测能力的验证,结果表明,使用遗传算法对拉格朗日扩散模型中的扩散参数进行修正,可明显提高扩散模型的预测能力。
In the radionuclide atmospheric diffusion model, the empirical diffusion parameters obtained based on the tracer experiments depend on specific experimental conditions. Under accident conditions, the empirical diffusion parameters are difficult to be accurately estimated due to differences in wind field, atmospheric turbulence, surface conditions and experimental conditions Reflect the actual proliferation process. In order to make up for this problem, empirical data can be used as a priori value and real-time dynamic correction can be made by using the observed data. Based on genetic algorithm, a dynamic correction model is established in this paper. The effects of four fitness functions on the correction results are obtained by numerical simulation. The results show that the fitness function with different weights set according to the observation error is better corrected. On this basis, the ability of model prediction is verified by using Kincaid experimental data set. The results show that using genetic algorithm to modify the diffusion parameters in Lagrange diffusion model can obviously improve the prediction ability of diffusion model.