欢迎访问《环境科学学报》编辑部网站！

快速检索			高级检索

研究论文

摘要：运用数值分类学手段建立了一种迭代的(iterative)、分裂的(divisive)、不加权的(non-weighting)、非交迭(nonoverlapping)的运算程式(IDNN模式),从实测原始数据中提取可近似作为纯源的样品和相应的性状,通过分析这些近似纯源样品及与之关系密切的其它特征样品的采样参数和成分特征可鉴别或识别源的类型,运用化学质量平衡法(CMB)求解受体样品的定量来源.在对模拟数据的验证处理中,IDNN模式显示了初步成功.

Abstract：An IDNN (iterative, divisive, non-weighting and nonoverlapping) model presented in this paper applies numerical taxonomy method to select the special samples, that can be seen as of pure sources, and corresponding attributes or variables from original data set.The model identifies Sources by analyzing the sampling parameters and the composition characters of the quasipure source samples,and calculates the quantitative contribution to the receptor samples for each source by CMB method. The model was applied to analyze two examples and satisfied results were obtained.