基于RCGA的PPC模型在化探异常识别与提取中的应用

为了进行地球化学异常的识别与提取,针对化探数据的特点,本文提出了一种将高维降维技术——投影寻踪分类(PPC)模型与实数编码遗传算法(RCGA)相结合的计算方法,分析了运用RCGA-PPC模型进行化探异常识别与提取的关键技术问题,并在MATLAB环境下开发了该方法的软件应用模块。以云南个旧地区水系沉积物地球化学数据为例,选取区域内Sn、Cu、Pb、Zn、As、Cd等主要成矿元素及与成矿关系密切的9种元素作为计算变量,利用RCGA-PPC模型对其进行处理和异常识别。研究表明:RCGAPPC模型中最佳投影值较高的地区与该区域实际矿床(点)吻合情况较好。该模型对化探异常的识别能力较强,是一种有效的化探多元素综合异常识别与提取方法。

Application of PPC Model Combined with RCGA to Identify and Extract Geochemical Anomaly

In order to identify and extract geochemical anomalies, a novel method, which combines projection pursuit classification (PPC), a dimension-reducing technology on high-dimensional data, with the real coded genetic algorithm (RCGA), has been proposed in this paper.We analyzed and discussed the key technological issues in geochemical anomaly identification and extraction using RCGA-PPC model based on the features of geochemical data.At the same time, an application software package was developed for the RCGA-PPC model on MATLAB.As a case study, stream sediment geochemical data in Gejiu, Yunnan, China, were tested with the RCGA-PPC model and Sn, Cu, Pb, Zn, As, Cd, etc.ore forming or/and mineralization associate elements were selected as the variables to recognize anomalies.The results show that most of the local higher optimal projection values are well coincident with the deposits location.Therefore, it can be concluded that RCGA-PPC model could be a useful method for geochemical anomaly identification and extraction.