【摘要】 化探异常是找矿的重要依据.传统地质统计方法具有无偏、最优等特点,但要求数据呈正态分布,而实际应用往往不符合统计假设;近年来分形理论被大量应用于地球化学异常确定,但存在需要平滑处理数据、不适合含特高品位值等问题;采用随机模拟进行空间分析往往忽视了数据空间分布的结构性特征.本研究利用基因表达式编程(Gene Expression Programming,GEP)在复杂数据建模方面的优势,提出GEP演化建模与空间结构分析有效结合的研究思路,通过克立格选择邻域样品,增强数据空间局部结构信息,采用GEP进行空间趋势分析,并利用多重演化建模技术修正趋势面模型.在云南个旧锡铜多金属矿床的应用实例表明,该研究充分利用了局部空间结构信息,强化局部区域的估值结果,提高建模精度,为有效圈定致矿异常提供新的解决途径.更多还原

【Abstract】 Geostatistics is the traditional algorithm to find the low limit of geochemical anomalies.But it requires normal distribution of data.The statistical assumption is often non-existent in the geological application.Fractal theory in finding geochemical anomalies needs smooth data and is sensitive in the samples of high grade.Random simulation analysis tends to ignore spatial structural characteristics.In order to determine geochemical anomalies effectively,this paper proposed a new approach for space interpolation by using two stages Gene Expression Programming(GEP)evolutionary algorithm based on geostatistics.In this approach Kriging was used to select adjacent samples and enhanced local spatial structure information of geological variables and weaken the noise.It presented a hybrid model which combines GEP evolution modeling with geostatistics and used the two stages GEP evolutionary algorithm to modify the spatial distribution model.Its application in determining the geochemical anomalies of the mineral deposits in Gejiu,Yunnan Province illustrated that the approach of two stages GEP evolutionary algorithm based on Kriging could make full use of local spatial structure information and improve the valuation accuracy of geochemical data in some local areas.更多还原