矿体几何制图数据的稳健处理

ROBUST DATA PROCESSINGIN OREBODY GEOMETRIZATION

本文讨论了矿体几何制图数据的特点及常用处理方法存在的问题，比较分析了游动中值与游动平均的修匀效果及优缺点，研究了基于Huber-M估计的稳健游动平均修匀与稳健趋势面分析方法。理论分析、实例计算、蒙特卡洛模拟结果表明：这两种方法对异值影响有较好的抑制能力，它们既保留了传统数据处理方法的优点，又克服了某些不足，在剩余不服从正态分布时亦能取得较好的效果。

The characteristics of the data in orebody geometrization and the drawbacks inhe- rent in the general data processing method are discussed. The smoothing effects of mo- ving averages and moving averaging as well as their merits and demerits are compared and analysed. Also studied are the Huber M-estimation based analysis method of robu- st smoothing and that of robust trend surface. Theoretic analysis. practical computation and Monte Carlo simulation indicate that these methods are more capable of deterring the effects of outliers; they retain the merits of general data processing mothod while overcome some of it's demerits. Better effects can also be achieved in the case where residuals don't obey the law of normal distribution.