基于遗传算法优化的支持向量机品位插值模型

Ore grade interpolation model based on support vector machines optimized by genetic algorithms

将支持向量机(SVM)和遗传算法(GA)集成应用到矿体品位插值问题中,利用遗传算法全局搜索的优势对支持向量机的三个关键参数——惩罚系数C、不敏感系数ε和核函数参数σ进行寻优,克服单纯支持向量机法中依靠经验确定参数的局限性.将优化参数代入到支持向量机中进行迭代训练,得到基于遗传算法参数优化的支持向量机(GA-SVM)矿体品位插值模型.以国内典型矿山的实际勘探数据为例,通过该品位插值模型计算结果与传统插值方法计算结果和矿山生产实际数据的对比分析,验证了其可行性和有效性.

An approach which integrates support vector machines （SVM） and genetic algorithms （GA） was pro- posed to do ore grade interpolation. With the global searching characteristics, GA was used to select the optimal parameters of SVM, including the penalty parameter C, the insensitive coefficient ε and the kernel function parameter a, so this approach can overcome the limitation of a pure SVM method in determining parameters by experience. The optimal parameters were substituted into SVM iterative training, and then an ore grade interpolation model based on SVM optimized by GA was established. Taking an underground mine in China as an example, the feasibility and validity of the ore grade interpolation model were verified by comparing the model calculation results with the actual data of mine production and the calculation results of traditional interpolation methods.