针对矿井底板突水预测精确度要求高的特点,提出了二值Probit回归矿井突水预测模型。依据概率论的相关知识构建模型参数的极大似然估计法,并通过遗传算法求模型参数的最优解。利用所建模型对建模样本的突水性进行反向识别,对测试样本的突水性进行预测识别,正确率分别为90%和100%,表明该模型具有很高的精确度。另外,所建模型结构简单、便于操作和使用,有一定的参考价值。%In allusion to characteristic of high requirements on mine floor water bursting prediction precision, a binary Probit regression mine water bursting prediction model has been put forward. According to relevant knowledge of probability theory modeling parameters maximum likelihood estimation and genetic algorithm find out optimal solution of modeling parameters. Using the established model carried out reverse identification of modeling samples water bursting property, and predicted identification of tested samples water bursting property, their accuracies are 90% and 100% respectively. Thus demonstrated that the model has very high definition. Be⁃sides, the model has simple structure, practical and easy to use;thus has certain reference value.
展开▼