摘要
论文提出了用遗传算法(GA)优化支持向量机(SVM)来预测矿区采空塌陷的方法。结合某矿区的实际塌陷情况,选取了17组采空塌陷数据作为训练样本,以覆盖层类型、覆盖层厚度、地质构造复杂程度、矿区倾角、采空体积率、采空区距地表的垂深和采空区空间叠置层数7个指标作为模型输入,采空区稳定程度作为模型输出,构建GA-SVM矿区采空塌陷的预测模型。然后利用该模型对工区5组采空塌陷数据进行预测,其预测结果与实际情况相符。为了验证提出的模型的优越性能,将得到的结果与BP神经网络模型和常规SVM预测的结果进行对比了结果表明GA-SVM预测模型比BP神经网络和常规SVM具有更高的精度,进一步验证了该模型的有效性和可靠性。
In order to predict the collapse of the goaf accurately, the support vector machine(SVM) method optimized by genetic algorithm(GA) is proposed to predict the mining collapse. Combined with the actual collapse of a mining area, 17 groups of goaf collapse data are chosen as the training samples to construct GA-SVM prediction model of mining collapse, in which 7 indexes, i.e. the coating types, coating thickness, complexity of geological structure, mining area angle of mined-out area away from the earth's surface, the air volume rate and the vertical depth and goaf spatial superimposed layer, are chosen as the model input and goaf stability as model output. Then this model is used to predict 5 groups of mining collapse data in the same mining area and the prediction results are in accordance with the actual situation. In order to verify the superiority of the proposed model, the results are compared with those obtained by BP neural network model. The results show that the GA-SVM prediction model has higher precision than BP neural network model, which further proves the validity and reliability of the model.
作者
于少将
YU Shao-jiang(Hebei GEO University,Shijiazhuang,Hebei 050031)
出处
《河北地质大学学报》
2018年第5期48-51,共4页
Journal of Hebei Geo University
基金
国家自然科学基金(41301015)
河北省教育厅重点项目(ZD2016038)
石家庄经济学院国家自然科学基金预研基金(syy201308)