地采诱发建筑物损伤识别的SVM分析模型

针对地采诱发建筑物损害预测中指标与建筑物损害的关系不确定性问题，综合应用相关分析法、鱼骨图理论及SVM原理构建地采诱发建筑物损害的分析模型。采用相关分析法及原因型鱼骨图模型分析指标与建筑物损害的关联度，计算各因素权重，用建筑物损害观测数据对指标加权的SVM模型进行训练和测试，测试结果良好。研究结果表明：鱼骨图模型可获得指标与建筑物损害的关系，量化输入指标的重要性；建筑物本身条件中与空区位置、建筑物状况的重要性明显高于其他指标，加大建筑物本身建设，可较好地改善建筑物抗损害能力；基于鱼骨图的SVM分析模型可以更好地考虑各指标对建筑物损害的综合影响，回估误判率较低。

SVM Model for Identifying Building Damage Induced by Underground Mining

In view of the uncertain relationships among indexes and buildings damage in mining-induced damage predicting,and combining with the characteristics of correlation analysis,fishbone diagram,and SVM,the analysis model for mining-induced buildings damage was established.Correlation analysis theory and fishbone diagram were used to analyze the correlations,and calculate the indexes′ weights.The SVM model were trained and tested by a series of data from observations of mining-induced and damage degree of buildings,and the test results was good.The results show that the fishbone diagram can obtain the relationships between indexes and buildings′ damage,and the importance of the quantified index；The building′s own conditions including its position in Gob and building conditions are more important than others obviously,and increasing the construction of building foundation can well improve the destruction resistance of buildings；SVM model based on fishbone diagram can better consider the effect of each index on damage degree induced by mining-induced,and get a lower ratio of mis-discrimination.