关键词: 残差型, 最小二乘支持向量机, 覆盖层冒落高度, 碴石厚度, 预测, 钻孔摄影

Abstract: Due to the complexity and randomness of the forming condition of the mined-out area and its change，the prediction accuracy for roof rock caving height and ballast stone thickness of the mined-out area by the traditional method is often lower.In this paper，combining with the advantages of various single forecasting methods based on artificial intelligence，a forecasting model of residual grey squares new support vector machine prediction model (G-LSSVM) is put forward and then applied to predicting the roof caving height and ballast stone thickness of an iron ore.The predicting results are very close to the actual drilling photography monitoring result.The tests show: the residual type G-LSSVM for predicting the covering caving height and ballast stone thickness is feasible and effective.

Key words: Residual type, Least squares support vector machine, Covering caving height, Ballast stone thickness, Prediction, Drilling photography