基于SVM的矿井通风阻力系数影响因素分析与预测

摘要

矿井通风阻力系数是通风安全最重要的基础参数之一,为了实现矿井通风阻力系数简单准确地预测,提出了利用支持向量机(SVM)来预测矿井通风阻力系数的方法.通过分析影响因子与矿井通风阻力系数的相关性关系,并利用MATLAB逐步建立单影响因素与矿井通风阻力系数、多影响因素与矿井通风阻力系数之间的SVM预测模型,对比分析各预测模型的相对误差,确定最佳矿井通风阻力系数预测模型,即当输入模型影响因素为巷道断面积、周长和支护方式时,预测相对误差小于10%的样本数占测试样本的76%,相对误差小于20%的样本数占测试样本的90%.结果表明:该预测方法在矿井通风阻力系数预测中是可行的,并具较高的准确性.%The mine ventilation resistance coefficient is one of the most important basic parameters of ventilation safety,and in order to realize the simple and accurate prediction of the mine ventilation resistance coefficient,a method to predict the mine ventilation resistance coefficient by using the support vector machine (SVM) was put forward. Through analyzing the correlation between the influence factors and the mine ventilation resistance coefficient,the SVM prediction models between the single influence factor and mine ventilation resistance coefficient and between the multiple influencing factors and mine ventilation resistance coefficient were established step by step by using MATLAB. The relative error of each prediction model were compared and analyzed,then the optimal prediction model of the mine ventilation resistance coefficient was determined, namely when the influence factors inputting the model were the sectional area,the perimeter and the support mode of road-way,the sample number with the relative error of prediction being less than 10% accounted for 76% of the test samples,and the sample number with the relative error of prediction being less than 20% accounted for 90% of the test samples. The re-sults showed that this prediction method is feasible and accurate in the prediction of mine ventilation resistance coefficient.