【摘要】 为了提高矿井瓦斯涌出量的预测精度,提出了一种基于支持向量机(SVM)的瓦斯涌出量非线性组合预测方法.该方法应用结构化风险最小化准则且具有在全局意义上逼近任意非线性函数特性的SVM,建立了一个多输入单输出的瓦斯涌出量非线性组合预测模型,通过样本学习和平均绝对百分比误差最小原则确定预测模型的参数,对双曲线回归、指数回归和灰色预测方法得到的3个不同的单项预测数据进行非线性组合作为最终预测结果.结果表明,该方法的平均绝对误差为6.92%,均方根误差为0.93 m3/t,其预测精度明显优于各个单项预测结果,大幅降低了预测风险,为提高瓦斯涌出量预测精度提供了一条新途径.更多还原

【Abstract】 In order to improve the forecast accuracy of gas emission amount in the coalmine,a nonlinear combined forecasting method using support vector machine(SVM) was proposed.The SVM is based on structural risk minimization(SRM) and can approximate any nonlinear function in the whole region.A multi-input and single-output(MISO) nonlinear combined forecasting model was built,and the parameters were tuned by training samples set and the principle of mean absolute percentage error(MAPE) minimization.Three original forecasting data from hyperbola regression forecast,exponential regression forecast and grey forecast,were combined as final forecasting result.The results show that the proposed method had much lower errors than other original individual forecast method,the MAPE is 6.92% and root mean square error(RMSE) is 0.93 m3/t,which provides a new approach to accurate forecast of gas emission amount.更多还原