煤与瓦斯突出的L-Isomap-KELM模型

Prediction Model for Coal and Gas Outburst Based on L-Isomap-KELM

煤与瓦斯突出预测是一个复杂多因素的、非线性的高维问题,传统的预测方法存在预测精度不高,预测速度慢等不足。针对上述问题,提出了将地标等距特征映射(Landmarks Isometric Mapping,L-Isomap)理论与核极端学习机(Kernel Extreme Learning Machine,KELM)相结合应用到煤与瓦斯突出预测中的新方法。首先,采用L-Isomap进行非线性降维,完成特征提取;然后,用KELM来融合煤与瓦斯突出风险与致突因素组成的特征向量之间的非线性关系,建立煤与瓦斯突出预测的L-Isomap-KELM模型,并将其与极端学习机(ELM)预测模型相比。仿真结果表明:L-Isomap-KELM预测模型能够达到97.31%的准确率,并且运算速度快,还具有很好的泛化能力。

Prediction of coal and gas outburst is a complex, multifactor, nonlinear high-dimensional problem. The traditional prediction method has the disadvantages of low prediction precision and slow prediction speed. Aiming at above problems, a new method of combining the landmarks isometric mapping(L-Isomap) theory with the kernel extreme learning machine(KELM) is applied to the prediction of coal and gas outburst. First, nonlinear dimensionality reduction is performed by L-Isomap, and feature extraction is completed. Then, KELM is used to fuse the nonlinear relationship between the coal and gas outburst risk and the characteristic vector of the sudden factors, and to establish the L-Isomap-KELM model for the prediction of coal and gas outburst, and it is comparee with the extreme learning machine(ELM) prediction model. The simulation results show that the L-Isomap-KELM prediction model can achieve 97.31% accuracy and has fast operation and good generalization ability.