Elman network dynamic prediction method of Gas Emission based on the correlation analysis theory and the local linear embedding theory was proposed,for the gas emission affected by many factors,and there is a com⁃plex nonlinear relationship between them which resulting in low prediction accuracy. On the basis of correlation analysis on monitoring indicators,the mapping from high-dimensional space to low-dimensional space of Gas Emis⁃sion factors was realized by locally linear embedding theory,then remodeling the effective factor as the input vector of Elman network predictive model to reduce the complexity of the model structure,and at the same time,the Elman model was optimized with the bat algorithm to improve prediction accuracy and generalization ability. The experi⁃mental results show that the dynamic predictive method proposed in this paper has high generalization ability and prediction accuracy,which is applicable to the actual work of the Gas Emission Prediction.%针对瓦斯涌出量受诸多因素影响，彼此间存在复杂的非线性关系导致预测精度不高这一问题，提出基于相关分析理论和局部线性嵌入理论的Elman网络瓦斯涌出量动态预测方法。在对监测指标进行相关性分析的基础上，用局部线性嵌入理论实现瓦斯涌出量影响因素从高维空间至低维空间的映射，进而重构影响瓦斯涌出量的有效因子，并将其作为Elman网络预测模型的输入矢量，以降低模型结构的复杂度，同时用蝙蝠算法全局优化Elman模型以提高预测的精度和泛化能力。试验结果表明该动态预测模型泛化能力强，预测精度高，适用于实际工作中对瓦斯涌出量的预测。
展开▼
