基于EMD-MFOA-ELM的瓦斯涌出量时变序列预测研究

为准确分析工作面绝对瓦斯涌出量的非平稳特征，实现瓦斯涌出量的准确预测，基于经验模态分解(EMD)、修正的果蝇优化算法(MFOA)和极限学习机(ELM)基本原理，构建瓦斯涌出量的EMD-MFOA-ELM多尺度时变预测模型。通过EMD将瓦斯涌出量时变序列进行深层次分解，获得多尺度本征模态函数(IMF)；采用MFOA-ELM对各IMF时变序列建立动态预测模型,等权叠加各预测值，得到模型最终预测结果。以晋煤某矿瓦斯涌出量监测时序样本为例进行研究分析，结果表明:EMD能充分挖掘出监测数据隐含信息，有效降低数据复杂度；该模型预测相对误差为0.024 3%~0.651 0%，平均值仅为0.252 6%，预测精度和泛化能力高于未经EMD分解模型，能很好地适用于非平稳时变序列预测。

Research on time-varying series forecasting of gas emission quantity based on EMD-MFOA-ELM

In order to accurately analyze the non-stationary characteristics of the absolute gas emission quantity in the working face and realize the accurate forecasting of the gas emission quantity, based on the basic principles of empirical mode decomposition (EMD), modified fruit fly optimization algorithm (MFOA) and extreme learning machine (ELM), a multi-scale time-varying series forecasting model of gas emission quantity based on EMD-MFOA-ELM was established. The time-varying series of gas emission quantity was deeply decomposed to obtain the multi-scale intrinsic mode function (IMF) by using EMD, and a dynamic forecasting model was established for each IMF time-varying series by using MFOA-ELM, then the final forecasting results of the model were obtained by superimposing each forecasting result with the equal weight. Taking the time series samples of gas emission quantity obtained by monitoring in a certain coal mine of Jincheng Coal Group as example, it showed that EMD could fully discover the implicit information and effectively reduce the complexity of the monitoring data. The relative error of the forecasting model was 0.0243%-0.6510%, and the average value was only 0.2526%. The prediction accuracy and generalization ability of the model were higher than that without EMD decomposition, and it can be well applied to non-stationary time-varying series forecasting.