基于经验模态分解的煤矿设备温度预测

Prediction of Coal Mine Equipment Temperature Based on Empirical Mode Decomposition

煤矿设备出现故障时,设备温度会迅速上升,表现出非线性和非平稳性的特点。为了较准确地预测温度异常,采用了基于经验模态分解(EMD)的神经网络方法对设备温度进行预测。该方法首先采用经验模态分解算法对设备温度时间序列进行分解,得到若干个平稳性较好的本征模态函数(IMF)分量和一个剩余量。然后分别对各分量及剩余量进行神经网络预测。仿真结果表明,基于EMD的神经网络预测方法比单一神经网络预测方法,预测精度更高,对于温度异常预测更有效。

Equipment temperature rises rapidly during coal mine equipment failure, and the equipment tempera- ture data are nonlinear and nonstationary. In order to predict temperature anomalies more accurately, the method of neural network based on empirical mode decomposition （EMD） is used. This method firstly uses empirical mode decomposition to decompose the equipment temperature time series into several intrinsic mode functions （IMF） and a residue. Then each component and the residue are predicted separately with neural network. The simulation re- sults show that the neural network prediction method based on EMD is more accurate than single neural network prediction method and more effective for the temperature anomaly forecast.