基于混沌理论和神经网络的瓦斯浓度预测研究

针对瓦斯浓度的非线性和混沌时间序列可预测的特点，根据Takens理论重构瓦斯浓度相空间，分别采用互信息最法计算时间延迟τ，G—P算法计算嵌入维数m，并采用BP神经网络对混沌时间序列进行预测，最后通过煤矿瓦斯浓度预测的实例说明预测结果，从而成功实现了对瓦斯浓度的预测。

Coalmine Gas Concentration Forecasting Based on Chaotic Theory and Neural Network Model

According to the non-linear of gas concentration and the predictability of the chaotic time series, gas concentration phase space was reconstructed by the Takens theory. In the first, the time delay 7 was attained by the mutual information method. Secondly the embedding dimension m was determined by GP algorithm and the chaotic time series was predicted by the BP neural network. Finally, an example is given which shows the forecast results could approximate the actual situation well, and accomplishing the forecast objection of gas concentration.