矿井涌水量的混沌时序预测研究

Prediction Study on Chaotic Time Series of Mine Water Inflow

为真实还原矿井涌水动态特征,开展了基于相空间重构理论的矿井涌水量混沌预测研究。采集某铅锌矿山日涌水量时间序列,运用互信息函数法和Cao氏方法确定重构参数,以此对原始序列展开相空间重构。根据对重构序列主分量谱图及最大Lyapunov指数的分析,确定了重构序列的混沌特征,并计算其有效预测时长为13d,在有效预测时长内,平均预测误差为4.81%,预测效果较好;超过有效预测时长后,预测精度迅速下降,平均预测误差为15.13%。结果表明,将混沌方法用于矿井涌水量预测具有原理简单、计算效率高等优点,但仅适用于短期预测。

In order to restore the dynamic characteristics of mine water inflow truly, chaotic prediction of mine water in- flow was carried out based on phase space reconstruction theory. Daily water inflow observation sequence in a lead-- zinc mine was collected, and the reconstruction parameters were determined by mutual information method and Cao method, which lead to phase space reconstruction of original sequence. The chaotic characteristics of reconstructed se- quence were confirmed by PCA and Lyapunov index analysis. And the effective prediction length was calculated as 13 days, in which the average prediction error with a good effect was 4.81%. Beyond effective prediction length, the ac- curacy was decreased rapidly, corresponding to average pre- diction error of 15.13%. The results indicated that applying chaotic method into mine water inflow prediction had simple principle, high calculation efficiency and other advantages,whieh was merely effeetive for short-term prediction.