摘要
针对马尔科夫链预测的局限性,提出了能够进行清晰定量计算的叠加矿井涌水量的马尔科夫链预测方法。基于2008年1月-2013年12月成庄煤矿72个月的涌水量资料,进行涌水量状态分级,计算状态转移矩阵,将不同步长转移矩阵求得的预测值进行叠加平均,进而建立了叠加马尔科夫链预测模型,分析拟合效果,预测了2014年1月-4月的涌水量,并与实测值进行了对比。结果表明,该模型的预测精度达到了94.84%,预测效果较好,从而为矿井涌水量的预测提供了一种新方法。
Superimposed Markov chain was proposed to predict mine water inflow quantitatively since the general Markov chain has limitations.. Based on the water inflow data in the Chengzhuang coat mine from January 2008 to December 2013 （72 months）, water inflow status was classified, state transition matrix was calculated, the predicted values from different step matrixes were superimposed and averaged,and thus the superimposed Markov chain model was built and the fitting results were analyzed. The water inflow data from January to April 2014 were predicted and compared with the observed data. The results showed that the model prediction accuracy is about 94. 84%, so this new method can be used for mine water inflow prediction.
出处
《南水北调与水利科技》
CAS
CSCD
北大核心
2015年第3期409-412,共4页
South-to-North Water Transfers and Water Science & Technology
基金
国家自然科学基金"基于水化学关键因子的相似矿区煤层底板突水水源的识别"(41272250)
关键词
叠加马尔科夫链
矿井涌水量
预测模型
状态分级
superimposed Markov chain
mine water inflow
prediction model
status classification
