煤矿百万吨死亡率动态无偏灰色马尔科夫预测

为掌握我国煤炭行业整体安全生产状况,构建动态无偏灰色马尔科夫模型,通过处理我国1999—2008年度煤矿百万吨死亡率数据,对2009—2011年度煤矿百万吨死亡率数据进行预测。用无偏灰色GM(1,1)模型代替传统灰色GM(1,1)模型,拟合得到煤矿百万吨死亡率数据的变化趋势,在此基础上应用马尔科夫模型进行预测,并在每一次预测中更新原始数据,最后对预测结果进行对比分析。计算结果表明:动态无偏灰色马尔科夫模型既能消除传统灰色GM(1,1)模型的固有偏差,又能提高预测精度尤其是中长期预测精度。其中,平均绝对误差为3.35%,平均相对误差为3.85%,均低于传统灰色GM(1,1)模型和一般灰色马尔科夫模型。动态无偏灰色马尔科夫模型对于煤矿百万吨死亡率数据的平均预测精度超过96%。

Application of Dynamic Unbiased Grey Markov Model in Prediction of Death Rate per Million-ton Coal

In order to accurately predict the development trend in work situation of coal industries in China,dynamic unbiased grey Markov model DUGM-Markov was proposed to predict death rates per million-ton coal from 2009 to 2011 by dealing with that from 1999 to 2008.Instead of traditional GM(1,1) model,unbiased GM(1,1) model was used to fit the development tendency of death rates per million-ton coal and then Markov model forecast the fluctuation along it;the newest data was gradually added while the oldest one removed from the original data sequence.The results show that DUGM-Markov eliminates the inherent bias of the traditional GM(1,1) model and that the prediction accuracy is much higher,especially for middle and long term prediction,than that gotten from traditional grey GM(1,1) model.The average absolute error by DUGM-Markov is 3.35% and the average relative error is 3.85%,which are both lower than those by the traditional GM(1,1) model and traditional Grey Markov model.The average prediction accuracy of DUGM-Markov on death rate per million-ton coal is over 96%.