煤矿事故预测的马尔可夫SCGM(1,1)_c模型的建立与应用

Establishment and application of Markov SCGM(1,1)_c model for accidents forecast in coal mining practice

煤矿安全是煤矿生产的重要保障,煤矿事故预测是煤矿安全评价和决策的基础。结合灰色SCGM(1,1)_c预测模型和马尔可夫链理论的优点,根据煤矿生产的特殊条件,提出了基于马尔可夫链的SCGM(1,1)_c预测模型。首先利用灰色SCGM(1,1)_c预测模型对我国1990—2010年的煤矿事故百万t死亡率进行初次预测,然后根据初次预测结果,利用残差模型对SCGM(1,1)_c模型预测结果进行修正。最后在修正模型的基础上,运用马尔可夫SCGM(1,1)_c模型对我国2011—2013年煤矿事故百万t死亡率进行了预测,并对两种模型的预测误差进行了对比分析。结果表明,马尔可夫SCGM(1,1)_c预测模型既能揭示煤矿事故百万t死亡率变化的总体趋势,又能克服随机波动性数据对预测精度的影响,具有较强的工程实用性。

The present paper is intended to propose Markov SCGM（ 1,1）_c model for accidents-forecasting in the coal mining practice. As is known,according to the gray system theory and Markov chain principle,gray SCGM（ 1,1）_c model is supposed to be available for predicting and forecasting the stochastic fluctuating dynamic process,such as those of the accidents of the systems with little data,short period and minor fluctuations. Therefore,in this paper,we would like to combine the advantages both of the gray system theory and Markov chain model to propose an innovated gray Markov SCGM（ 1,1）_c model to meet the needs of the special conditions of the coal mining production practice. To achieve the aforementioned purpose,we have first of all taken the coal mine accidents that have taken place during the period from1990 to 2010 in China as the starting prediction indicators for our analysis,including the death rate accidents per million-ton coal production by using the single gene system cloud gray SCGM（ 1,1）_c model to imitate the developmental tendency of the forecast system in combination with Markov model. And,then,we have suggested that the residual error model be used to amend the SCGM（ 1,1）_c model on the basis of the primary forecast results in hoping to improve the forecast expectancy. And,finally,it is suggested that Markov SCGM（ 1,1）_c can be applied to the forecast of the death rate per million ton of coal mine accidents from 2010 to 2013 years in China based on the residuals-amended model. Thus,comparing the two versions of the forecast models,we have found that the Markov SCGM（ 1,1）_c model can not only help us to disclose the trend of the death rate per million ton of coal mine accidents,but it also is in a position to overcome the random fluctuation of the data that may affect the statistical precision. Hence,it can be concluded that the forecast model we have innovated is endowed with the capability to forecast and control the coal-mine accidents,it can also serve as a strong reference to predicting and defending the engineering practicability.