我国矿业系统安全态势预测研究

发布时间：2018-10-05 06:31

【摘要】：矿业系统安全态势是衡量矿业系统安全管理效果和制定安全管理政策的重要依据，矿业系统安全态势的预测研究是确定安全投入和制定安全对策措施的重要基础之一。以我国矿业系统安全事故发生起数为系统安全态势表征值，依其趋势性、周期性、随机波动性等特征，研究建立了季节指数灰色马尔科夫加权组合预测模型，提出了灰色预测、马尔科夫预测串联组合再与季节指数预测加权并联组合的混合组合模式；以平均绝对误差最小为判断准则，选用幂指数方程作为季节指数预测的趋势方程，既较好地体现了安全态势非线性发展规律，也可提高趋势预测精度；依据预测平均绝对误差值最小原则，，确定了马尔科夫预测的状态数为4；提出了基于观测值与灰色预测值比值的“灰色马尔科夫预测组合方法”；以预测误差平方和倒数构建目标函数，优化权重系数，构建了灰色马尔科夫预测与季节指数预测加权组合模型。加权组合模型较季节指数预测模型和灰色马尔科夫预测模型精度分别提高了7.8%，5.6%。基于VB6.0开发平台，开发了季节指数灰色马尔科夫组合预测模型系统软件，实现了马尔科夫状态数确定、季节长度确定、趋势方程选择、单一模型预测、组合模型预测、组合精度计算、数据的动态更新与动态建模等功能，提高了预测工作的效率和准确度。
[Abstract]:The safety situation of mining system is an important basis to measure the effect of safety management in mining system and to formulate safety management policy. The prediction and study of safety situation of mining system is one of the important bases for determining safety input and formulating safety countermeasures. According to the characteristics of trend, periodicity and random fluctuation, the grey Markov weighted combination forecasting model of seasonal index is established, and the grey forecast is put forward according to the characteristics of trend, periodicity and random fluctuation of mining system safety accidents in China. The mixed combination model of Markov prediction series combination and seasonal index forecasting weighted parallel combination is adopted as the criterion of minimum mean absolute error and power exponent equation is chosen as the trend equation of seasonal index prediction. It not only reflects the nonlinear development law of security situation, but also improves the precision of trend prediction, according to the principle of minimum average absolute error of prediction, the state number of Markov prediction is determined to be 4. Based on the ratio of observed value to grey prediction value, a combined method of grey Markov prediction is proposed, and the objective function is constructed from the sum of square of prediction error, and the weight coefficient is optimized. A weighted combination model of grey Markov prediction and seasonal index prediction is constructed. The precision of the weighted combination model is 7.856% higher than that of the seasonal index model and the grey Markov model, respectively. Based on the VB6.0 development platform, the system software of seasonal exponential grey Markov combined prediction model is developed, which realizes the determination of Markov state number, seasonal length determination, trend equation selection, single model prediction, combined model prediction, and so on. The efficiency and accuracy of prediction are improved by combining precision calculation, dynamic updating of data and dynamic modeling.
【学位授予单位】：武汉科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：X936

【参考文献】