未确知聚类理论在顶板沉降量预测中的应用

 从距离判别的思想出发，对未确知聚类理论中的置信度准则进行优化，并将该理论应用于顶板沉降量预测中。选取岩体抗拉强度、埋藏深度、暴露面积等9 个影响因子，构建顶板沉降量预测的未确知聚类预测模型。根据收集的15 组样本数据确定了未确知测度函数，并利用熵权法计算指标权重，预测得出顶板沉降的分类等级及顶板沉降量。经计算验证，该方法的平均预测误差为7.38%，较模糊数学、灰色关联及神经网络3 种方法有更高的预测精度。为进一步验证其实用性，以辰州矿业沃溪矿区为例，采用该方法对4142 采场进行顶板沉降量预测。结果表明，预测结果与实际监测结果相吻合，证明该方法用于采场顶板沉降量预测是客观合理的，可为矿山安全生产提供决策依据。

Application Research of Unascertained Clustering Theory in Prediction of Roof Settlement

The confidence criterion of the unascertained clustering theory is optimized on the basis of distance discriminant method, then this theory is applied to the prediction of roof settlement. Firstly, nine factors of roof settlement, such as the tensile strength of rock mass, stope depth, and exposed area, are selected to establish the model for unascertained clustering prediction. Then the unascertained measurement function is obtained based on fifteen groups of sample data, and the index weights of each factor are calculated by entropy theory. Finally, the results of the classification grade and prediction of roof settlement are obtained. According to the computing validation, the average prediction error of this method is controlled within 7.38%, which is improved greatly compared with other three methods including the fuzzy mathematics method, the grey relation method and the neutral network method. Taking Woxi mining area of Chenzhou mine for example, this optimized method is applied to the prediction of roof settlement of No.4142 stope in order to further inspect its practicability. The results show that the prediction results are consistent with the actual situation, and that the optimized method of roof settlement prediction is objective and reasonable and can provide a decision-making basis for the safety production on mine.