To address the challenges associated with the probability distribution and weight allocation processes of evaluation indicators for the stability of artificial pillars(backfill)using the segmented empty-field subsequent filling method,an evaluation indicator system was developed.This system considers the influences of five categories of information sources:Engineering geological environment,rock mass quality,mechanical response indicators,acoustic emission parameter characteristics,and geometric shapes.The resulting evaluation framework encompasses 17 factors that influence stability.Based on this analysis,the recognition framework and evaluation index grading standards for the stability of artificial pillars(backfill)were established utilizing Dempster-Shafer(D-S)evidence theory.By incorporating the normal membership function and an enhanced Kullback-Leibler(K-L)distance,the probability distribution and weight allocation of indicator factors were optimized.Consequently,an improved multi-source information fusion model,grounded in D-S evidence theory,was developed for the assessment of artificial pillar(backfill)stability.The applicability of the model was verified through five typical artificial pillar(backfill)in a mine with segmented open stoping subsequent filling mining.The results indicate that the stability evaluation levels for multi-source information fusion were ranked as follows:AP-2=AP-3=AP-5(Level Ⅱ)>AP-4(Level Ⅲ)>AP-1(Level Ⅳ).These stability evaluation out-comes are largely consistent with previous assessments based on displacement,acoustic emission on-site monitoring,and numerical simulation analysis.This congruence validates the scientific efficacy of the evaluation model in practical applications.Consequently,this model offers a more comprehensive approach for analyzing the stability of artificial pillars(backfill).
关键词
分段空场嗣后充填法/人工矿柱(充填体)/多源信息融合/稳定性评价模型/D-S证据理论/隶属度函数
Key words
segmented open stoping with subsequent filling method/artificial pillar(backfill)/multi-source information fusion/stability evaluation model/D-S evidence theory/membership function