充填体强度设计知识库模型

综合分析了国内外充填矿山的工程实践资料,并用分形维数来表征充填材料的级配特性,将矿体埋藏深度、矿体长度、矿体厚度、充填体暴露高度、充填体暴露面积及充填材料分形维数6个因素作为输入,充填体设计强度作为输出,建立了充填体强度设计的神经网络知识库模型.为了避免神经网络易陷入局部极小以及收敛速度慢等缺陷,采用遗传算法对神经网络进行优化,提高了神经网络学习效率.通过对知识库模型进行检验,发现其具有很高的计算精度.研究表明,随着开采深度的增大,充填体设计强度必须随之增大;充填体侧向暴露面积越大,所要求的充填体强度越高.将该模型应用于三山岛金矿,设计了合理的充填体强度.

Knowledge bank model of design of backfill strength

The engineering practices in domestic and oversea filling mines were comprehensive analyzed,and the grading characteristics of filling material were researched with fractal dimension.Adopting neural network,which input data were deposit depth,ore body length,ore body thickness,exposure height of backfill,exposure areas of filling body and fractal dimension of filling materials,and output was design strength of backfill,a knowledge bank model of filling strength design was established.In order to avoid neural network easily fall into the local minimum and slow convergence and other defects,the neural network model was optimized by genetic algorithm,which improved the learning efficiency.Through inspecting the calculation results,the knowledge bank model had high precision.Researches show that,with the increase of mining depth,the filling design strength must be increased;and the more larger lateral exposure area of filling body was,the higher strength of filling body was required.Applying the knowledge bank model in Sanshandao Gold Mine,the rational backfill strength was obtained.