摘要
为了更精确地对矿岩可爆性进行预测分级,对BP神经网络评价方法进行优化,建立了主成分分析法和改进BP神经网络相结合的矿岩可爆性分级评价模型。以具体矿山为例,考虑影响矿岩可爆性的10项评判指标,统计15个实际矿山的样本数据。利用SPSS软件对样本数据进行主成分分析,将输出结果作为改进BP神经网络的输入因子,矿岩的爆破等级作为输出因子,得到的分级预测结果更加准确且精度更高。结果表明:该模型对矿岩可爆性分级的相对误差都控制在6%以内,与未经主成分分析的BP神经网络预测误差相比,预测精度显著提高。该组合预测模型为矿岩的可爆性分级提供了一种更加完善的评价体系。
In order to predict the rock mass blastability classification more accurately,the method of BP neural network was optimized,the rock mass blastability classification evaluation model was established according to the principal component analysis and the improved BP neural network. Taking specific mine as an example,ten evaluation indexes influencing the rock mass blastability were considered,and the sample data in 15 actual mines were counted. The sample data were analyzed with principal component method by SPSS. The result was used as input factor of the improved BP neural network,and level of ore rock mass blastability was used as output factor. The prediction of ore rock mass blastability was more accurate. Results indicate that the relative errors of predicting outcomes from the model of rock mass blastability classification are all controlled within 6%. Compared with the prediction errors by BP neural network without principal components analysis,the prediction precision is improved greatly and the relative prediction error is obviously decreased. The combining prediction model provides a better system for evaluation of rock mass blastability classification.
出处
《爆破》
CSCD
北大核心
2016年第1期19-25,83,共8页
Blasting
基金
国家科技支撑计划项目(2013BAB02B05)
关键词
矿岩可爆性预测
评判指标
预测精度
组合模型
评价体系
rock mass blastability prediction
evaluation indexes
prediction precision
combining model
evaluation system