基于GRNN的金山店铁矿爆破振动峰值速度预测

为研究爆破振动对金山店铁矿地表构筑物和井下巷道的影响,引入广义回归神经网络(GRNN)的方法,分别以地表、井下部分振动监测数据为学习样本对GRNN进行训练,构建地表、井下爆破振动峰值速度的GRNN预测模型,以剩余振动监测数据为检测样本对地表和井下GRNN预测模型进行检验,并将GRNN模型的预测结果与BPNN、基函数回归法和经验公式法作对比。同时,将地表、井下GRNN模型的预测结果与以地表和井下综合训练数据为学习样本构建的综合GRNN预测模型进行对比。研究结果表明:对于地表监测点,四种方法的预测误差率依次为12.1%、18.9%、30.3%、43.7%;对于井下监测点,四种方法的预测误差率依次为14.0%、16.2%、19.9%、23.0%。GRNN的预测精度最高,其为爆破振动峰值速度的预测提供了一种新方法,且采用GRNN对地表、井下质点爆破振动峰值速度分别进行预测更加合理。

Prediction of Peak Particle Velocity of Blasting by GRNN in Jinshandian Iron Mine

In order to study the influence of blasting vibration on the surface structure and underground roadway in Jinshandian iron mine,the method of generalized regression neural network(GRNN) was introduced and trained by the study samples which are part of surface,underground vibration monitoring data respectively.Then,the surface and underground GRNN models of blasting peak particle velocity are constructed and tested,of which the tested sample is the monitoring data of remainder vibration.Comparison of prediction results is made among GRNN,BPNN,basis function and the empirical formula.Meanwhile,prediction results of the surface and underground GRNN models are compared with the comprehension GRNN of which the study samples are comprehensively trained data of the surface and underground GRNN models.The result shows the prediction error rates of the four methods with 12.1%、 18.9%、 30.3%、 43.7% for the surface monitoring points,respectively,and the prediction error rates of the four methods are 14.0%,16.2%,19.9%,23.0% for the underground monitoring points.Prediction accuracy of the GRNN model is the highest of the compared models,which provides a new method for predicting blasting peak particle velocity and it is much reasonable to predict the peak blasting particle velocity of surface and underground particle,respectively.