中文摘要：

为探索有效的爆破振动参数预测方法，以新疆某铁矿爆破振动测试为依托，在对露天开采爆破振动参数影响因素分析的基础上，运用人工神经网络，以总药量、最大单段药量、单位药耗、高程差、爆心距为影响爆破振动的主要因素，构建BP神经网络模型1；同时结合矿区工程地质调查结论，引入测振区域RMR值为地质条件表征值，作为爆破振动影响因素，构建修正后BP神经网络模型2，分别对振动速度峰值、振动主频和振动持续时间进行预测。研究结果表明：模型2的预测精度较模型1提高了5％～8％，且2个模型预测精度较萨道夫斯基公式所得精度均有提高。

英文摘要：

The influencing factors of blasting vibration parameters were analyzed based on the test of a iron mine of Xinjiang province and the BP neural network model 1 was established with total charge, maximum charge per interval, unit explosive consumption, elevation difference and explosive distance as main factors impacting the blasting vibrations. Also, modified BP neural network model 2 was established by introducing RMR value, the geological condition character, as the influence factor of the blasting vibration. The two models were adopted to predict the peak vibration velocity, dominant frequency, and the time of duration, respectively. The results reveal that the forecasting precision of model 2 is 5%-8% higher compared with model 1 and both are better than that obtained by Sadowsky formula.