根据BP神经网络理论,对煤矿主斜井爆破开挖产生的振动强度进行了预测。通过建立合理的网络预测模型,以影响爆破振动特征参量的主要因素作为模型的输入,在模型误差小于0.001的前提下,对21组数据进行了重点预测。参量预测误差均小于10%,可以满足该复杂场地条件下安全施工的预测要求。 According to the theory of BP neural network, the vibration intensity of blasting excavation in main inclined shaft of coal mine is predicted. By establishing a reasonable network prediction model and taking the main factors that affect the characteristic parameters of blasting vibration as the input of the model, 21 sets of data are mainly predicted with the model error less than 0.001. The error of parameter prediction is less than 10%, which can meet the prediction requirements of safety construction in this complex site.