Over-break caused by smooth blasting operation has been extensively considered the main factor affecting not only the safety of tunneling but also the cost of construction. Well-understanding of the occurrence of over-break and its impact level is virtually conductive to improve the quality control of blasting. This work aims to map the presence law of over-break under the site conditions on the basis of field testing and to develop a probabilistic neural network (PNN)-based modeling approach for estimating the impact level of over-break. The statistical analysis of measured over-breaks reveals that the probabilistic distribution of over-break distance follows the normal distribution (i.e., Gaussian distribution), at confidence level of 95%. In this work, effects of over-break distance and corresponding area are counted into the evaluation of impact level of over-break. The rating of impact level is determined by the analytic hierarchy process (AHP) method and K-means clustering. Predictions of impact level conducted by the proposed PNN-based estimation method are reliable and applicable, with accuracy of 96%. Results of this investigation are expected to further enhance the understanding of over-break caused by smooth blasting and then to improve the design and operation of blasting in rock tunneling.

中文翻译：

---

概率神经网络评估爆破破坏的影响程度

人们广泛认为，由爆破作业引起的爆破不仅是影响隧道安全的主要因素，而且还影响到施工成本。充分理解爆破的发生及其影响程度实际上有助于改善爆破的质量控制。这项工作旨在在现场测试的基础上绘制现场条件下突围的存在规律，并开发一种基于概率神经网络（PNN）的建模方法来估算突围的影响程度。对测量的突越的统计分析表明，突越距离的概率分布遵循正态分布（即高斯分布），置信度为95％。在这项工作中 过冲距离和相应区域的影响被计入过冲影响水平的评估中。影响程度的等级由层次分析法（AHP）和K-means聚类法确定。所提出的基于PNN的估算方法进行的影响程度的预测是可靠且适用的，准确性为96％。这项研究的结果有望进一步增进对平滑爆破造成的过度破坏的认识，从而改善岩石隧道爆破的设计和操作。