基于人工神经网络模型的涔天河面板堆石坝爆破试验分析

Blasting Experiment Analysis of Centian River Concrete Faced Rock-fill Dam Based on Artificial Neural Network Model

为了消除堆石坝坝料爆破试验不准确造成的影响,以涔天河面板堆石坝扩建工程的爆破试验为例,建立了基于多工程的坝料爆破试验数据的人工神经网络模型,将不同工程的爆破试验数据作为训练样本对模型进行训练,以使模型达到最优,进而对作为检验样本的涔天河爆破试验数据进行预测。结果表明,该方法可利用各工程爆破试验之间的差异排除影响爆破结果的不确定性,研究结果适用于确定涔天河扩建工程面板堆石坝坝料开采的爆破设计参数。

In order to eliminate the inaccurate impact of blasting test of rockfill dam material,taking burst test of Centian river rockfill dam extension project for an example,artificial neural network model based the data of dam material blasting test was established.Then the burst test data of different projects were used as training samples to train and optimize the model.The blasting test data of Centian River as test samples were predicted.The results show that this method can take advantage of different laws between blasting test results to exclude the impact of inaccuracies caused by the blasting.The research results can determine the blasting design parameters for exploitation of dam material of the extension project of Centian River concrete faced rock-fill dam.